ChE TRIAD e-LECTURE SERIES

𝝏𝑪 𝑨
𝑺𝑺

𝝏𝒕
, 𝟏𝑫

+𝑽
: 𝑽

.𝛁
Lectures can be downloaded from www.iitk.ac.in/che/nv.htm

�
Fin

𝑪𝑨
𝒅𝑪 𝑨 +

𝒅𝑪 𝑨 =
𝒅𝑿
�⃗𝒈

𝒅𝑿

=
𝝆

𝑫𝛁
ite
+

𝑫
𝒅
�⃗

𝟐
𝑽

𝟐 𝑪𝑨
𝟎

𝒅
𝝁𝛁 𝟐

𝟐 𝑪𝑨
𝒅𝑿

𝑪𝑨
𝒅𝑿
D

𝟐
𝟏

+(
+𝑪

−𝒌
iffe
𝒑+
�= =𝟎

na

𝟐𝒊

𝒓𝑨
𝑪𝑨
B

𝑪𝑨

)
�⃗

−𝛁

−𝟏
𝑽

𝑽⃗

=𝟎
me

re n
𝛁. �

,
e no

ce-
�⃗
𝛁𝑽

𝑩=
𝑽.

Ba

𝒊+

− 𝑽
𝝏𝒕 +

Ph

𝑫��
𝝏𝑽�⃗

𝟏
𝑺)

se d
(𝑺
𝝆�

𝑵−

;𝑪
rt

=
𝟏
Nu

𝒌
spo

�𝑽
, 𝚫
𝑵
m
𝑟 = 𝑘𝐶�

𝒉=
n

eri
Tra

𝑳� 𝑵
cal
Me
th o
ds
A
Heterogeneous Chemical Reaction Engineering
𝑪𝑨,𝒃 𝑹𝒑

𝒌𝒎

𝟏
𝑹𝒐𝒃𝒔 𝒏
𝑹𝒐𝒃𝒔 = 𝒂𝒌𝒎 �𝑪𝑨,𝒃 − � � �
𝒌𝜼𝒊𝒏𝒕𝒓𝒂

𝒌𝑪𝒏−𝟏
𝑨,𝒃
𝜼𝒊𝒏𝒕𝒓𝒂 = 𝜼𝒊𝒏𝒕𝒓𝒂 (𝒏, 𝝓); 𝝓 = 𝑳�
𝑫𝒑𝒐𝒓𝒆

Nishith Verma
Indian Institute of Technology Kanpur
(nishith@iitk.ac.in, vermanishith@gmail.com)
Transport Phenomena in
Chemical Engineering

1
 Lecture #01
Reference books

1. Transport Phenomena by BSL

2. Analysis of Transport Phenomena by Deen
Selective examples are also taken from
3. Boundary Layer Theory by Schlichting and Gersten
4. Physicochemical Hydrodynamics by Levich
5. Transport Phenomena Fundamentals by Joel Plawsky
6. Convective Heat and Mass Transfer by Kays, Crawford, Weigand

Transport Phenomena (continuum, laminar flow, incompressible fluid or flow at low Mach
number, Newtonian fluid)

Course in three parts: (1) Set up the equations describing the physical change in nature (vector
analysis, divergence theorem, etc.)

(2) Prototype problems/examples in momentum, heat and mass transport (analytical techniques:
Sturm Liouville, PDE solver etc.). These are relatively simpler problems or examples, dealing with
the laminar flow of an incompressible, Newtonian fluid.

(3) Approximate solution to complex problems (physics and math). These deal with flow at high
Reynolds numbers (boundary layer theory), and that at high Peclet numbers (concentration and
thermal boundary layers). The analytical solutions often use stream function. In this part of the
course, there are one or two introductory lectures on turbulent flow.

⇒ Scalars, vectors and tensors

Scalar – Numeric value such as mass, temperature

Vector – Magnitude and direction: 𝑣⃗ , 𝑞⃗, 𝑔⃗

eg. 𝑣⃗ = 𝑖𝑣𝑥 + 𝑗𝑣𝑦 + 𝑘𝑣𝑧 ; 𝑞⃗ = 𝑖𝑞𝑥 + 𝑗𝑞𝑦 + 𝑘𝑞𝑧 ; 𝑔⃗ = 𝑘𝑔𝑧

𝑚𝑎𝑔𝑛𝑖𝑡𝑢𝑑𝑒𝑠(𝑣𝑥 , 𝑣𝑦 , 𝑣𝑧 )

- Flux of a scalar is a vector (all the time?).

(𝑒𝑛𝑒𝑟𝑔𝑦, 𝑐𝑎𝑙)
𝑐𝑎𝑙 ⁄𝑠 − 𝑚2
𝑛̂ 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑖𝑠 𝑠𝑒𝑡 𝑏𝑦 𝑡ℎ𝑒 𝑛𝑜𝑟𝑚𝑎𝑙 𝑡𝑜 𝑡ℎ𝑒 𝑎𝑟𝑒𝑎.
𝑞𝑛 = |𝑞𝑛 |𝑛̂

- Tensor: 𝜏 𝑒𝑔. 𝑠ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑠𝑠 ≡ 𝑓lux of a vector (momentum); 𝑎𝑙𝑙 𝑡ℎ𝑒 𝑡𝑖𝑚𝑒?

2
 (𝑖𝑡 𝑖𝑠 𝑎 ℎ𝑖𝑔ℎ𝑒𝑟 𝑜𝑟𝑑𝑒𝑟 𝑞𝑢𝑎𝑛𝑡𝑖𝑡𝑦)

Let us understand stress tensor in the context of solid mechanics (fluid dynamics will be taken up later):

𝜏 in 𝑥 direction will have vectors in 𝑥, 𝑦, 𝑧 directions.

𝜏𝑥𝑥 𝜏𝑥𝑦 𝜏𝑥𝑧
𝜏
𝜏 = { 𝑦𝑥 𝜏𝑦𝑦 𝜏𝑦𝑧 } ; 𝜏𝑖𝑗 = 𝑔𝑒𝑛𝑒𝑟𝑎𝑙 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡
𝜏𝑧𝑥 𝜏𝑧𝑦 𝜏𝑧𝑧

All it means is that 𝜏 has a magnitude, direction it acts along, and the area upon which it acts.
𝑦
𝑧
𝜏𝑖𝑗
𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛
𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓
𝑡ℎ𝑒 𝑎𝑟𝑒𝑎 𝑥
𝑜𝑟 𝑖𝑡𝑠 ⊥ (𝑛𝑜𝑟𝑚𝑎𝑙)
(𝜏𝑦𝑥 , 𝜏𝑧𝑥 , 𝜏𝑥𝑥 )
𝑜𝑛𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 (𝑥)
𝑎𝑛𝑑 3 𝑝𝑙𝑎𝑛𝑒𝑠 (𝑣𝑒𝑐𝑡𝑜𝑟𝑠)

Conventionally , 𝜏𝑥𝑥 ≡ 𝜎 𝑥𝑥 , 𝜏𝑦𝑦 ≡ 𝜎 𝑦𝑦 , 𝜏𝑧𝑧 ≡ 𝜎 𝑧𝑧

𝑛𝑜𝑟𝑚𝑎𝑙 𝑠𝑡𝑟𝑒𝑠𝑠 𝐴
𝑠ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑠𝑠

Scalar Product: 𝐴⃗. 𝐵

⃗⃗ = |𝐴||𝐵|𝑐𝑜𝑠𝜃
𝜃
= 𝐴𝑥 𝐵𝑥 + 𝐴𝑦 𝐵𝑦 + 𝐴𝑧 𝐵𝑧
𝐵
Operator ⇒ Divergence

𝜕 𝜕 𝜕
⃗⃗ : ∇ ≡ 𝑣𝑒𝑐𝑡𝑜𝑟 = 𝑖 + 𝑗 + 𝑘
𝑒𝑔. ∇ . 𝑉 𝜕𝑥 𝜕𝑦 𝜕𝑧
(𝑑𝑒𝑙)
∇.𝑉⃗⃗ = 𝜕𝑉𝑥 + 𝜕𝑉𝑦 + 𝜕𝑉𝑧 ; 𝑉⃗⃗ = 𝑖𝑉𝑥 + 𝑗𝑉𝑦 + 𝑘𝑉𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧

⃗⃗ = 𝐶⃗ = |𝐴||𝐵|𝑠𝑖𝑛𝜃 𝑘̂
Vector Product: 𝐴⃗ × 𝐵

where, 𝑘 is the unit vector normal to both 𝐴 & 𝐵 (plane);

𝐶 𝐵 𝑖 𝑗 𝑘
𝑘̂ 𝐴⃗ × 𝐵
⃗⃗ = |𝐴𝑥 𝐴𝑦 𝐴𝑧 |
𝐵𝑥 𝐵𝑦 𝐵𝑧

3
 𝑖 𝑗 𝑘
𝜕 𝜕 𝜕
Curl Operator ⃗⃗ =
∇×𝑉 |𝜕𝑥 | ⃗⃗ 𝑐𝑎𝑛 𝑏𝑒 1𝐷 𝑜𝑟 2𝐷 𝑜𝑟 3𝐷)
(𝑉
𝜕𝑦 𝜕𝑧
𝑉𝑥 𝑉𝑦 𝑉𝑧

eg. vorticity 𝜔 ⃗⃗ (𝑐𝑢𝑟𝑙 𝑜𝑓 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦) = ∇ × 𝑉

⃗⃗ = 𝑐𝑢𝑟𝑙 𝑉 ⃗⃗

Vorticity is zero, implying that the element on fluid does not rotate about its own axis ⇒ irrotational flow.

⃗⃗ ≠ 0
∇×𝑉 ⃗⃗ = 0
∇×𝑉

′𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛′ ′𝑛𝑜 𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛′

A “zero-mass” object, for example, a piece of paper marked with cross sign does not rotate in an
irrotational flow.

𝜕𝜓 𝜕𝜓 𝜕𝜓
∇𝜓 ≡ 𝑔𝑟𝑎𝑑 𝜓 ≡ 𝑣𝑒𝑐𝑡𝑜𝑟 ≡ 𝑖 𝜕𝑥 + 𝑗 𝜕𝑦 + 𝑘 𝜕𝑧

(𝑠𝑐𝑎𝑙𝑎𝑟)
Recall: Curl of a grad (scalar) or ∇ × ∇𝜓 = 0 (in a conservative field, say gravitational field)

𝜕2 𝜕2 𝜕2
∇2 = ∇. ∇≡ (𝜕𝑥 2 + 𝜕𝑦2 + 𝜕𝑧 2 ) ≡ 𝑠𝑐𝑎𝑙𝑎𝑟

Substantial or material derivative

𝐷𝑓 𝜕
⃗⃗ . ∇ ) 𝑓
= (𝜕𝑡 + 𝑉 (𝑓 ≡ 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦)
𝐷𝑡

scalar

𝜕 𝜕 𝜕 𝜕
= (𝜕𝑡 + 𝑉𝑥 𝜕𝑥 + 𝑉𝑦 𝜕𝑦 + 𝑉𝑧 𝜕𝑧 ) 𝑓

Fixed (𝑥, 𝑦, 𝑧) other coordinates are fixed.

eg. Moving in the flow field with the fluid’s velocity (𝑉𝑥 , 𝑉𝑦 , 𝑉𝑧 ): time – rate of change in the property as
𝐷𝑓
seen/noted/observed by the observer ≡ 𝐷𝑡

Total derivative

𝑑𝑓 𝜕 𝑑𝑥 𝜕 𝑑𝑦 𝜕 𝑑𝑧 𝜕
𝑑𝑡
= (𝜕𝑡 + 𝑑𝑡 𝜕𝑥 + 𝑑𝑡 𝜕𝑦 + 𝑑𝑡 𝜕𝑧 ) 𝑓

chosen velocities by the observer

Moving in the flow field with chosen velocities (𝑉𝑥 , 𝑉𝑦 , 𝑉𝑧 )

𝑑𝑥 𝑑𝑦 𝑑𝑧 𝑑𝑓 𝐷𝑓
(If observer is moving along a fluid, 𝑑𝑡
= 𝑉𝑥 , 𝑑𝑡
= 𝑉𝑦 , 𝑑𝑡 = 𝑉𝑧 ; 𝑑𝑡
= 𝐷𝑡
)

4
eg. 𝐶 fish in a stream/lake: 𝐶(𝑡, 𝑥, 𝑦, 𝑧)

𝜕𝐶
(1) 𝑥, 𝑦, 𝑧 is fixed | (standing on a bridge over the lake)
𝜕𝑡 𝑥,𝑦,𝑧
𝐷𝐶
(2) 𝐷𝑡
(take a boat and let it move with fluid or switch off the engine)
𝑑𝐶
(3) 𝑑𝑡
(take a boat and choose your own velocity)

All three log-books will look alike

𝑦𝑜𝑢 𝑐𝑎𝑛𝑛𝑜𝑡
𝑡 − − − −
} 𝑡𝑒𝑙𝑙 𝑤ℎ𝑖𝑐ℎ
𝐶 − − − −
𝑑𝑒𝑟𝑖𝑣𝑎𝑡𝑖𝑣𝑒𝑠

Revisit: Tensors/ Dyadics

Scalar: Zero order 1 Component

Vector: 1st order 3 Components

Tensors: 2nd order 9 Components.

𝑇𝑥𝑥 𝑇𝑥𝑦 𝑇𝑥𝑧

𝑇
𝑇 = { 𝑦𝑥 𝑇𝑦𝑦 𝑇𝑦𝑧 } ; 𝑇ℎ𝑒 𝑠𝑐𝑎𝑙𝑎𝑟𝑠 𝑇𝑖𝑗 𝑎𝑟𝑒 𝑡ℎ𝑒 components of the tensor 𝑇
𝑇𝑧𝑥 𝑇𝑧𝑦 𝑇𝑧𝑧

Multiply (?) two vectors:

𝐴𝑥 𝐵𝑥 𝐴𝑥 𝐵𝑦 𝐴𝑥 𝐵𝑧
⃗ ⃗⃗
𝐴𝐵(also a tensor) = { 𝑦 𝐵𝑥
𝐴 𝐴𝑦 𝐵𝑦 𝐴𝑦 𝐵𝑧 } ⇒ 𝑑𝑖𝑎𝑑𝑖𝑐𝑠 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 : there are 9 scalar components
𝐴𝑧 𝐵𝑥 𝐴𝑧 𝐵𝑦 𝐴𝑧 𝐵𝑧

𝑁𝑜 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟 ′. ′ 𝑜𝑟 ′ × ′ in-between

Thus, in a fluid-flow field of velocity vector 𝑢⃗⃗, ρ𝑢

⃗⃗𝑢
⃗⃗ represents a tensor (convective momentum flux)
having usual 9 scalar components, with the general ρuiuj component representing the ρuj or j-momentum
transferred across the i-plane or in the direction perpendicular to j-k plane by the convective flow (ui):
𝑢𝑥 𝑢𝑥 𝑢𝑥 𝑢𝑦 𝑢𝑥 𝑢𝑧
{ 𝑦 𝑢𝑥
𝑢 𝑢𝑦 𝑢𝑦 𝑢𝑦 𝑢𝑧 }
𝑢𝑧 𝑢𝑥 𝑢𝑧 𝑢𝑦 𝑢𝑧 𝑢𝑧

Similar to the previous example of 𝜏𝑖𝑗 , the momentum (vector) of the fluid flow creates momentum flux
(tensors), parallel and normal to the flow in the fluid element, and therefore, there are 9 scalar
components. Such scenario will be discussed later for turbulent flow.

⃗⃗ . ∇⃗⃗⃗⃗
eg. 𝑉 𝑉 appears in NS equation.

See this term: ∇⃗⃗⃗⃗

𝑉 ≡ It is also a tensor with 9 components. ∇𝑖 𝑉𝑗 ≡ scalar components of the tensor

5
 𝜕 𝜕 𝜕
𝜕𝑋𝑖
𝑉𝑖 𝜕𝑋𝑖
𝑉𝑗 𝜕𝑋𝑖
𝑉𝑘

𝜕 𝜕 𝜕
𝜕𝑋𝑗
𝑉𝑖 𝜕𝑋𝑗
𝑉𝑗 𝜕𝑋𝑗
𝑉𝑘

𝜕 𝜕 𝜕
𝜕𝑋𝑘
𝑉𝑖 𝜕𝑋𝑘
𝑉𝑗 𝜕𝑋𝑘
𝑉𝑘

𝑖 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑗 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑘 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 (column-wise)

I leave this to you to give a suitable interpretation of the different scalar components.

Kronecker deltas (convenient to use for mathematically handling tensors operation)

𝐶onventional notations: ⃗⃗⃗

𝐴. ⃗⃗⃗
𝐵 = |𝐴||𝐵|𝑐𝑜𝑠𝜃 = ∑𝑖=𝑥,𝑦,𝑧 𝐴𝑖 𝐵𝑖 = 𝐴𝑥 𝐵𝑥 + 𝐴𝑦 𝐵𝑦 + 𝐴𝑧 𝐵𝑧

𝛿𝑖𝑗 = 1 𝑖 = 𝑗
∑𝑖 𝛿𝑖 𝐴𝑖 . ∑𝑗 𝛿𝑗 𝐵𝑗 = ∑𝑖 ∑𝑗 𝛿𝑖𝑗 𝐴𝑖 𝐵𝑗 = ∑𝑖 𝐴𝑖 𝐵𝑖 ( ) : Kronecker deltas
0 𝑖≠𝑗

= 𝐴𝑥 𝐵𝑥 + 𝐴𝑦 𝐵𝑦 + 𝐴𝑧 𝐵𝑧 same as before.

𝛿𝑖 , 𝛿𝑗 𝑎𝑟𝑒 𝑡ℎ𝑒
𝑢𝑛𝑖𝑡 𝑣𝑒𝑐𝑡𝑜𝑟𝑠;
𝛿𝑖𝑗 𝑖𝑠 𝑡ℎ𝑒 𝐾𝑟𝑜𝑛𝑒𝑐𝑘𝑒𝑟 𝑑𝑒𝑙𝑡𝑎.

𝛿𝑖𝑗 = +1 𝑖𝑓 𝑖 = 𝑗
Rules:
𝛿𝑖𝑗 = 0 𝑖𝑓 𝑖 ≠ 𝑗

𝜖𝑖𝑗𝑘 = +1 𝑖𝑓 𝑖𝑗𝑘 = 123, 231, 𝑜𝑟 312

= −1, = 321, 132, 𝑜𝑟 213 }
= 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

𝛿𝑖 . 𝛿𝑗 = 𝛿𝑖𝑗 (𝑠𝑐𝑎𝑙𝑎𝑟) 𝑖
𝑘 𝑗

𝛿𝑖 × 𝛿𝑗 = ∑3𝑘=1 𝜖𝑖𝑗𝑘 𝛿𝑘 (𝑣𝑒𝑐𝑡𝑜𝑟); 𝛿𝑖 𝛿𝑗 : 𝛿𝑘 𝛿𝑙 = 𝛿𝑖𝑙 𝛿𝑗𝑘 (𝑠𝑐𝑎𝑙𝑎𝑟)

𝛿𝑖 𝛿𝑗 𝛿𝑙
Verify ⃗⃗
and also, 𝑉 × 𝜔
⃗⃗ = | 𝑉1 𝑉2 𝑉3 | =
𝜔1 𝜔2 𝜔3

= ∑𝑖 𝛿𝑖 𝑉𝑖 × ∑𝑗 𝛿𝑗 𝜔𝑗 = ∑𝑖 ∑𝑗 ∑𝑘 𝜖𝑖𝑗𝑘 𝛿𝑘 𝑉𝑖 𝜔𝑗 = ∑𝑖 ∑𝑗 ∑𝑘 𝜖𝑖𝑗𝑘 𝛿𝑖 𝑉𝑗 𝜔𝑘 same as

𝑇𝑥𝑥 𝑇𝑥𝑦 𝑇𝑥𝑧

𝑇
𝑇 = { 𝑦𝑥 𝑇𝑦𝑦 𝑇𝑦𝑧 } (9 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡𝑠, 3 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛𝑠)
𝑇𝑧𝑥 𝑇𝑧𝑦 𝑇𝑧𝑧

6
 can be written as 𝑇𝑖𝑗 = ∑𝑖 ∑𝑗 𝛿𝑖 𝛿𝑗 𝜏𝑖𝑗 (9 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡𝑠, 3 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛𝑠)

⃗⃗⃗⃗𝐵
⇒ Multiplication of two vectors (𝐴 ⃗⃗⃗⃗ & 𝐵
⃗⃗) or dyadic product (Note there is no 𝑋 𝑜𝑟 . 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝐴 ⃗⃗⃗⃗ ) is a
tensor or dyad.

𝐴𝑥 𝐵𝑥 𝐴𝑥 𝐵𝑦 𝐴𝑥 𝐵𝑧
⃗⃗⃗⃗𝐵
𝐴 ⃗⃗ = ∑𝑖 ∑𝑗 𝛿𝑖 𝛿𝑗 𝐴𝑖 𝐵𝑗 = {𝐴𝑦 𝐵𝑥 𝐴𝑦 𝐵𝑦 𝐴𝑦 𝐵𝑧 }
𝐴𝑧 𝐵𝑥 𝐴𝑧 𝐵𝑦 𝐴𝑧 𝐵𝑧

Consider a dot product of a vector & a tensor to show:

𝐴. 𝑇 (𝐴̅. 𝑇̿) = (𝐴𝑥 𝑇𝑥𝑥 + 𝐴𝑦 𝑇𝑦𝑥 + 𝐴𝑧 𝑇𝑧𝑥 )𝑖 + (𝐴𝑥 𝑇𝑥𝑦 + 𝐴𝑦 𝑇𝑦𝑦 + 𝐴𝑧 𝑇𝑧𝑦 )𝑗 + (𝐴𝑥 𝑇𝑥𝑧 + 𝐴𝑦 𝑇𝑦𝑧 + 𝐴𝑧 𝑇𝑧𝑧 )𝑘

Similarly, show that ∇. (𝐴⃗𝐵 ⃗⃗⃗⃗. ∇)𝐵

⃗⃗) = (𝐴 ⃗⃗(∇. ⃗⃗⃗⃗
⃗⃗ + 𝐵 𝐴)

Vector Tensor

⇒Many of such tensor operations such as 𝑉 ⃗⃗⃗⃗. ∇𝑉

⃗⃗⃗⃗, ∇. 𝜏 . 𝑉
⃗⃗⃗⃗, 𝜏 : ∇𝑉
⃗⃗⃗⃗are performed using Kronecker deltas.
See the appendix of BSL and check the following equalities, as an exercise:
(1) 𝜎 : 𝜏 = ∑ ∑ 𝜎𝑖𝑗 𝜏𝑗𝑖
(2) ∇. (𝐴⃗𝐵 ⃗⃗⃗⃗. ∇)𝐵
⃗⃗) = (𝐴 ⃗⃗ + 𝐵 ⃗⃗⃗⃗)
⃗⃗(∇. 𝐴
(3) – 𝜏 : ∇𝑉 ⃗⃗ ≡ +𝑣𝑒 = 𝜇𝜙𝑉
(4) 𝜏 : ∇𝑉⃗⃗ = ∇.(𝜏. 𝑉
⃗⃗ ) - 𝑉
⃗⃗ . (∇. 𝜏) 𝑖𝑓 𝜏𝑖𝑗 = 𝜏𝑗𝑖 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝑡𝑒𝑛𝑠𝑜𝑟)

Component wise

𝜕 𝜕 𝜕
𝐴𝐵 = (𝐴𝑖 ) 𝐵𝑗 + 𝐵𝑗 ( 𝐴)
𝜕𝑋𝑖 𝑖 𝑗 𝜕𝑋𝑖 𝜕𝑋𝑖 𝑖

⃗⃗ ≠ 𝑉
∇. 𝑉 ⃗⃗ . ∇
Also show that }
⃗⃗ ≠ 𝑉
and ∇ 𝑉 ⃗⃗ ∇
𝜎11 𝜎12 𝜎13 𝜏11 𝜏12 𝜏13
Example: 𝜎 = {𝜎21 𝜎22 𝜎23 } 𝜏 = {𝜏21 𝜏22 𝜏23 }
𝜎31 𝜎32 𝜎33 𝜏31 𝜏32 𝜏33

𝜎 = ∑𝑖 ∑𝑗 𝛿𝑖 𝛿𝑗 𝜎𝑖𝑗 ; 𝜏 = ∑𝑘 ∑𝑙 𝛿𝑘 𝛿𝑙 𝜏𝑘𝑙

𝜎 : 𝜏 = ∑𝑖 ∑𝑗 𝛿𝑖 𝛿𝑗 𝜎𝑖𝑗 : ∑𝑘 ∑𝑙 𝛿𝑘 𝛿𝑙 𝜏𝑘𝑙 = ∑𝑖 ∑𝑗 ∑𝑘 ∑𝑙(𝛿𝑖𝑙 𝛿𝑗𝑘 ) 𝜎𝑖𝑗 𝜏𝑘𝑙

𝑤ℎ𝑒𝑛 𝑙 = 𝑖
= ∑𝑖 ∑𝑗 ∑𝑘 𝛿𝑗𝑘 𝜎𝑖𝑗 𝜏𝑘𝑙 → { 𝛿𝑖𝑙 = 𝛿𝑖𝑖 = 1
𝑎𝑛𝑑 𝜏𝑘𝑙 = 𝜏𝑘𝑖

𝑤ℎ𝑒𝑛 𝑘 = 𝑗
= ∑𝑖 ∑𝑗 𝜎𝑖𝑗 𝜏𝑗𝑖 → { 𝛿𝑗𝑘 = 𝛿𝑗𝑗 = 1 = (𝜎11 𝜏11 + 𝜎12 𝜏21 + 𝜎13 𝜏31 ) + (𝜎21 𝜏12 + 𝜎22 𝜏22 +
𝑎𝑛𝑑 𝜏𝑘𝑖 = 𝜏𝑗𝑖
𝑠𝑐𝑎𝑙𝑎𝑟
𝜎23 𝜏32 ) + (𝜎31 𝜏13 + 𝜎32 𝜏23 + 𝜎33 𝜏33 )

7
 Lecture #02
General Property Balance (continuum)

Some basics first:

Line Integral
⃗
1 𝜃𝐴 2
⃗⃗⃗⃗ = ∫2 𝐴 𝑐𝑜𝑠𝜃 𝑑𝑙 = 𝑤𝑜𝑟𝑘 𝑑𝑜𝑛𝑒 𝑖𝑓 𝐴⃗ = 𝐹𝑜𝑟𝑐𝑒
∫1 𝐴⃗. 𝑑𝑙 1
𝑑𝑙
2

Surface Integral Magnitude of a property-flux, 𝐴⃗ through ds = (𝐴⃗ . 𝑛̂)𝑑𝑠

𝑛̂
Total property, surface integral = ∬𝑆(𝐴⃗ . 𝑛̂)𝑑𝑠
𝜃
𝐴⃗
𝑑𝑠 = ∬𝑆 𝐴 𝑐𝑜𝑠𝜃 𝑑𝑠
𝑠𝑢𝑟𝑓𝑎𝑐𝑒
(𝑛̂ 𝑖𝑠 ⊥ 𝑟 𝑡𝑜 𝑑𝑠 𝑎𝑛𝑑 𝑖𝑠 𝑐𝑜𝑛𝑠𝑖𝑑𝑒𝑟𝑒𝑑 𝑡𝑜 𝑏𝑒 𝑡ℎ𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑑𝑠⃗ )

Integral Theorems

If ∀ is the closed region in a space, whose outer surface is S,

Then ∬𝑆(𝐴⃗ . 𝑛̂)𝑑𝑆 = ∭∀(∇. 𝐴⃗)𝑑∀

𝑠𝑝𝑎𝑡𝑖𝑎𝑙 𝑐ℎ𝑎𝑛𝑔𝑒 over 𝑑∀

(Divergence Theorem/Green’s Theorem)

𝑆 𝑛̂

∀
𝑑𝑆
𝐴⃗

Likewise for a scalar 𝜓

∬𝑆(𝜓 𝑛̂)𝑑𝑆 = ∭∀(∇𝜓)𝑑∀ (𝑣𝑒𝑐𝑡𝑜𝑟)

For a tensor 𝜏

∬𝑆( 𝜏 . 𝑛̂)𝑑𝑆 = ∭∀(∇ . 𝜏) 𝑑∀ (𝑣𝑒𝑐𝑡𝑜𝑟)

8
Stoke’s Theorem:

∬𝑆(∇ × 𝐴⃗). 𝑛̂ 𝑑𝑆 = ∮𝑙 𝐴⃗. 𝑑𝒍 (curl of a conservative vector is zero)

𝑛̂
𝐴⃗
𝑑𝑆
𝑆

𝑙
Frame of Reference ( Coordinate Framework)

⇒ Eularian co-ordinate framework

Identify a volume, ∀ in space at (x, y, z). Fluid flows through the volume. Identity time-changes across the
volume.
∀

different fluid elements. In this case, independent variables would be (𝑡, 𝑥, 𝑦, 𝑧)

⇒ Lagrangian co-ordinate framework

Identify a piece of mass (fluid element) and describe what happens to the body (as if the fluid element is
tagged and you are following it). Independent variable would be time only because (𝑥, 𝑦, 𝑧) ≡
𝑓(𝑡) 𝑜𝑟 (𝑥, 𝑦, 𝑧) are time-dependent.

eg. cannon balls projectile

𝑠𝑎𝑚𝑒 𝑏𝑎𝑙𝑙

𝑡2 𝑡3 𝑡4
𝑡1

General Property Balance Equation (see H.W. -1.5)

𝑦 𝑛̂ Take arbitrary volume ∀ fixed in space, bounded by surface S.

𝑧
𝑑𝑆
𝑥
∀, 𝑆
Fluid flow
⃗⃗ (𝑥, 𝑦, 𝑧, 𝑡)
𝑉
fixed in space ∀ ≠ ∀(𝑡)

9
 𝑚𝑎𝑠𝑠, 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚, 𝑒𝑛𝑒𝑟𝑔𝑦, 𝑚𝑜𝑙𝑒𝑠
𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦
Consider some property, 𝜓 ( )
𝑐𝑚3

𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦
Flux, = 𝑐𝑚2 −𝑠

𝜕
Time rate of change of property within volume: 𝜕𝑡 (∭∀ 𝜓𝑑∀ )

Net flow (transfer) of property across surface:

∬𝑆( . 𝑛̂) 𝑑𝑆

Generation term (rate at which property is generated in the volume ∀):

𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦
= ∭∀ 𝜓𝑔 𝑑∀ 𝜓𝑔 = 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 𝑜𝑓 𝜓, ( )
𝑠−𝑐𝑚3

𝜕
Balance: 𝜕𝑡
(∭∀ 𝜓𝑑∀ ) = − ∬𝑆( . 𝑛̂) 𝑑𝑆 + ∭∀ 𝜓𝑔 𝑑∀

Consists of at least two parts:

⃗⃗ (𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑜𝑤)
a) Convection, 𝜓 is carried by velocity 𝑉
𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦
⃗⃗ (
Flux ≡ 𝜓𝑉 )
𝑠−𝑐𝑚2
b) Some 𝜓 is transferred by non-convection (diffusion, radiation) ⇒ Call this flux

𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦
⃗⃗ +
= 𝜓𝑉 ( )
𝑠−𝑐𝑚2

convective part non- convective part

Substitute,

𝜕
∭∀ 𝜓𝑑∀ ⃗⃗ +
= − ∬𝑆( 𝜓𝑉 ). 𝑛̂ 𝑑𝑆 + ∭∀ 𝜓𝑔 𝑑∀
𝜕𝑡

𝜕
⇒ (∭∀ 𝜓𝑑∀ ⃗⃗ ). 𝑛̂ 𝑑𝑆 = − ∬ . 𝑛̂ 𝑑𝑆 + ∭ 𝜓𝑔 𝑑∀
) + ∬𝑆( 𝜓𝑉
𝜕𝑡 𝑆 ∀

𝜕 𝜕𝜓
𝜕𝑡
(∭∀ 𝜓𝑑∀ ) = (∭∀ 𝜕𝑡 𝑑∀ )

Applying Divergence theorem,

⃗⃗ ). 𝑛̂ 𝑑𝑆 = ∭ (∇. 𝜓𝑉
∬𝑆( 𝜓𝑉 ⃗⃗ )𝑑∀
∀

∬𝑆 . 𝑛̂ 𝑑𝑆 = ∭∀ ∇. 𝑑∀

𝜕𝜓
⃗⃗ ] 𝑑∀ = ∭ [−∇. + 𝜓𝑔 ]𝑑∀
∭∀ [ 𝜕𝑡 + ∇. 𝜓𝑉 ∀

𝐶𝑜𝑛𝑠𝑖𝑑𝑒𝑟𝑖𝑛𝑔 𝑡ℎ𝑎𝑡 ∀ 𝑖𝑠 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦,

10
 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝜕𝜓
∶ ⃗⃗ = −∇. + 𝜓𝑔
+ ∇. 𝜓𝑉 ⇒ 𝑔𝑒𝑛𝑒𝑟𝑎𝑙 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝑏𝑎𝑙𝑎𝑛𝑐𝑒 in CV
𝑠−𝑐𝑚3 𝜕𝑡

𝑢𝑛𝑠𝑡𝑒𝑎𝑑𝑦 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛

𝑡𝑒𝑟𝑚 𝑛𝑜𝑛 −
𝑠𝑡𝑎𝑡𝑒 𝑡𝑒𝑟𝑚
𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑡𝑒𝑟𝑚
𝑔
(1) 𝜓 = 𝜌 (𝑐𝑚3)
𝑔−𝑐𝑚⁄𝑠
⃗⃗ (
(1) 𝜓 = 𝜌𝑉 )
𝑐𝑚3
(2) 𝜓 = 𝑃𝐸 + 𝐾𝐸 + 𝑈 (𝑐𝑎𝑙⁄𝑐𝑚3 )
(3) 𝜓 = 𝐶𝑖 (𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3 ) (𝑠𝑝𝑒𝑐𝑖𝑒𝑠)

(1) Conservation of mass: Continuity (single (pure) fluid; no mixture)

𝜓 = 𝜌, 𝜓𝑔 = 0, = 0 (𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑚𝑜𝑑𝑒) 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡
𝜕𝜌 ⃗⃗ = 0
+ ∇. 𝜌𝑉
𝜕𝑡
𝜕𝜌 𝜕
or + 𝜕𝑥 𝜌𝑣𝑖 = 0 (𝐸𝑖𝑛𝑠𝑡𝑒𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛/𝑛𝑜𝑚𝑒𝑛𝑐𝑙𝑎𝑡𝑢𝑟𝑒)
𝜕𝑡 𝑖
(Einstein’s nomenclature: whenever a subscript is repeated, it implies summation)

(mass of a pure fluid in a CV varies with time because of convective transport only)

⃗⃗ = 𝜌(∇. 𝑉
∇. 𝜌𝑉 ⃗⃗ ) + (𝑉
⃗⃗ . ∇)𝜌

𝜕 𝜕𝑣 𝜕𝜌
or, 𝜕𝑥 𝜌𝑣𝑖 = 𝜌 𝜕𝑥𝑖 + 𝑣𝑖 𝜕𝑥
𝑖 𝑖 𝑖

𝜕𝜌 𝐷𝜌
⃗⃗ . ∇)𝜌 + 𝜌∇. 𝑉
+ (𝑉 ⃗⃗ = 0 ⇒ ⃗⃗ = 0
+ 𝜌∇. 𝑉
𝜕𝑡 𝐷𝑡

𝐷𝜌
(𝑚𝑜𝑣𝑖𝑛𝑔 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑓𝑙𝑢𝑖𝑑)
𝐷𝑡

If 𝜌 = 𝑐𝑜𝑛𝑠𝑡 (𝑖𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑)

𝐷𝜌
= 0; 𝜌 ≠ 𝜌(𝑥, 𝑦, 𝑧)
𝐷𝑡

⃗⃗ = 0 or 𝑑𝑖𝑣 𝑉
∇. 𝑉 ⃗⃗ = 0

𝜕𝑣𝑖
or =0 (𝑚𝑎𝑠𝑠 𝑖𝑠 𝑐𝑜𝑛𝑠𝑒𝑟𝑣𝑒𝑑; 𝑎𝑙𝑠𝑜 𝑛𝑜𝑡𝑒 𝑤𝑒 ℎ𝑎𝑣𝑒 𝑢𝑠𝑒𝑑 𝐸𝑖𝑛𝑠𝑡𝑖𝑒𝑛′ 𝑠 𝑛𝑜𝑚𝑒𝑛𝑐𝑙𝑎𝑡𝑢𝑟𝑒)
𝜕𝑥𝑖

In cylindrical or spherical coordinate system, a different type of velocity gradient terms appear in
the equation. You may not memorize these terms but for sure should refer the appendix of BSL to
become familiar with all terms in different coordinate systems. Also, the emphasis should be on
𝜕(𝑟𝑣𝑟)
the physical meaning of the terms. For example, in a 2D flow-field, 𝜕𝑟 represents the rate of
change in mass flowrate in r-direction (if you multiply it with 2𝜋𝜌), which must be the same as
that in x-direction for total mass flowrate to be conserved.
11
 Lecture #03
Linear momentum balance

⃗⃗ momentum/unit volume (a vector)

𝜓 =𝜌𝑉

Therefore, general property balance is a vector equation (unlike the previous continuity
equation), and therefore, flux term of the vector-property must be one order higher than the
property, which is a tensor or dyad.
𝜕𝜓
⇒ ⃗⃗ = −∇. 𝜓⃗⃗ + 𝜓⃗⃗𝑔 (general property balance)
+ ∇. 𝜓 𝑉
𝜕𝑡

⃗⃗)
𝜕(𝜌 𝑉
⃗⃗ 𝑉
+ ∇. 𝜌 𝑉 ⃗⃗ = −∇. 𝑃 + 𝐹⃗ (thus, there will be three scalar equations)
𝜕𝑡
𝑏𝑜𝑑𝑦 𝑓𝑜𝑟𝑐𝑒
𝑡𝑖𝑚𝑒 𝑟𝑎𝑡𝑒 𝑜𝑓 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑢𝑥 𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑢𝑥
𝑐ℎ𝑎𝑛𝑔𝑒 𝑜𝑓 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑜𝑓 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚
𝑜𝑓 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚
(𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑓𝑜𝑟𝑐𝑒)
𝜕𝜌𝑣𝑖 𝜕𝜌(𝑣𝑘 𝑣𝑖 ) 𝜕𝑃𝑘𝑖
+ = − + 𝐹𝑖 ⇒ Einstein notation (repeat indices indicate summation)
𝜕𝑡 𝜕𝑥𝑘 𝜕𝑥𝑘

Note: Newton’s 2nd law of motion defines 𝐹⃗ = net body force on fluid , i. e. gravity (restriction)

⃗⃗ 𝑉
Further simplification: ∇. 𝜌 𝑉 ⃗⃗ = 𝑉
⃗⃗ (∇. 𝜌𝑉
⃗⃗ ) + (𝜌𝑉
⃗⃗ . ∇)𝑉
⃗⃗ (recall previous lecture on Kronecker
deltas)
⃗⃗)
𝜕(𝜌 𝑉 ⃗⃗
𝜕𝑉 𝜕𝜌
Also, =𝜌 ⃗⃗
+𝑉 ⃗⃗
: ∇. 𝜌𝑉
𝜕𝑡 𝜕𝑡 𝜕𝑡

⃗⃗
𝜕𝑉
LHS = 𝜌 ⃗⃗ . ∇)𝑉
+ (𝜌𝑉 ⃗⃗ ( 𝜕𝜌 + ∇. 𝜌𝑉
⃗⃗ + 𝑉 ⃗⃗ )
𝜕𝑡 𝜕𝑡

𝜕𝑉 ⃗⃗
⃗⃗ . ∇)𝑉
= 𝜌 ( 𝜕𝑡 + (𝑉 ⃗⃗ ) 𝑧𝑒𝑟𝑜 𝑓𝑟𝑜𝑚 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦
(𝜌 ≠ 𝑐)
⃗⃗
𝜕𝑉
⃗⃗ ) = −∇. 𝑃 + 𝜓⃗⃗𝑔
⃗⃗ . ∇)𝑉
Equation becomes 𝜌 ( 𝜕𝑡 + (𝑉

(Assumption: 𝜌 is not constant)

Einstein equation:
𝜕𝑉 𝜕𝑉 𝜕𝑃𝑘𝑖
𝜌 ( 𝜕𝑡 𝑖 + 𝑉𝑘 𝜕𝑥 𝑖 ) = − + 𝜓𝑔𝑖 : i - momentum balance using Einstein’s nomenclature
𝑘 𝜕𝑥𝑘

𝜕𝑉 𝜕𝑉𝑥 𝜕𝑉𝑥 𝜕𝑉𝑥 𝜕𝑃𝑥𝑥 𝜕𝑃𝑦𝑥 𝜕𝑃𝑧𝑥

𝑖 = 𝑥: 𝜌 ( 𝜕𝑡𝑥 + 𝑉𝑥 + 𝑉𝑦 + 𝑉𝑧 )=− − − + 𝜓𝑔𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

12
 𝜕𝑉𝑦 𝜕𝑉𝑦 𝜕𝑉𝑦 𝜕𝑉𝑦 𝜕𝑃𝑥𝑦 𝜕𝑃𝑦𝑦 𝜕𝑃𝑧𝑦
= 𝑦; 𝜌 ( + 𝑉𝑥 + 𝑉𝑦 + 𝑉𝑧 )=− − − + 𝜓𝑔𝑦
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

Restrict to body force = gravity

𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑓𝑜𝑟𝑐𝑒
𝜓⃗⃗𝑔 = = = 𝜌 𝑔⃗
𝑣𝑜𝑙𝑢𝑚𝑒−𝑠 𝑢𝑛𝑖𝑡 𝑣𝑜𝑙𝑢𝑚𝑒

𝜕𝜌
So far, ⃗⃗ = 0 (𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦)
+ ∇. 𝜌𝑉
𝜕𝑡

⃗⃗
𝜕𝑉
⃗⃗ . ∇𝑉
𝜌 ( 𝜕𝑡 + 𝑉 ⃗⃗ ) = −∇. 𝑃 + 𝜌 𝑔⃗ ⇒ ( linear momentum balance or equation of motion)

(𝜌 ≠ 𝑐 and 𝑁𝐹 𝑜𝑟 𝑁𝑁𝐹 and 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑜𝑟 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡)

⃗⃗ and 𝜌(𝑥, 𝑦, 𝑧) unless we get 𝑃 in terms of 𝑉

We still cannot solve 𝑉 ⃗⃗ (velocity and density fields)

Therefore, we need insight into 𝑃 (surface forces/non-convective flux/pressure tensor)

Assume 𝑃 contains two parts:

a) One part which is non-zero for fluid at rest.

b) The other part which depends on fluid motion.

𝑃 =𝑝𝛿+𝜏 𝑝 + 𝜏𝑥𝑥 𝜏𝑥𝑦 𝜏𝑥𝑧

𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑠𝑡𝑟𝑒𝑠𝑠 = { 𝜏𝑦𝑥 𝑝 + 𝜏𝑦𝑦 𝜏𝑦𝑧 }
Thermodynamic 𝜏𝑧𝑥 𝜏𝑧𝑦 𝑝 + 𝜏𝑧𝑧
pressure (due to 1 0 0
{ 0 1 0}
state of the fluid) 0 0 1

Recall: 𝜏𝑖𝑗 ≡ Flux of 𝑗 − momentum transferred in 𝑖 direction or across j-k surface or plane?
𝑣𝑒𝑐𝑡𝑜𝑟

𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓
𝑓𝑙𝑢𝑥 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚

(or, force acting in 𝑗 direction on ′𝑖′ plane or ⊥ to 𝑗 − 𝑘 surface or plane?)

𝑣𝑒𝑐𝑡𝑜𝑟
eg. i

fluid above the j − k surface is sheared and moves

Further discussion on 𝜏𝑖𝑗 :

13
Apply conservation of angular momentum to a fluid element of volume 𝑑∀=
𝑑𝑥𝑑𝑦𝑑𝑧 & 𝑙𝑒𝑡 𝑑𝑥, 𝑑𝑦, 𝑑𝑧 → 0,

Show 𝑃𝑥𝑦 = 𝑃𝑦𝑥 𝑜𝑟 𝑃𝑖𝑗 = 𝑃𝑗𝑖 𝑜𝑟 𝜏𝑖𝑗 = 𝜏𝑗𝑖 𝑜𝑟 𝜏 = 𝜏 𝑇 (see old version of BSL)

⇒ Shear stress is symmetric (a fluid element cannot rotate or spin because it is a fluid)

⇒ Deen: For the overwhelming majority of fluids, including gases, homogeneous liquids of low
molecular weight, polymeric liquids, and most suspensions it is safe to assume that the stress
tensor is symmetric. Exceptions are in ferro fluids. The limitation on the conclusion that 𝜏 is
symmetric due not to any violation of the general principle of conservation of angular
momentum, but rather to the fact that some external torques may still exist in fluids.

⇒ To this end, one still cannot solve the equation of motion (consequence of linear momentum),
unless we know how 𝜏 depends on 𝑉 ⃗⃗ , which requires a ‘constitutive’ equation relating the non-
⃗⃗ in some ways.
convective term (flux), 𝜏 to 𝑉

It is found that in simple real fluids, 𝜏 depends only on the local first derivative of velocity i.e. on
the relative motion of neighboring particles. Convective momentum cannot be transferred if fluid
layers are at rest wrt each other.

𝑉 ⃗⃗ = ( 𝜕 𝑉
⃗⃗ ′ − 𝑉 ⃗⃗ ) . 𝜹𝐫 + approx.
𝑃′ ⃗⃗ ′ 𝜕𝑟⃗
𝑉
𝛿𝑟⃗ ⃗⃗ ). 𝜹𝐫
= (∇𝑉

𝑃 𝑡𝑒𝑛𝑠𝑜𝑟 ≡ 𝐷 (velocity gradient dyad)

𝜕𝑣𝑗 ⃗⃗ = 𝑉𝑥 , 𝑉𝑦 , 𝑉𝑧
𝑉
⃗⃗
𝑉 𝐷 ≡ 𝐷𝑖𝑗 = (9 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡𝑠 𝑜𝑓 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡𝑠) }
𝜕𝑥𝑖 𝑟⃗ = 𝑥, 𝑦, 𝑧

We cannot have a linear dependence between 𝜏 & 𝐷 ( 𝜏 ≠ 𝑘 𝐷) because 𝜏 is symmetric and 𝐷

is non-symmetric. 𝐷 can be written as the sum of a symmetric tensor and a non-symmetric
tensor.
1 1
𝐷=2(𝐷+𝐷 𝑇)+ (𝐷- 𝐷𝑇)
2

𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝑛𝑜𝑛 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐

1 1
𝑜𝑟, 𝐷𝑖𝑗 (≠ 𝐷𝑗𝑖 ) = ( 𝐷𝑖𝑗 + 𝐷𝑗𝑖 ) + 2 ( 𝐷𝑖𝑗 − 𝐷𝑗𝑖 )
2

14
 𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑣
(𝜕𝑥𝑗 + 𝜕𝑥𝑖 ) (𝜕𝑥𝑗 − 𝜕𝑥𝑖 )
𝑖 𝑗 𝑖 𝑗

𝑎𝑙𝑤𝑎𝑦𝑠 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝑎𝑙𝑤𝑎𝑦𝑠 𝑛𝑜𝑛 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐

𝑁𝑜𝑡𝑒:
𝐷𝑖𝑗 = 𝐸𝑖𝑗 + Ω𝑖𝑗 {𝑣𝑜𝑟𝑡𝑖𝑐𝑖𝑡𝑦, 𝜔
⃗⃗
=∇×𝑉 ⃗⃗
𝑟𝑎𝑡𝑒 𝑜𝑓 𝑣𝑜𝑟𝑡𝑖𝑐𝑖𝑡𝑦 𝑡𝑒𝑛𝑠𝑜𝑟
𝑠𝑡𝑟𝑎𝑖𝑛 𝑡𝑒𝑛𝑠𝑜𝑟 (𝑛𝑜𝑛 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)
(𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐) /𝑎𝑛𝑡𝑖 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐.

The first tensor relates to the local rate of deformation and the second relates to rotation.

Thus, 𝜏 can depend only on the symmetric portion of 𝐷 i.e. on 𝐸𝑖𝑗 .

( 2𝐷 Ω𝑖𝑗 implies pure rotation about own axis of a fluid element ≡ rigid body rotation / non-
deformation. Ω𝑖𝑗 = 0 ⇒ 𝑁𝑜 𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 in a fluid ) See chapter 5.2 of Deen’s book, including

examples 5.2.1 and 5.2.2, for details.

On 𝐸𝑖𝑗 (no uniform translation or rigid-body rotation)

𝐸𝑖𝑗 = {𝛼 0
}
Suppose 𝛼 off diagonal terms are zero.
(2𝐷) 0

This describes the following type of deformation:

𝑡=0 (for incompressible fluid, areas are same: linear

deformation)
𝑡=𝑡

0 𝛽
Suppose Ω𝑖𝑗 = { } : only off diagonal elements/terms are non-zero & they have to be
−𝛽 0
𝜕𝑉 𝜕𝑉𝑦
same ( 𝜕𝑦𝑥 = )
𝜕𝑥

(angular deformation or rotation only ⇒ lengths are the same;

𝑡=0

𝑡=𝑡 for incompressible fluid, areas must be the same)

15
Check out some of the introductory courses on fluid dynamics. In general, the fluid element
translates, dilates, rotates, and distorts, the latter two considered as angular deformation. Some
textbooks (Schlitchting) terms the kinds of fluid motions as extension, shear deformation,
deformation, and rigid body rotation.

To this end, for a class of fluids called Newtonian fluid

a) Incompressible fluid
𝜏 = −𝜇 𝐸 = −𝜇[∇𝑉 ⃗⃗ + ∇𝑉
⃗⃗ 𝑇 ] (Newton’s law of viscosity): constitutive type of equation

Here, 𝜇 is a scalar quantity (viscosity) and does not depend on 𝐸

b) Compressible fluid
2
𝜏 = −𝜇 𝐸+ 𝜇(∇. 𝑉 ⃗⃗ ) 𝛿
3

𝜕𝑉𝑥 2
𝜏𝑥𝑥 = −𝜇 (2 ⃗⃗ )
) + 3 𝜇(∇. 𝑉
𝜕𝑥

𝜕𝑉𝑦 2
𝜏𝑦𝑦 = −𝜇 (2 ⃗⃗ )
) + 3 𝜇(∇. 𝑉
𝜕𝑦

𝜕𝑉𝑥 𝜕𝑉𝑦
𝜏𝑥𝑦 = 𝜏𝑦𝑥 = −𝜇 ( + )
𝜕𝑦 𝜕𝑥

Equation of motion (for a compressible Newtonian fluid)

𝜕𝑉 ⃗⃗
⃗⃗ . ∇)𝑉
𝜌 ( 𝜕𝑡 + (𝑉 ⃗⃗) = −∇𝑝 − ∇. 𝜏 + 𝜌𝑔⃗

2
⃗⃗ + ∇𝑉
= −∇𝑝 + ∇. 𝜇(∇𝑉 ⃗⃗ 𝑇 ) − ∇. 𝜇(∇. 𝑉
⃗⃗ ) 𝛿 + 𝜌𝑔⃗
3

( ∇. 𝑝𝛿 = ∇𝑝)

⃗⃗ (𝑟⃗, 𝑡), 𝜌 (𝑟⃗, 𝑡), 𝑝(𝑟⃗, 𝑡)and 𝜇(𝑟⃗, 𝑡)

Unknowns are 𝑉

- We need three more equations to solve

𝜇 = 𝜇(𝜌) − (1)

𝜌 = 𝜌(𝑝) − (2)
𝜕𝜌
⃗⃗ = 0
+ ∇. 𝜌𝑉 − (3)
𝜕𝑡

In Einstein notation,

16
 𝜕𝑣𝑖 𝜕 𝜕 𝜕 𝜕 𝜕 2 𝜕 𝜕𝑣
𝜌
𝜕𝑡
+ 𝜌 (𝑣𝑗
𝜕𝑥𝑗
) 𝑣𝑖 = − 𝜕𝑥 𝑝 + 𝜕𝑥 𝜇 {𝜕𝑥 𝑣𝑖 + 𝜕𝑥 𝑣𝑗 } − 3 𝜕𝑥 𝜇 𝜕𝑥𝑖 + 𝜌𝑔𝑖
𝑖 𝑗 𝑗 𝑖 𝑗 𝑖

⃗⃗ = 0 𝑜𝑟 𝜕𝑣𝑖 = 0 (𝑖𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑)

𝐼𝑓 𝜌 = 𝑐𝑜𝑛𝑠𝑡 ⇒ ∇. 𝑉 𝜕𝑥 𝑖

𝜇 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

𝜕𝑣𝑖 𝜕 𝜕 𝜕 𝜕 𝜕 𝜕
𝜌 + 𝜌 (𝑣𝑗 ) 𝑣𝑖 = − 𝜕𝑥 𝑝 + 𝜇 𝜕𝑥 𝑣𝑖 + 𝜇 𝑣 + 𝜌𝑔𝑖
𝜕𝑡 𝜕𝑥𝑗 𝑖 𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑖 𝑗

Schwarz’s Rule:

𝜕2 𝑦 𝜕2 𝑦
= 𝜕𝑥
𝜕𝑥𝑖 𝜕𝑥𝑗 𝑗 𝜕𝑥𝑖

𝜕 𝜕 𝜕 𝜕
Therefore, 𝑣𝑗 = 𝜕𝑥 (𝜕𝑥 𝑣𝑗 ) = 0
𝜕𝑥𝑗 𝜕𝑥𝑖 𝑖 𝑗
𝜕𝑣𝑗
⃗⃗ = 0 𝑜𝑟
∇. 𝑉 =0
𝜕𝑥𝑗

(constant fluid density 𝜌 = 𝑐; incompressible fluid)

Therefore,
𝜕𝑣𝑖 𝜕 𝜕𝑝 𝜕2 𝑣𝑖
𝜌 + 𝜌 (𝑣𝑗 𝜕𝑥 𝑣𝑖 ) = − 𝜕𝑥 + 𝜇 𝜕𝑥 + 𝜌𝑔𝑖
𝜕𝑡 𝑗 𝑖 𝑗 𝜕𝑥𝑗
𝜕𝑉 ⃗⃗
⃗⃗ . ∇𝑉
𝜌 ( 𝜕𝑡 + 𝑉 ⃗⃗ ) = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗ + 𝜌𝑔⃗ ⇒ 𝑵𝒂𝒗𝒊𝒆𝒓 − 𝑺𝒕𝒐𝒌𝒆𝒔 𝒆𝒒𝒖𝒂𝒕𝒊𝒐𝒏

- Conservation of linear momentum for const 𝝆, const 𝝁, NF

 Component- wise, the NS equation is as follows:

𝜕𝑉 𝜕𝑉𝑥 𝜕𝑉𝑥 𝜕𝑉𝑥 𝜕𝑝 𝜕2 𝑉 𝜕2 𝑉𝑥 𝜕2 𝑉𝑥
𝜌 ( 𝜕𝑡𝑥 + 𝑉𝑥 + 𝑉𝑦 + 𝑉𝑧 ) = − 𝜕𝑥 + 𝜇 ( 𝜕𝑥 2𝑥 + + ) + 𝜌𝑔𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦 2 𝜕𝑧 2

You should become familiar with similar expressions in cylindrical and spherical coordinate
systems. These expressions are listed in the appendix of all books on transport phenomena.

17
Homework 1- 6
(These should be submitted as the course progresses. The instructor will inform.)

HW – 1
Q1. (a) Show that if 𝜏 is symmetric, i.e. 𝜏𝑖𝑗 = 𝜏𝑗𝑖

⃗⃗ = ∇. (𝜏 . 𝑉
𝜏 ∶ ∇𝑉 ⃗⃗ ) − 𝑉
⃗⃗ (∇. 𝜏)

3 2 −1 5
(b) If 𝜏 ≡ { 2 2 ⃗⃗ . [𝜏. 𝑉
1 } ; 𝑣̅ = { 3 }, evaluate 𝑉 ⃗⃗ ] , 𝜏: 𝜏, ⃗⃗⃗⃗
𝑉𝑉 ⃗⃗
−1 1 4 −2
Show that
⃗⃗ = (∇𝑠. 𝑉
(c )∇. 𝑠𝑉 ⃗⃗ ) + 𝑠(∇. 𝑉
⃗⃗ )

⃗⃗ = ∇(∇. 𝑉
(d) ∇. ∇𝑉 ⃗⃗ ) − [∇ × [∇ × 𝑉
⃗⃗ ]
1
⃗⃗ = ∇(𝑉.
(e) 𝑉. ∇𝑉 ⃗⃗⃗⃗ 𝑉
⃗⃗ ) − [𝑉
⃗⃗ × [∇ × 𝑉
⃗⃗ ]
2
𝜃(𝑡)
2: Consider the cylindrically symmetric flow field given by 𝑉𝑟 (𝑉, 𝑡) = , 𝑤ℎ𝑒𝑟𝑒 𝜃(𝑡) is a time-
2𝜋𝑟

dependent source at the origin, 𝑟 = 0 (𝑎 𝑠𝑖𝑛𝑔𝑢𝑙𝑎𝑟 𝑝𝑜𝑖𝑛𝑡). The velocity of a material point,
starting at (or near) 𝑟 = 0 decays experimentally with time from a value of 𝑣𝑜 . What function of
𝑡 must 𝜃(𝑡) be in order for this to be true?

3. A sensitive electronic instrument on board a balloon must not experience a rate of change in
temperature larger than ≠ 0.01 𝐾/𝑠 for proper functioning. The atmosphere temperature changes
with the height above the ground, and also, with time after sunrise, as given by
𝑇 = (288.2 − 6.5 × 10−3 𝑧)(2 − 𝑒 −0.01𝑡 )𝐾, 𝑤ℎ𝑒𝑟𝑒 𝑧 𝑖𝑠 𝑡ℎ𝑒 ℎ𝑒𝑖𝑔ℎ𝑡 𝑖𝑛 ′𝑚′ above the ground and
′𝑡′ is the time in hour after sunrise. Find the maximum permissible rate of ascent of balloon when
it is about to leave the ground immediately after sunrise?

4. The relation between shear stress, 𝜏 and the rate of deformation tensor, 𝐸 for a NNF is given by

1
𝜏𝑖𝑗 = −𝜂 𝐸𝑖𝑗 , 𝑤ℎ𝑒𝑟𝑒 𝜂 𝑖𝑠 𝑔𝑖𝑣𝑒𝑛 𝑏𝑦 √ (𝐸: 𝐸)
2

18
 ⃗⃗ = 3𝑥𝑦𝑖̂ + 4𝑥 2 𝑧𝑗̂ − 𝑦𝑧𝑘̂ , at the location
Calculate, ′𝜂′ of the fluid in a velocity field given by 𝑉
(1,-1,1) in the flow.

(5) Derive the general property balance for a control volume whose boundaries are moving with
fluid velocity V. Use Leibnitz’s theorem:
𝜕 𝜕𝜓
∭ 𝜓 𝑑𝑉 = ∭ 𝑑𝑉 + ∬ 𝜓 (𝑣𝑠 . 𝑛)𝑑𝑆
𝜕𝑡 𝑉 𝑉 𝜕𝑡 𝑆

where, V = V(t) and S = S(t). 𝑣𝑠 is the surface velocity vector.

HW – 2
(1) Determine the force required to move a thin plate of 30×60 𝑐𝑚2 size through a liquid of
𝑘𝑔
viscosity, 𝜇 = 0.05 𝑚−𝑠 at a velocity of 0.40 m/s. The liquid is filled between two long

parallel plates as shown in figure below:

𝑉 = 0.4 𝑚/𝑠
1.5 𝑐𝑚 60 𝑐𝑚 1.0 𝑐𝑚

(2) Consider an isothermal, incompressible fluid flowing radially between two concentric
porous spherical shells. Assume steady laminar flow with 𝑉𝑟 = 𝑉𝑟 (𝑟).
(a) Show by use of the equation of continuity that 𝑟 2 𝑉𝑟 = 𝐶𝑜𝑛𝑠𝑡.
(b) Show that the radial pressure distribution may be expressed in terms of 𝑝 as
1 𝑅 4
(𝑝 − 𝑝𝑅 ) = 𝜌𝑉𝑅2 (1 − ( ) )
2 𝑟

(3) In a long cylindrical tube, fluid is initially stationary. At 𝑡 = 0+ , the fluid is allowed to
flow vertically down under the influence of gravity. Assume axis symmetry, and solve for
𝑉𝑧 (𝑡, 𝑟), where 𝑧 is the direction of gravity and 𝑟 is the radial co-ordinate.
𝑚2
If 𝜈 = 3.45 × 10−4 , 𝑅 = 0.7 cm, determine how long it will take for centre – line
𝑠

velocity to reach 90% of the maximum velocity.

19
 (4) Consider a rectangular volume element within the fluid as shown in the following figure:

𝑦
Δ𝑥

τy𝑥 (x, y, z) Δ𝑦
τ𝑥𝑦
𝑥
Show that 𝜏𝑦𝑥 = 𝜏𝑥𝑦 . Both 𝜏𝑦𝑥 𝑎𝑛𝑑 𝜏𝑥𝑦 are shear stresses acting on point (𝑥, 𝑦, 𝑧), i.e.,
show the stresses 𝜏𝑖𝑗 & 𝜏𝑗𝑖 are symmetric.

(5) Obtain a solution for the unsteady tangential flow in a coaxial annulus when the fluid is at
rest for 𝑡 < 0, and the outer cylinder rotates with angular velocity 𝜔 to cause laminar
viscous flow for 𝑡 > 0. Solve for both SS and transient flows.

𝑅1
𝜔
𝑅2

(6) Obtain a similar solution for the unsteady axial flow in a coaxial annulus, when 𝑡 > 0, a
constant Δ𝑃 (𝑃𝑜 − 𝑃𝐿 ) is applied across the cylinder.

(7) For an incompressible, NF and 2D planner flow, Show that vorticity 𝜔 satisfies diffusion
equation:
𝐷𝜔
= 𝜈∇2 𝜔 , similar to the other diffusion equations, namely,
𝐷𝑡
⃗⃗
𝑫𝑽 𝑫𝑻 𝑫𝑪
⃗⃗ ;
= 𝝂𝛁 𝟐 𝑽 = 𝜶𝛁 𝟐 𝐓 ; = 𝑫𝛁 𝟐 𝑪
𝑫𝒕 𝑫𝒕 𝑫𝒕

HW – 3
Q1: It is desired to estimate the rate of heat loss from the meteorological installation shown in
figure below from an experiment on a small geometrically similar model of 1/ 5 the linear
dimensions. The desired surface temperature of the installation and the expected air temperature
and wind velocity and direction are known. Under these expected condition, both free and forced
convection are probably important, and it, therefore, appears reasonable that the reduced velocity
distribution will depend upon both 𝑅𝑒 𝑎𝑛𝑑 𝐺𝑟. It is desired to use air in the model experiment and

20
to maintain similarity between the model and full-scale apparatus by varying air pressure and
velocity. The temperature dependence of fluid properties may be ignored.
(a) What pressure and relative air velocity should be used in the model experiment to maintain
dynamic similarity?
(b) What will be the relative heat – flux form the model?

Q2: Develop governing equations and solve for temperature profiles for unsteady state heat
conduction in solids (a) sphere and (b) cylinder. Initial temperature of solid is 𝑇𝑜 . The outside
surface temperature of solids are maintained constant at 𝑇1 .
Q3. Oil is flowing in the Trans Alaska pipeline in laminar flow. The pipe is fully insulated
(adiabatic walls) in order not to melt the permafrost. Viscous heating is important, since 𝐵𝑟 is
large. The temperature at 𝑧 = 0 is 𝑇𝑜 everywhere. Derive an expression for SS temperature profile
as a function of axial and radial distances. Sketch the temperature profile for large and short values
of z.

HW - 4
BSL Chapter 18
Q1. 18 A.7
Q2. 18 B.2
Q3. 18 C.1. (a, c, d, e). Do not do part (b)!
Q4. 18 C.3.

HW – 5
Q 1. Re-solve BSL 2.5 (flow of two adjacent immiscible fluids)
2. Lecture 17 (condensation of a saturated steam over a flat vertical plate)
a. Solve velocity profile, 𝑣𝑧 (𝑥)
b. Derive the expression for 𝛿𝑓 (𝑥)

21
 c. Calculate local Nusselt number, 𝑁𝑢(𝑥)
4
d. Show that ℎ̅ = 3 (ℎ𝑓 𝑎𝑡 𝑥 = 𝐿)

3. Creeping flow (BSL example 4.2-1)

a. Solve for 𝑉𝑟 , 𝑉𝜃
b. Solve for pressure distribution on sphere
c. Determine Stoke’s law
(Go as far as possible. It is a tedious problem)
𝑧 𝑅
4. (a) Starting with the complex potential 𝜔(𝑧) = −𝑉∞ 𝑅 (𝑅 + 𝑧 ) for the potential flow around a

circular cylinder of radius R, when the approach velocity is 𝑉∞ , show that the form drag is zero on
the cylinder – BSL example 4.3-1.
Re-solve the problem in polar coordinate 𝜓(𝑟, 𝜃) - Deen example 8.3 – 1
5. What is the flow field for the problem of opposite impingement of two incompressible, infinitely
wide, axi-symmetrical cylindrical jets. What is the stream function, and the potential function?
Sketch them. Consider the case where the stagnation plane is flat.

HW – 6
Q1. Relative magnitude of molecular and eddy viscosity:- determine µ(t)/µ(l) at s = R/2 for water
flowing at SS in a long smooth round tube under the following conditions:
𝑘𝑔(𝑓) 𝑘𝑔
R = tube radius = 100 mm, 𝜏0 = 1650 , 𝜌 = 1.25 𝑚3 , 𝜈 = 15 × 10−6 𝑚2 /𝑠𝑒𝑐
𝑚2

Q 2. Estimate the energy dissipation rate in cumulus cloud both per unit mass and for the entire
cloud. Base your estimates on velocity and length scales typical of cumulus clouds. Also estimate
the Kolmogorov microscale, ln. Use 𝜌 = 1.25 𝑘𝑔/𝑚3 and 𝜈 = 15 × 10−6 𝑚2 /𝑠𝑒𝑐
Q 3. The large eddies in turbulent flow have length scale 𝑙, a velocity scale 𝑉(𝑙) = 𝑢, and a time
𝑙
scale t(𝑙) = 𝑢 . The smallest eddies have length scale 𝑛, a velocity scale, 𝜈 and a time scale t.

Estimate the characteristic velocity 𝑉(𝑟) and the characteristic time 𝑡(𝑟) of eddies of size 𝑟, where
𝑟 is any length in the range 𝑛 < 𝑟 < 𝑙. Do this by assuming that 𝑉(𝑟) and 𝑡(𝑟) are determined by
𝜖 and 𝑟 only. Show that your results agree with the known velocity and time scales of 𝑟 =
1
𝑙 and 𝑟 = 𝑛. The energy spectrum of turbulence is a plot of 𝐸(𝐾) = 𝐾 −1 𝑉 2 (𝑘), where 𝐾 = 𝑟 is

the “wave number” associated with eddies of size r. Find an expression for E(K).

22
 Lecture #04
Notes: (1) Property balance is independent of choosing reference frames.

(2) General property balance can also be derived starting with the material rate of change and
using the Reynolds transport equation;

𝐷𝜓 𝜕
= ∭𝑐𝑣 𝜓𝑑∀ + ∬𝑆 𝜓(𝑉𝑠 . 𝑛̂) 𝑑𝑆, where the last term can be written as ∭∀(∇. 𝜓𝑉𝑠 )𝑑∀
𝐷𝑡 𝜕𝑡

So far,

(1) NS equation (𝜌 = 𝑐𝑜𝑛𝑠𝑡, 𝜇 = 𝑐𝑜𝑛𝑠𝑡, 𝑁𝐹)

𝜕𝑉 ⃗⃗
⃗⃗ . ∇)𝑉
𝜌 ( 𝜕𝑡 + (𝑉 ⃗⃗ ) = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗ + 𝜌𝑔⃗ (𝑣𝑒𝑐𝑡𝑜𝑟 𝑒𝑞 𝑛 )
⃗⃗ = 0 : equation of continuity
(2) ∇. 𝑉

⃗⃗ (𝑉𝑥 , 𝑉𝑦 , 𝑉𝑧 ) and
Now, there are 3 momentum (scalar components) and 1 continuity equations. So, 𝑉
𝑝 can be solved.

⃗⃗ using continuity alone without momentum balance, we

Notes: (1) When it is possible to solve 𝑉
have a kinematic solution. In such case, we have a 1st order ODE and we need only one bc for
each non-zero component.

⃗⃗ & 𝑝.
(2) In general, NS and continuity must be solved simultaneously to determine 𝑉
⇒ (a) PDE
(b) Non-linearity on dependent variable: (𝑉⃗⃗⃗. ∇) 𝑉
⃗⃗⃗
𝜕𝑉𝑥
(𝑉𝑥 , 𝑒𝑡𝑐)
𝜕𝑥
Leads to non-stable, chaotic behavior: difficult to solve
Turbulence
⃗⃗
𝜕𝑉
(unsteady state) ⇒ 𝑝𝑎𝑟𝑎𝑏𝑜𝑙𝑖𝑐 𝑤𝑟𝑡 𝑡𝑖𝑚𝑒
𝜕𝑡
In space (𝑥, 𝑦, 𝑧) NS makes it an elliptic equation (difficult to solve)

(𝑐) 𝜇∇2 ⃗⃗⃗

𝑉 = 𝜇∇ (∇. ⃗⃗⃗
𝑉) − 𝜇∇ × (∇ × ⃗⃗⃗
𝑉) = −𝜇∇ × (∇ × ⃗⃗⃗
𝑉) (show it as an exercise)

Viscous term (𝜌 = 𝑐): 𝜇 ∇2 𝑉

⃗⃗⃗ = 0 𝑖𝑓 𝑒𝑖𝑡ℎ𝑒𝑟 𝜇 = 0 (inviscid fluid)
⃗⃗ = 0 𝑜𝑟 𝑏𝑜𝑡ℎ.
𝑜𝑟 ∇ × 𝑉

irrotational

Go one step back, ∇. 𝜏 = 0 : 𝑛𝑜 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑓𝑙𝑢𝑥 (𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒) (shear forces are zero)

23
 ⃗⃗
𝐷𝑉
𝜌 = −∇𝑝 + 𝜌𝑔⃗
𝐷𝑡

⃗⃗⃗
𝜕𝑉 ⃗⃗⃗. ∇) 𝑉
⇒ 𝜌 ( 𝜕𝑡 + (𝑉 ⃗⃗⃗) = −∇𝑝 + 𝜌𝑔
⃗⃗⃗ (𝑛𝑜𝑛 𝑙𝑖𝑛𝑒𝑎𝑟 𝑜𝑛⃗⃗⃗⃗
𝑉)

- Euler’s equation (no viscous effect or term, i.e. fluid is inviscid or irrotational? or both??)

Under SS, integrate along a streamline to derive Bernoulli’s equation (see BSL; to be revisited later in the
course).

Some more on viscosity:

1 𝑇
𝜏 depends on 𝐸 = ( 𝐷 + 𝐷 )
2

Rate of strain-tensor

For NF and 𝜌 = 𝑐:

𝜏 = −𝜇 𝐸 , where scalar 𝜇 ≠ 𝑓(𝐸) ⇒ 𝑙𝑖𝑛𝑒𝑎𝑟. It may be a function of (𝑇, 𝑃)

NNF:

𝜏 depends non-linearly on 𝐸

𝜏 = 𝜂( 𝐸 ) 𝐸

a scalar function of the tensor 𝐸

Thus, 𝜂 must depend on scalar invariants of 𝐸

Such as 𝐼1 = 𝐸 : 𝛿 - scalar

𝐼2 = 𝐸 : 𝐸 - do-

𝐼3 = Det 𝐸 |𝐸| - do-

It turns out that 𝐼1 = 0 for incompressible fluids

⃗⃗ = 0
Since, 𝐸1 : 𝛿 = 2∇. 𝑉

𝜂 = 𝜂 (𝐼2 , 𝐼3 )

Usually, influence of 𝐼3 is negligible,

𝜂 = 𝜂( 𝐸 : 𝐸 )
𝑛−1
eg. 𝜂 = 𝑚(√𝐼2 ) (pseudoplastic, thixotropic)

24
 - Osborn de waals model (BSL)

Note: No elastic property considered so far, only viscosity

Like in viscoelastic fluid.

Dimensional Analysis

The governing conservation equations should be non-dimensionalized before solving:

(a) Dimensionless numbers (groups) resulting out of non-dimensionalization can be used to

make approximation. However, in doing so, the characteristic variables (length, time, etc)
must be correct (physically consistent).
(b) Such parameters can be used as scaling parameters which allows one equation to be used
for both scales of phenomenon.

Reverting to NS equation:

(𝜇1 , 𝜌1 ) (𝜇2 , 𝜌2 )

𝑉2
𝑉1 𝐿1 𝐿2

Geometrical similarity

⃗⃗ )
But two different fluids (𝜇, 𝜌) and velocities (𝑉

Non-dimensionalize as far as possible by choosing the characteristic values of parameters such that each
variable is scaled to be of the order of 1.
𝑥
Define 𝑥 ∗ = 𝑙 : characteristic length 𝑙 is the length over which change takes place in 𝑥 direction, such that
0 < 𝑥 ∗ <1.

25
 𝑦 𝑧
𝑦∗ = , 𝑧∗ =
𝑙 𝑙
𝑡
𝑡𝑉 𝑙
𝑡∗ = 𝑡 𝑙
=
; 𝑡𝑐ℎ𝑎𝑟 = 𝑉
𝑐ℎ𝑎𝑟
𝑉 𝑉 𝑉
𝑉𝑥∗ = 𝑥 , 𝑉𝑦∗ = 𝑦 , 𝑉𝑧∗ = 𝑧 𝐹⁄
𝑉 𝑉 𝑉
𝑝 𝑝 𝑔 𝑚𝑎𝑠𝑠
𝑝∗ = 𝜌𝑉 2 ( ⁄𝑝𝑜 ? ), 𝑔∗ = 𝑔
𝜕 𝜕 𝑙 𝜕
𝑐
9.8 𝑚⁄𝑠 2
{ 𝜕𝑡 ∗ = 𝜕𝑡⁄𝑡𝑐ℎ𝑎𝑟 = 𝑉 𝜕𝑡

{ ∇∗ = 𝑙∇, ∇∗ 2 = 𝑙 2 ∇2 }

⃗⃗
𝐷𝑉
⃗⃗ + 𝜌𝑔⃗
NS: 𝜌 𝐷𝑡 = −∇𝑝 + 𝜇∇2 𝑉

𝜌V2 𝐷𝑉
⃗⃗ ∗ 𝜌V2 ∗ ∗ 𝜇𝑉 ∗ 2 ∗
=− ∇𝑝 + ∇ 𝑉⃗⃗ + 𝜌𝑔∗ 𝑔𝑐
𝑙 𝐷𝑡 ∗ 𝑙 𝑙2

⃗⃗ ∗
𝐷𝑉 𝜇 ∗ 𝑔 𝑙
𝐷𝑡 ∗
= −∇∗ 𝑝∗ + [𝜌𝑉𝑙] ∇∗ 2 𝑉
⃗⃗ ∗ + [ 𝑐2 ] 𝑔∗
𝑉
𝑂(1)
𝑂(1)
𝑂(1)
𝑠𝑐𝑎𝑙𝑒𝑑 𝑡𝑜 𝑜𝑟𝑑𝑒𝑟 1

𝜌𝑉𝑙
(1) [ ] = 𝑅𝑒 (𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑙𝑒𝑠𝑠; 𝑅𝑒𝑦𝑛𝑜𝑙𝑑𝑠 #)
𝜇
𝑉2
(2) [𝑔 𝐿 ] = 𝐹𝑟 (𝐹𝑟𝑜𝑢𝑑𝑒 #)
𝑐

(A) If two different systems have

(a) same dimensionless BCs
(b) same dimensionless equations
They have same dimensionless solution: 𝑉 ∗ (𝑥 ∗ )

(a) is true when they have geometric similarity:

or or

(b) is true when they have dynamic similarity (same 𝑅𝑒 & 𝐹𝑟)

In such case, two systems are consistent with each other as far as scaling up or down is concerned.

(B) 𝑅𝑒 indicates the relative importance of different terms/effects in the equation:

⃗⃗ ∗
𝐷𝑉
𝑖𝑛𝑒𝑟𝑡𝑖𝑎𝑙 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 𝐷𝑡 ∗⁄ 𝜌𝑉𝑙
≈ 𝜇 ≈( )
𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 [𝜌𝑉𝑙 ] ∇∗ 2 𝑉
⃗⃗ ∗ 𝜇

𝑜𝑟𝑑𝑒𝑟 𝑜𝑟𝑑𝑒𝑟

26
 Fr indicates the relative importance of different terms/effects in the equation:
⃗⃗ ∗
𝐷𝑉
𝑖𝑛𝑒𝑟𝑡𝑖𝑎𝑙 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 𝐷𝑡 ∗⁄ 𝑔 𝑙
≈ 𝑔𝑐 𝑙 ≈ ( 𝑉𝑐2 )
𝑔𝑟𝑎𝑣𝑖𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 [ 𝑉2 ] 𝑔∗

𝑜𝑟𝑑𝑒𝑟 𝑜𝑟𝑑𝑒𝑟

So, the non-dimensionalized NS equation becomes

∗
𝐷⃗⃗⃗
𝑉 1 ∗2 ∗ 1 ∗
= −∇ ∗ ∗
𝑝 + ∇ ⃗⃗⃗
𝑉 + 𝑔
𝐷𝑡 ∗ 𝑅𝑒 𝐹𝑟

Case 1: If 𝑅𝑒 ≪ 1 & 𝐹𝑟 ≫ 1 (gravitational effect is negligible)

Then the NS equation becomes

∗ 𝑉2
∇∗ 2 𝑉
⃗⃗⃗ = 0 ⇒ ∇2 𝑉 = 0 (𝐹𝑟 ≡ 𝑔 𝐿)
𝑐

In the above equation, all terms are of the order 1, except coefficient. They determine whether the effects
of terms are negligible or not. If 𝑅𝑒 = 0.1 𝑝𝑒𝑟 𝑠𝑎𝑦, the order becomes 10 and all other terms are ~1 ⇒
𝑐𝑟𝑒𝑒𝑝𝑖𝑛𝑔 𝑓𝑙𝑜𝑤 (viscouse effect dominates)

Case 2: Likewise, 𝑖𝑓 𝑅𝑒 ≫ 1
∗
⃗⃗⃗
𝐷𝑉 1 ∗ 1 ∗ 2 ⃗⃗⃗∗
= −∇∗ 𝑝∗ + 𝑔 ( ∇ 𝑉 drops out, and the modified equation is Euler’s equation )
𝐷𝑡 ∗ 𝐹𝑟 𝑅𝑒

Note: In the latter case 2BCs are gone. Also, recall that, to be able to decide correctly whether 𝑅𝑒 ≫
1 𝑜𝑟 ≪ 1, leading to dropping or retaining the viscous term in the equation, characteristic lengths must
be physically consistent (correct), else there will be enormous error in physical results/data --- to be
discussed later in the context of boundary layer theory.

27
 Lecture #05
Mechanical/Kinetic Energy Balance
𝐽
Balance of mechanical energy: (1⁄2 𝜌𝑉 2 ; 𝑐𝑎𝑙⁄ 3 𝑜𝑟 ⁄ 3 )
𝑚 𝑚

General balance equation: property/cm3-s

If we put 𝜓 = 1⁄2 𝜌𝑉2; the equation cannot distinguish between (non-conductive mode of KE
⃗⃗𝑔 (generation or source of KE) ⇒ needs a different approach
transport) and 𝜓

⃗⃗ into the linear momentum balance or NS equation:

Alternative: Take a dot 𝑉

⃗⃗
𝐷𝑉
⃗⃗ . [𝜌
𝑉 = −∇𝑝 − ∇. 𝜏 + 𝜌𝑔⃗ ]
𝐷𝑡

𝐷𝑉
𝜌𝑉 ⃗⃗ . ∇𝑝 − 𝑉
= −𝑉 ⃗⃗ . ∇. 𝜏 + 𝑉
⃗⃗ . 𝜌𝑔⃗ - eq. (1)
𝐷𝑡

𝐷𝑉 𝐷1⁄2𝑉2 𝜕1⁄2𝑉2 𝑉2
L.H.S. ⇒ 𝜌𝑉 =𝜌 = 𝜌( ⃗⃗ . ∇)
+𝑉
𝐷𝑡 𝐷𝑡 𝜕𝑡 2

𝜕(1⁄2𝜌𝑉2 ) 𝜕𝜌 𝑉2
= + ∇. (1⁄2 𝜌𝑉2 )𝑉
⃗⃗ : add ( + ∇. 𝜌𝑉
⃗⃗ = 0) × (𝜌 ≠ 𝑐)
𝜕𝑡 𝜕𝑡 2

⃗⃗ − 𝑝(−∇. 𝑉
R.H.S. ⇒ −∇. 𝑝𝑉 ⃗⃗ ) − ∇. [𝜏. 𝑉
⃗⃗ ] + [𝜏: ∇𝑉
⃗⃗ ] + 𝜌𝑉𝑔

1𝑠𝑡 𝑡𝑒𝑟𝑚 2𝑛𝑑 𝑡𝑒𝑟𝑚

⃗⃗ = ∇. [𝜏. 𝑉
(H.W. : Show that 𝜏: ∇𝑉 ⃗⃗ ] − 𝑉
⃗⃗ . ∇. 𝜏 )

ME balance:

𝜕(1⁄2𝜌𝑉2 )
⃗⃗ (1⁄ 𝜌𝑉2 ) = −∇. 𝑝𝑉
+ ∇. 𝑉 ⃗⃗ − 𝑝(−∇. 𝑉
⃗⃗ ) − ∇. [𝜏. 𝑉
⃗⃗ ] − [− 𝜏: ∇𝑉
⃗⃗ ] + 𝜌𝑉𝑔 : eq. (2)
𝜕𝑡 2

work done by gravity

𝑢𝑛𝑠𝑡𝑒𝑎𝑑𝑦 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑤𝑜𝑟𝑘 𝑑𝑜𝑛𝑒 𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒 𝑤𝑜𝑟𝑘 𝑑𝑜𝑛𝑒
𝑠𝑡𝑎𝑡𝑒 𝑡𝑒𝑟𝑚 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑏𝑦 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑡𝑜 𝑏𝑦 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑖𝑟𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛
𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦 𝑡𝑜 𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦
𝑓𝑜𝑟𝑐𝑒𝑠
(𝑣𝑖𝑠𝑐𝑜𝑢𝑠 ℎ𝑒𝑎𝑡𝑖𝑛𝑔)
⃗⃗⃗) = 𝜙𝑉2
(−𝜏: ∇𝑉
Now, let us seek the equation of change for internal energy, or total energy (?):
𝑖𝑠 𝑎𝑙𝑤𝑎𝑦𝑠 + 𝑣𝑒
𝐺𝑒𝑛𝑒𝑟𝑎𝑙 (𝑠𝑒𝑒 𝐻. 𝑊. )
+ 𝜓⃗⃗𝑔 ∶ 𝑐𝑎𝑙⁄𝑠 − 𝑚3
𝜕𝜓 ⃗⃗⃗ = −∇.
𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 } ⇒ 𝜕𝑡
+ ∇. 𝜓 𝑉 {
𝑏𝑎𝑙𝑎𝑛𝑐𝑒
28
 1
𝐿𝑒𝑡 𝑢𝑠 𝑠𝑡𝑎𝑟𝑡 𝑤𝑖𝑡ℎ 𝜓 = 𝑇𝑜𝑡𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦 = (𝐼𝐸 + 𝐾𝐸 + 𝑃𝐸)⁄𝑣𝑜𝑙𝑢𝑚𝑒 = 𝜌 (𝑈 + 2 𝑣2 + 𝜙) (J/m3)

𝜓𝑔 = 0 ; = 𝑒𝑛𝑒𝑟𝑔𝑦 𝑓𝑙𝑢𝑥 (𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒); it contains two terms:

(a) 𝑞⃗: 𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛, 𝑟𝑎𝑑𝑖𝑎𝑡𝑖𝑜𝑛 (𝑐𝑎𝑙⁄𝑠 − 𝑚2 )

(b) 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑤𝑜𝑟𝑘 𝑃 . 𝑉⃗⃗ (𝑐𝑎𝑙⁄
𝑠 − 𝑚2 )
𝑑∀ 𝑑𝑙
(Analogous to 𝑝 ( 𝑑𝑡 ) = ∬ 𝑝 (𝑑𝑡) 𝑑𝑆 = ∬ 𝑝𝑉𝑑𝑆 )

⃗⃗ = (𝑝 𝛿 + 𝜏) . 𝑉
𝑃 .𝑉 ⃗⃗ = 𝑝𝑉
⃗⃗ + 𝜏. 𝑉
⃗⃗

⃗⃗ = ∇. 𝑝𝑉
∇. 𝑃 . 𝑉 ⃗⃗ + ∇. 𝜏. 𝑉
⃗⃗

And equation is
𝜕 1 1
⃗⃗ (𝑈 + 𝑣 2 + 𝜙) = −∇. 𝑞⃗ − ∇. 𝑝𝑉
𝜌 (𝑈 + 𝑣 2 + 𝜙) + ∇. 𝜌𝑉 ⃗⃗ + ∇. 𝜏. 𝑉
⃗⃗
𝜕𝑡 2 2

1
𝐷(𝑈+2𝑣2 +𝜙) 𝜕𝜌
𝑜𝑟 𝜌 = −∇. 𝑞 ⃗⃗⃗ + ∇. 𝜏. 𝑉
⃗⃗ − ∇. 𝑝𝑉 ⃗⃗ = 0) 𝐟𝐫𝐨𝐦 LHS term]
⃗⃗⃗ [subtract ( + ∇. 𝜌𝑉
𝐷𝑡 𝜕𝑡

Let 𝝓 (𝑷𝑬) be due to gravity:

𝑔⃗ = −∇𝜙
𝑔𝑟𝑎𝑑 𝑜𝑓
𝑎 𝑠𝑐𝑎𝑙𝑎𝑟

𝐷𝜙 𝜕𝜙 ⃗⃗⃗. ∇𝜙 ⃗⃗⃗. ⃗𝑔
𝜌 𝐷𝑡
=𝜌 𝜕𝑡
+ 𝜌𝑉 = −𝜌𝑉 ⃗⃗

𝐷 1 2 ⃗⃗⃗ + ∇. 𝜏. 𝑉
⃗⃗⃗ + 𝜌𝑉
⃗⃗⃗. 𝑔
𝜌 𝐷𝑡 (𝑈 + 2 𝑉 ) = −∇. 𝑞
⃗⃗ − ∇. 𝑝𝑉 ⃗⃗⃗ - (3)

𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑟 𝑖𝑠 𝑚𝑜𝑣𝑖𝑛𝑔
𝑤𝑖𝑡ℎ 𝑓𝑙𝑢𝑖𝑑

Subtract ME eq. (2) from eq. (3)

𝜕𝑈
𝜌 𝜕𝑡
+ 𝜌 (⃗⃗⃗
𝑉. ∇) 𝑈 ⃗⃗ − 𝑝∇. ⃗⃗⃗
= −∇. 𝑞 ⃗⃗⃗
𝑉 − 𝜏: ∇𝑉

𝑖𝑟𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒
𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛
𝑢𝑛𝑠𝑡𝑒𝑎𝑑𝑦 𝐼𝐸 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟
𝐼𝐸 𝑤𝑜𝑟𝑘 𝑑𝑜𝑛𝑒 𝑡𝑜 𝐼𝐸
𝑐ℎ𝑎𝑛𝑔𝑒 ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟
𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝐼𝐸

29
Inspect equations (2) and (3) (or IE and ME balances)

⃗⃗ is always +ve (IE always increases because of viscous heating; there is nothing
(1) −𝜏: ∇𝑉
called viscous cooling! ME always decreases on the other hand.
− Such energy-change is irreversible.
− When we speak of an isothermal system in a flowing system, it is clear that it is an
assumption. There is no true isothermality per say. Flowing fluid will always be heated.
(2) 𝑝 (−∇. ⃗⃗⃗
𝑉) appears in both eqs as +ve and -ve, but it can be +ve or –ve. Fluid expands: IE ↓ ME ↑
(work done by fluid) or fluid is compressed, IE ↑ ME ↓ (work done on fluid)

Considering T is measurable, IE balance is used as the working equation:

𝐷𝑈
𝜌 ⃗⃗ − 𝑝∇. ⃗⃗⃗
= −∇. 𝑞 ⃗⃗⃗
𝑉 − 𝜏: ∇𝑉
𝐷𝑡 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 ℎ𝑒𝑎𝑡𝑖𝑛𝑔,
𝑖𝑟𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒
(𝑐ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑜𝑛 𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒 𝑤𝑜𝑟𝑘
𝐼𝐸) (𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 (𝑒𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛 𝑜𝑟
𝑓𝑙𝑢𝑥) 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑜𝑛)

(Note that this equation allows chemical reactions, change in composition, 𝜌 ≠ 𝑐)

For now, restrict to no-chemical reaction or non-reactive flow:

Let 𝑈 = 𝑈(𝑇, ∀)

𝜕𝑈 𝜕𝑈
𝑑𝑈 = ( ) 𝑑∀ + ( ) 𝑑𝑇
𝜕∀ 𝑇 𝜕𝑇 ∀

𝜕𝑝
= [−𝑝 + 𝑇 (𝜕𝑇) ] 𝑑∀ + 𝐶𝑉 𝑑𝑇
∀

𝐷𝑈 𝜕𝑝 𝐷∀ 𝐷𝑇
𝜌 𝐷𝑡
= 𝜌 [−𝑝 + 𝑇 (𝜕𝑇) ] 𝐷𝑡 + 𝜌𝐶𝑉 𝐷𝑡
∀

𝐷∀ 𝐷(1⁄𝜌) 𝐷𝜌 1
Now, = = − 1⁄𝜌2 ⃗⃗ ]
= − 𝜌2 [−𝜌∇. 𝑉
𝐷𝑡 𝐷𝑡 𝐷𝑡

Substituting,
𝐷𝑇 ⃗⃗⃗] − 𝑝∇. 𝑉𝜕𝑝 1
⃗⃗⃗ − 𝜏: ∇𝑉
⃗⃗⃗
𝜌𝐶𝑉 𝐷𝑡 = −∇. 𝑞
⃗⃗ − 𝜌 [−𝑝 + 𝑇 ] [ ∇. 𝑉
𝜕𝑇 𝜌 ∀

𝐷𝑇 𝜕𝑝
⃗⃗ − 𝑇 ( ) ∇. ⃗⃗⃗
𝜌𝐶𝑉 𝐷𝑡 = −∇. 𝑞 ⃗⃗⃗
𝑉 − 𝜏: ∇𝑉 (no reaction)
𝜕𝑇 ∀

One can also show,

30
 𝐷𝑇 𝑇 𝜕∀ 𝐷𝑝 ⃗⃗⃗
𝜌𝐶𝑝
𝐷𝑡
= −∇. 𝑞
⃗⃗ − ( )
∀ 𝜕𝑇
− 𝜏: ∇𝑉
𝑝 𝐷𝑡

⃗⃗ :
Now, we need to relate 𝑞⃗ 𝑡𝑜 𝑇 via the constitutive equation, similar to 𝜏 relating to 𝑉

⃗⃗ + ∇𝑉
eg. 𝜏 = 𝜇[∇𝑉 ⃗⃗ 𝑇 ] (𝐹𝑜𝑟 𝑁𝐹)

⃗⃗ = +𝜇[∇𝑉
and −𝜏: ∇𝑉 ⃗⃗ + ∇𝑉
⃗⃗ 𝑇 ]: ∇𝑉
⃗⃗ = +𝑣𝑒 (𝜙𝑉2 )

Fourier’s Law, 𝑞⃗ = −𝑘∇𝑇𝑛̂

property of material

𝐷𝑇 𝜕𝑝 ⃗⃗⃗ + 𝜙𝑉2
𝜌𝐶𝑉 𝐷𝑡 = ∇. 𝑘∇𝑇𝑛
̂ − 𝑇 ( ) ∇𝑉
𝜕𝑇
(4a)
∀

⃗⃗ = 0 𝑓𝑜𝑟 𝑖𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑)

(∇. 𝑉

Therefore, if 𝜌 = 𝑐, 𝑘 = 𝑐𝑜𝑛𝑠𝑡, 𝜙2𝑉 = negligible,

𝐷𝑇 𝜕𝑇
𝜌𝐶𝑉 𝐷𝑡 = 𝜌𝐶𝑉 ( 𝜕𝑡 + ⃗⃗⃗
𝑉. ∇T) = 𝑘∇2 𝑇 (4b)

If we allow for a chemical reaction, the equation is easily modified as

𝐷𝑇 𝜕𝑇 ⃗⃗⃗. ∇T) = 𝑘∇2 𝑇 + (𝑟)(−∆𝐻) (cal/s-cm3)

𝜌𝐶𝑉
𝐷𝑡
= 𝜌𝐶𝑉 ( 𝜕𝑡 + 𝑉 (4c);

, where r is the rate of reaction in moles/s-cm3 and ∆𝐻 is the heat of reaction in cal/moles, which could
be +ve or –ve depending on the endothermic or exothermic reaction, respectively. However, we require
an additional species conservation equation (to be discussed later). Features of the equation are

(1) PDE
⃗⃗ )
(2) linear in T unlike momentum equation (non-linear on 𝑉
(3) elliptic in space
(4) scalar equation
Recall NS:

𝜕⃗⃗⃗
𝑉 ⃗⃗⃗. ∇)𝑉
⃗⃗⃗) = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗⃗ + 𝜌𝑔
𝜌 ( 𝜕𝑡 + (𝑉 ⃗⃗⃗

⃗⃗ , 𝑝, 𝑇 (𝑛𝑜𝑛 − 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙, 𝑏𝑢𝑡 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛, 𝑤ℎ𝑦? )

Now, the variables are 𝑉

𝑡ℎ𝑒 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠 are also 3: Continuity, NS, Energy

And,
𝑝ℎ𝑦𝑠𝑖𝑐𝑎𝑙 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦: 𝜇 = 𝜇(𝑇), 𝑘 = 𝑘(𝑇)
𝐺𝑎𝑠 𝑙𝑎𝑤: 𝑝 = 𝜌𝑅𝑇 𝑜𝑟 𝑝 = 𝑝(𝜌, 𝑇) }
𝑜𝑟 𝜌 = 𝜌𝑜 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 / 𝑖𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑, 𝑜𝑟 𝑙𝑖𝑞𝑢𝑖𝑑

31
 Lecture #06
Coupling between energy & momentum equations: Free convection or buoyance - driven flow

⃗⃗ = 0)
(Incompressible fluid: ∇. 𝑉

⃗⃗
𝐷𝑉
𝜌 𝐷𝑡 = −∇. 𝜏 − ∇𝑝 + 𝜌𝑔⃗ (1) (𝑝 is the actual pressure and not dynamic pressure)

Two extreme situations of coupling (or no-cpupling)

(1) In forced convection, ′∇𝑝′ is due to an externally applied source term for momentum generation.
In free convection, no external source or pressure is imposed and ∇𝑝 is the same with or without
flow. In such case, solve for velocity fields from NS equation and superimpose the solutions or
velocity fields on energy balance equation---thus there is no coupling.
(2) In free convection, difference in 𝜌(𝑥, 𝑦, 𝑧) arises because of ∇𝑇 (or concentration in a mixture),
but without 𝑔⃗ natural convection can not take place.
⇒ ∇𝑝 = 𝜌𝑔 ≡ static pressure difference
mean density 𝑎𝑡(𝑇) → 𝑢𝑛𝑘𝑛𝑜𝑤𝑛 (to be solved; specific to the situation)
Flow is not induced because of ′∇𝑝′ per say, but because of 𝜌 𝑔⃗ relative to 𝜌𝑔⃗, where 𝜌 is the local
fluid density. In other words, each fluid layer rises (or falls) because of its density that is different
from the neighboring fluid layer ( 𝜌 )
𝜌(𝑧)
Therefore,
𝜌
⃗⃗
𝐷𝑉
𝜌 𝐷𝑡 = −𝜌𝑔⃗ − ∇. 𝜏 + 𝜌𝑔⃗ 𝑔

= −∇. 𝜏 − 𝑔⃗(𝜌 − 𝜌)

Δ𝜌
Boussinesq approximation (1): ≪1 maximum change in 𝜌
𝜌

𝜕𝜌
Now, 𝜌 = 𝜌 + ( ) (𝑇 − 𝑇 ) - Taylor’s series (leading term only) and assuming 𝜌 ≠ 𝜌(𝑝)
𝜕𝑇 𝑝

= 𝜌 − 𝜌 𝛽(𝑇 − 𝑇 )

1 𝜕𝜌
coeff. of volume expansion. (𝛽 = − 𝜌 (𝜕𝑇 ) )
𝑇

The modified momentum conservation equation or NS equation

⃗⃗
𝐷𝑉
(1) 𝜌 = −∇. 𝜏 − 𝜌 𝛽𝑔⃗(𝑇 − 𝑇 ) (2)
𝐷𝑡
buoyancy force due to Δ𝑇 (do not mix: 𝜌 ≠ 𝜌̅ )
⃗⃗ = 0 (𝜌 = constant wrt 𝑝 but not with respect to 𝑇)
(2) ∇. 𝑉

32
Boussinesq approximation (2): 𝜌 can be replaced everywhere by the constant value 𝜌̅ , except in the
buoyancy term. Therefore, for a constant 𝜌 and 𝜇 fluid:

𝐷⃗⃗⃗
𝑉 ⃗⃗⃗ − 𝑔
𝐷𝑡
= 𝜈∇2 𝑉 ⃗⃗⃗𝛽(𝑇 − 𝑇 ) (free convection)

Note: 𝑇 is unknown and so is 𝜌(𝑇) and is different from situation (problem) to situation (problem). Energy
balance equation is required to solve for 𝑇 𝑎𝑛𝑑 𝑇:

𝐷𝑇
𝜌𝐶𝑝 = 𝑘∇2 𝑇 + 𝜙𝑉2 (𝑁𝑒𝑤𝑡𝑜𝑛𝑖𝑎𝑛 𝑓𝑙𝑢𝑖𝑑)
𝐷𝑡

𝐷𝑇
= 𝛼∇2 𝑇 (neglecting 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛 𝑡𝑒𝑟𝑚)
𝐷𝑡

⇒ Without Boussinesq approximation 2, momentum conservation equation for free convection:

⃗⃗⃗
𝐷𝑉
𝜌 𝐷𝑡 = 𝜇∇2 ⃗⃗⃗
𝑉−𝜌𝑔⃗⃗⃗𝛽 (𝑇 − 𝑇) (𝜌 ≠ 𝜌)

Note: 𝜌 is not cancelled out, and as per Deen, all solutions have considered approximation 2

A general conservation equation can also be written for mixed convection (forced + free convection):

𝐷⃗⃗⃗
𝑉 ∇p
=− − ⃗⃗⃗𝛽(𝑇 − 𝑇 ) + 𝜈∇2 ⃗⃗⃗
𝑔 𝑉
𝐷𝑡 𝜌

where, p ≡ defined as 𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 ⇒ ∇p = ∇𝑝 − 𝜌𝑔 (note difference between p and p).

This also implies that dynamic pressure difference is zero in a static fluid.

To this end, for a constant 𝜌 & 𝜇 (𝑁𝐹) fluid:

⃗⃗ = 0
Continuity: ∇. 𝑉

⃗⃗⃗
𝐷𝑉
Momentum: 𝜌 𝐷𝑡 = −∇𝑝 + 𝜇∇2 ⃗⃗⃗
𝑉 + 𝜌𝑔
⃗⃗⃗ (𝑓𝑜𝑟𝑐𝑒𝑑 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛)

⃗⃗ − 𝜌 𝛽𝑔⃗(𝑇 − 𝑇 ) (𝑓𝑟𝑒𝑒 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛)

= 𝜇∇2 𝑉
𝐷𝑇
Energy: 𝜌𝐶𝑝 𝐷𝑡 = 𝑘∇2 𝑇 + 𝜙𝑉2

⃗⃗ , 𝑝, 𝑇 for forced convection

Variables are: 𝑉

⃗⃗ and 𝑇 for free convection and 𝜌 (𝑇 )

and 𝑉

Dimensional Analysis:

(A) Forced Convection: Need Characteristic variables for 𝑙, 𝑉, 𝑡, 𝑝, 𝑇

𝑉 𝑥,𝑦,𝑧 𝑡 𝑇−𝑇
𝑣 ∗ = 𝑉 ; 𝑥 ∗, 𝑦 ∗, 𝑧 ∗ = ; 𝑡 ∗ = 𝑙 ; 𝑇 ∗ = 𝑇 −𝑇𝑜 : O(1)
𝑙 ⁄ 1 𝑜
𝑉

33
 ∗
Continuity: ∇∗ . ⃗⃗⃗
𝑉 =0

𝐷𝑉 ∗ 1 1
NS: = ∇∗ 2 𝑉 ∗ − ∇∗ 𝑝 ∗ + 𝑔∗
𝐷𝑡 ∗ 𝑅𝑒 𝐹𝑟

𝐷𝑇 ∗ 1 𝐵𝑟
Energy: = ∇∗ 2 𝑇 ∗ + 𝜙∗2
𝑉
(show it as an exercise)
𝐷𝑡 𝑅𝑒𝑃𝑟 𝑅𝑒𝑃𝑟
2 2
𝑙𝑐 𝑉 𝜌 𝑉 𝜌𝑉 ⁄𝑙𝑐 𝜇 ⁄𝜌 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑑𝑖𝑓𝑓. 𝑐𝑜𝑒𝑓𝑓.
𝑅𝑒 = , 𝐹𝑟 = = , 𝑃𝑟 = 𝜈⁄𝛼 = =
𝜇 𝑔𝑐 𝑙𝑐 𝜌𝑔𝑐 𝑘 ⁄𝜌𝐶𝑝 𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑑𝑖𝑓𝑓. 𝑐𝑜𝑒𝑓𝑓.

𝑔𝑎𝑠𝑒𝑠 𝑝𝑟 ≈ 0.7, 𝑙𝑖𝑞𝑢𝑖𝑑 (𝑤𝑎𝑡𝑒𝑟, 𝐻𝑔) 𝑃𝑟 ≈ 1000

Get physical meanings of the dimensionless groups (relative importance or effect of different terms in
the transport governing equation or conservation equations):
2
𝐵𝑟 ∗2 1 𝜇𝑉
𝐵𝑟 ≡ 𝐵𝑟𝑖𝑛𝑘𝑚𝑎𝑛 # = 𝜙𝑉 / ∇∗ 2 𝑇 ∗ =
𝑅𝑒𝑃𝑟 𝑅𝑒𝑃𝑟 𝑘(𝑇1 − 𝑇𝑜 )
𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑 𝑏𝑦 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛
=
𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑜𝑓 𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑒𝑛𝑒𝑟𝑔𝑦

𝐷𝑇 ∗ 1 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑒𝑓𝑓𝑒𝑐𝑡
𝑅𝑒𝑃𝑟 = 𝑃𝑒𝑐𝑙𝑒𝑡 # = ⁄ ∇∗ 2 𝑇 ∗ = (𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑒𝑓𝑓𝑒𝑐𝑡) of heat transport
𝐷𝑡 ∗ 𝑅𝑒𝑃𝑟

1 1 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡

𝑃𝑟 = (∇∗ 2 𝑉 ∗ )⁄ ∇∗ 2 𝑇 ∗ = (another defn? same as before )
𝑅𝑒 𝑅𝑒𝑃𝑟 ℎ𝑒𝑎𝑡−𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡

Non-dimensionalization: (a) Ascertains relative importance of each feature/effect in equation.

(b) Determines scale between prototype and full dynamical similarity
(c) Determines analogy between momentum and heat transfer. Needs identical non-dimensional
equations.

(B) Free convection: Needs characteristic length, 𝑙 (vertical layer), 𝑎𝑛𝑑 velocity, 𝑣𝑐 =?
𝜌𝑉𝑙
1. 𝑅𝑒 = 𝜇
≈ 1 (viscosity prevents convective current from increasing; a steady – state 𝑉𝑠 is attained:
𝜇
𝑉𝑠 ≈ )
𝜌𝑙

𝜌𝑙2
2. 𝑡 = 𝑙⁄𝑉 = is the time to reach 𝑆𝑆 velocity 𝑉𝑠
𝑠 𝜇

𝐷𝑉 ∗
Get 𝐷𝑡 ∗
= ∇∗ 2 𝑉 ∗ − 𝑇 ∗ 𝐺𝑟𝑔∗ (dimensionless form of momentum balance in free convection)

2
𝑔𝜌 𝛽(𝑇1 −𝑇𝑜 )𝑙3 𝑇−𝑇
where, 𝐺𝑟 ≡ 𝐺𝑟𝑎𝑠ℎ𝑜𝑓𝑓 # = vertical length; (𝑇 ∗ = 𝑇 −𝑇𝑜 )
𝜇2 1 𝑜

(relative importance of buoyancy with respect to viscose effects in natural convection.

34
 Lecture #07
Equation of Change for Species
𝑚𝑜𝑙𝑒𝑠 𝑔
Property, 𝜓 ≡ 𝐶𝑖 ( ), 𝜌𝑖 ( ), 𝑖 = 1,2,3 𝑠𝑝𝑒𝑐𝑖𝑒𝑠
𝑐𝑚3 𝑐𝑚3

𝑔 𝑚𝑜𝑙𝑒𝑠
= 𝑓𝑙𝑢𝑥 𝑜𝑓 𝜓 = 𝑛𝑖 (𝑐𝑚2 −𝑠) 𝑜𝑟 𝑁𝑖 (𝑐𝑚2 −𝑠) ; 𝑖𝑛𝑐𝑙𝑢𝑑𝑒𝑠 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑎𝑛𝑑 𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑚𝑜𝑑𝑒𝑠

General property balance:

𝜕𝜌𝑖 𝜕𝐶𝑖
𝜕𝑡
+ ∇. 𝑛𝑖 = 𝑟𝑖 𝑜𝑟 𝜕𝑡
+ ∇. 𝑁𝑖 = 𝑅𝑖 (𝑛𝑖 , 𝑁𝑖 𝑎𝑟𝑒 𝑣𝑒𝑐𝑡𝑜𝑟𝑠) eq. (1)

𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒⃗⃗⃗⃗⃗
𝜓𝑔
( 𝑜𝑓 𝑠𝑝𝑒𝑐𝑖𝑒𝑠 𝑖 𝑖𝑛 ); g/s-cm3 or moles/s-cm3
𝑐ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛

∑𝑖( ) = ∑𝑖(𝑟𝑖 ) = 0 ∑𝑖( ) = ∑𝑖(𝑅𝑖 ) ≠ 0 : sum over all species (i) in reaction rates

(mass is conserved and not moles)

⇒ As noted earlier, 𝑛𝑖 and 𝑁𝑖 contain both convective and diffusion fluxes.

𝒈⁄𝒔 − 𝒄𝒎𝟐 : 𝑛𝑖 = 𝜌𝑖 𝑣𝑖 = 𝜌𝑖 ⃗V⃗ + 𝑗𝑖 = 𝜔𝑖 ∑ 𝑛𝑖 + 𝑗𝑖 = 𝜔𝑖 𝑛 𝑇 + 𝑗𝑖

𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑜𝑓 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑚𝑎𝑠𝑠 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛

′𝑖 ′ 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑓𝑙𝑢𝑥 𝑓𝑙𝑢𝑥 𝜌𝑖 /𝜌

∑ 𝜌𝑖 𝑣𝑖
That brings definition for ⃗V⃗ = 𝑙𝑜𝑐𝑎𝑙 𝑚𝑎𝑠𝑠 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = ∑𝑖 𝜌𝑖 (=𝜌)
(bulk flow velocity)

Sum over all species,

𝜕𝜌𝑖 𝜕𝜌
∑𝑖 ( + ∇. 𝑛𝑖 ) = ∑ 𝑟𝑖 ⇒ + ∇. 𝑛 𝑇 = 0; : summation over all species in eq. (1)
𝜕𝑡 𝜕𝑡

This equation is similar to the mass conservation equation for a pure fluid

𝒎𝒐𝒍𝒆𝒔⁄𝒔 − 𝒄𝒎𝟐 : 𝑁𝑖 = 𝐶𝑖 𝑣𝑖 = 𝐶𝑖 𝑉 ⃗⃗ ∗ = 𝑚𝑜𝑙𝑎𝑟 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = ∑ 𝐶𝑖 𝑣𝑖

⃗⃗ ∗ + 𝐽𝑖 ⇒ 𝑉
∑𝐶 𝑖

molar diffusion flux

𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑎𝑠 ∑ 𝐽𝑖 = 0 (and ∑ 𝑗𝑖 = 0)

35
 𝜕𝐶𝑖 𝜕𝐶
∑𝑖 ( + ∇. 𝑁𝑖 ) = ∑ 𝑅𝑖 ⇒ + ∇. 𝑁𝑇 = ∑ 𝑅𝑖 ≠ 0 summation over all species in eq. (1)
𝜕𝑡 𝜕𝑡

(mass is conserved and not moles)

Binary diffusion (A & B only):

𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝐴
𝐽𝐴 = −𝐶 𝐷𝐴𝐵 ∇𝑦𝐴 → 𝐹𝑖𝑐𝑘 ′ 𝑠 1𝑠𝑡 𝑙𝑎𝑤 ( 𝑠−𝑐𝑚2
)
𝑜𝑟 𝐹𝑖𝑐𝑘 ′ 𝑠 𝑙𝑎𝑤 𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑡𝑜 𝑡ℎ𝑒
𝑔 𝑜𝑓 𝐴
𝑗𝐴 = −𝜌 𝐷𝐴𝐵 ∇𝜔𝐴 = −(𝐶 2 ⁄𝜌)𝑀𝐴 𝑀𝐵 𝐷𝐴𝐵 ∇𝑦𝐴 ; (𝑠−𝑐𝑚2 ) 𝑜𝑏𝑠𝑒𝑟𝑣𝑒 𝑚𝑜𝑣𝑖𝑛𝑔 𝑎𝑡
𝑓𝑙𝑢𝑖𝑑′ 𝑠 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦
(𝐶, 𝜌 ≡ 𝑡𝑜𝑡𝑎𝑙)
}

⃗⃗); 𝐽1 = 𝐶1 (𝑣1 − V
Also, note that 𝑗𝑖 = 𝜌𝑖 (𝑣𝑖 − V ⃗⃗ ∗ ) (diffusion flux is relative to bulk avg velocity;
intertial frame)

⇒ Show that BSL Table 17.8 – 2: given terms for 𝑗𝐴 , 𝑗𝐴 ∗ , etc are equivalent for binary mixture.

General species balance:

𝜕𝜌𝐴
𝑔⁄𝑠 − 𝑐𝑚3 : 𝜕𝑡
+ ∇. 𝜌𝐴 ⃗V⃗ = ∇. 𝜌 𝐷𝐴𝐵 ∇𝜔𝐴 + 𝑟𝐴
𝑗𝐴
mass avg. velocity

𝜕𝐶𝐴
𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑐𝑚3 : ⃗⃗ ∗ = ∇. 𝐶 𝐷𝐴𝐵 ∇𝑦𝐴 + 𝑅𝐴
+ ∇. 𝐶𝐴 V
𝜕𝑡

molar avg. velocity

Special cases:

(A) Const 𝜌, 𝐷𝐴𝐵 (𝑙𝑖𝑞𝑢𝑖𝑑 𝑚𝑎𝑦 𝑏𝑒 𝑎𝑡 𝑟𝑒𝑠𝑡)

𝜕𝜌𝐴
𝜕𝑡
+ ρA ∇. ⃗V⃗ + ⃗V⃗ . ∇ρA = ∇ DAB ∇ρA + rA
( ∇. ⃗V⃗ = 0)

𝜕𝜌𝐴
+ ⃗V⃗. ∇ρA = DAB ∇2 ρA + rA
𝜕𝑡 𝑔⁄𝑠 − 𝑐𝑚3
Divided by 𝑚𝜔 (𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑤𝑡 𝑜𝑓 𝐴)

𝜕𝐶𝐴
⃗⃗ . ∇CA = DAB ∇2 CA + R A
+V 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑐𝑚3
𝜕𝑡 : This is often used as working equation for species balance
(𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒)
(B) Const C & 𝐷𝐴𝐵 (dilute gas at const T & p)
𝜕𝐶𝐴
𝜕𝑡
+ CA ∇.⃗⃗⃗⃗⃗
V ∗ + ⃗V⃗ ∗ . ∇CA = 𝐶𝐷𝐴𝐵 ∇2 yA + R A 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑐𝑚3
1
⃗⃗ ∗ ≠ 0 =
But ∇. V (𝑅𝐴 + 𝑅𝐵 )
𝐶

36
 If 𝑅𝐴 = −𝑅𝐵 (special case, equal molar) ⇒ ∇. ⃗V⃗ ∗ = 0

𝜕𝐶𝐴 𝐶𝐴
Therefore, 𝜕𝑡
+ ⃗V⃗ ∗ . ∇𝐶𝐴 = 𝐶𝐷𝐴𝐵 ∇2 𝑦𝐴 + 𝑅𝐴 − 𝐶
(𝑅𝐴 + 𝑅𝐵 )

⃗⃗ → 0 𝑜𝑟 V
(C ) 𝑉 ⃗⃗ ∗ → 0 and no reaction, constant properties.

𝜕𝐶𝐴
𝜕𝑡
= 𝐷𝐴𝐵 ∇2 𝐶𝐴 − Fick’s 2nd law! (Nothing special; it should not be termed law)

𝜕𝑥𝐴
𝑜𝑟, = 𝐷𝐴𝐵 ∇2 𝑥𝐴
𝜕𝑡

Let us explore coupling between momentum and species balance equation:

⃗⃗ = 0
Continuity: ∇. 𝑉 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝜌)

⃗⃗
𝐷𝑉
⃗⃗ − ∇𝑝 + 𝜌𝑔⃗
Momentum (NS): 𝜌 𝐷𝑡 = 𝜇∇2 𝑉 (Forced convection)

⃗⃗
𝐷𝑉
⃗⃗ − ∇p − 𝜌̅ 𝛽𝑔⃗(𝑇 − 𝑇𝑜 ) (Mixed convection) (note difference between p and p)
𝜌 𝐷𝑡 = 𝜇∇2 𝑉

⃗⃗
𝐷𝑉
𝜌 ⃗⃗ − 𝜌 𝑔⃗𝛽(𝑇 − 𝑇)
= 𝜇∇2 𝑉 (Free convection)
𝐷𝑡

Recall:

𝜕𝜌 𝜕𝜌
𝜌 = 𝜌𝑜 + ( ) (𝑇 − 𝑇𝑜 ) + ( ) (𝐶𝑖 − 𝐶𝑖𝑜 ) (Taylor series expansion of multivariables)
𝜕𝑇 𝑇𝑜 ,𝐶𝑖0 𝜕𝐶𝑖 𝑇 ,𝐶
𝑜 𝑖0

If T = constant, we have multi-component mixture under isothermal condition)

⃗⃗
𝐷𝑉
𝜌 ⃗⃗ − 𝜌
= 𝜇∇2 𝑉 ̅𝛽𝐴 ⃗𝑔
⃗⃗(𝑥𝐴 − 𝑥𝐴𝑜 ) (Free convection)
𝐷𝑡

1 𝜕𝜌
( )
𝜌 𝜕𝐶𝐴 𝑇 ,𝐶
𝑜 𝐴0

𝑔𝜌̅ 2 𝛽𝐴 (𝑥𝐴 − 𝑥𝐴𝑜 )l3

(𝐺𝑟 = , another Grashoff number based on concentration difference)
𝜇𝑓2

Note: constant 𝜌, 𝜇, 𝑘, 𝐷𝐴𝐵 𝑓𝑙𝑢𝑖𝑑 𝑖𝑠 termed constant physical properties fluid.

⃗⃗ ∗ = 0
Continuity ∇∗ . 𝑉 (1)
⃗⃗ ∗
𝐷𝑉 1 ∗ 1
NS = −∇∗ 𝑝∗ + 𝑅𝑒 ∇∗ 2 V
⃗⃗ + 𝑔∗ (2)
𝐷𝑡 𝐹𝑟

𝐷𝑇 ∗ 1 𝐵𝑟
Energy = 𝑅𝑒𝑃𝑟 ∇∗ 2 𝑇 ∗ + 𝑅𝑒𝑃𝑟 𝜙𝑟∗ (3)
𝐷𝑡

37
 𝐷𝐶𝐴∗ 1
Species = 𝑅𝑒𝑆𝑐 ∇∗ 2 𝐶𝐴∗ (no reaction) (4a) (by analogy, check)
𝐷𝑡

𝐶 −𝐶 𝜇
where, 𝐶𝐴∗ = 𝐶 𝐴 −𝐶𝐴𝑜 , similar to T* defined earlier and 𝑆𝑐 = ⁄𝜌𝐷
𝐴1 𝐴𝑜

- Analogy between (3) and (4) if reaction included as the source term (RA*) in eq. (4a)

If 𝑷𝒓 ≈ 𝑺𝒄 and in the absence of viscous heating and chemical reaction- there is a good
similarity between heat & mass transport.

𝐷𝑇 ∗ 1 ∗2 ∗ 𝐷𝐶𝐴∗ 1 ∗2 ∗
= ∇ 𝑇 and = ∇ 𝐶𝐴
𝐷𝑡 𝑃𝑒ℎ 𝐷𝑡 𝑃𝑒𝑚

(𝑃𝑒ℎ & 𝑃𝑒𝑚 𝑎𝑟𝑒 𝑃𝑒𝑐𝑙𝑒𝑡#)

𝜇 𝜇
𝑆𝑐 = ⁄𝜌𝐷 , 𝑃𝑟 = ⁄𝜌𝛼 ≈ 0.7 (𝑔𝑎𝑠𝑒𝑠)

𝜈⁄ , 𝜈⁄ ≈ 1000 (𝑙𝑖𝑞𝑢𝑖𝑑)
𝐷 𝛼

1⁄ (∇∗ 2 𝑉
⃗⃗ ∗ ) 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡
𝑆𝑐 = 𝑅𝑒 =
1⁄ ∗2 ∗ 𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 (𝑠𝑝𝑒𝑐𝑖𝑒𝑠) 𝑏𝑦 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛
𝑅𝑒𝑆𝑐 (∇ 𝐶𝐴 )

𝐺𝑒𝑛𝑒𝑟𝑎𝑙 (𝑐𝑜𝑚𝑚𝑜𝑛)𝑓𝑜𝑟𝑚 𝑜𝑓 𝑠𝑝𝑒𝑐𝑖𝑒𝑠 𝑐𝑜𝑛𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛:

𝐷𝐶𝐴∗ 1 ∗
𝐷𝑡
= 𝑃𝑒 ∇∗ 2 𝐶𝐴∗ + 𝑅𝐴 (4b)
𝑚

(A definition for 𝑃𝑒𝑚 can also be written in the similar fashion as was written for 𝑃𝑒ℎ in the previous
lecture on the non-dimensionalized form of energy balance)

This lecture brings an end to the development of four conservation equations used in TP, namely:
continuity, momentum, energy, and species, to solve for four variables p, V, T, and C, with additional
gas law, if required, ρ = ρ(p, T). The conservation equations we developed also contain three
constitutive types of equations for diffusion or conduction flux:

𝜏 = −𝜇∇𝑉 = −𝜐∇(ρ𝑉) (momentum/s-cm2): Newton’s 1st law of viscosity

𝑞 = −𝑘∇𝑇 = −𝛼∇(𝑇𝜌𝐶𝑝) (cal/s-cm2): Fourier’s 1st Law of conduction

𝐽𝐴 = −𝐷𝐴𝐵 ∇𝐶𝐴 (moles/s-cm2): Fick’s first law of diffusion

(Note that the last term in each equation shows the grad of property-concentrations: momentum,
thermal energy, and moles each expressed per unit volume. Thus, the flux of the property is
proportional to the grad of property-concentration, where the proportionality constant is diffusion
coefficient or diffusivity having the unit of cm2/s in each case)

38
 Lecture #08
Prototype example 1: Rayleigh problem or Stoke’s first problem
(Flow near a plate suddenly set in motion; Jerked plate problem)

(t = 0-: a long horizontal plate and fluid on top of it, both at rest)

t = 0: plate is moved horizontally at a constant velocity; fluid is at rest.

𝑦
𝑡 𝑣𝑥 (𝑡, 𝑦) = ? 𝑡 > 0
(𝑢𝑛𝑠𝑡𝑒𝑎𝑑𝑦 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑟𝑜𝑓𝑖𝑙𝑒𝑠)

𝑥
𝑉𝑝
𝜕𝑣𝑥 𝜕𝑣𝑦 𝜕𝑣𝑥
⃗⃗ = 0;
Continuity: ∇. 𝑉 + =0⇒ = 0 (symmetric from −∞ 𝑡𝑜 ∞ along 𝑥 )
𝜕𝑥 𝜕𝑦 𝜕𝑥

𝜕𝑣𝑦
𝜕𝑦
= 0 ⇒ 𝑣𝑦 ≠ 𝑓(𝑦) = 𝑐𝑜𝑛𝑠𝑡 ⇒ 𝑣𝑦 = 0 𝑎𝑡 𝑦 = 0 (𝑛𝑜 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛) ⇒ 𝑣𝑦 = 0 everywhere. With
vz also zero, we are, therefore, discussing a unidirectional unsteady-state flow, viz 𝑣𝑥 (𝑡, 𝑦) = ?

𝜕⃗⃗⃗
𝑉 ⃗⃗⃗. ∇ 𝑉
⃗⃗⃗) = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗⃗ + 𝜌𝑔
NS: 𝜌 ( 𝜕𝑡 + 𝑉 ⃗⃗⃗ (𝜌 = 𝑐, 𝜇 = 𝑐, 𝑁𝐹)

𝜕𝑣 𝜕𝑣𝑥 𝜕𝑣𝑥 𝜕𝑣𝑥 𝜕𝑝 𝜕2 𝑣 𝜕2 𝑣𝑥 𝜕2 𝑣𝑥

𝑥 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ 𝜕𝑡𝑥 + 𝑣𝑥 + 𝑣𝑦 + 𝑣𝑧 ] = − 𝜕𝑥 + 𝜇 [ 𝜕𝑥2𝑥 + + ] + 𝜌𝑔𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦2 𝜕𝑧2

2𝐷 (𝑧 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐) (no external pressure imposed;

symmetric along x-direction)
no

𝜕𝑣𝑥 𝜕2 𝑣𝑥 𝜕𝑣𝑥 𝜕2 𝑣𝑥
𝜌 =𝜇 ⇒ =𝜈
𝜕𝑡 𝜕𝑦 2 𝜕𝑡 𝜕𝑦 2

Some inspection: Parabolic PDE, t = 0- 𝑣𝑥 = 0 for all 𝑦 ≥ 0; 𝐼𝐶

on time

39
 𝑛𝑜 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛

𝑡=0 𝑦 = 0 𝑣𝑥 = 𝑉𝑝
} 𝐵𝐶𝑠
→ ∞ 𝑣𝑥 = 0

very far from the

{ plate, fluid remains
undisturbed/stationary

Some words on no-slip condition: It is an experimental observation that under continuum conditions, the
relative velocity between the solid and fluid in contact with the solid is zero. If solid is stationary, fluid (last
layer) is also stationary. If solid moves, the layer in contact assumes the velocity of the solid.

What method? Note semi-infinite domain (𝑦 = 0 ∶ ∞);

𝑎𝑛𝑑, 𝐼𝐶 = one of the two BCs ⇒ Combination of variables

𝑡 𝑎𝑛𝑑 𝑦; combine them to one, 𝜂.

(Separation of variables method will not work)

𝑣𝑥
First, non-dimensionalize 𝑣𝑥 ⇒ 𝜙 =
𝑉𝑝

𝜕𝜙 𝜕2 𝜙
=𝜈
𝜕𝑡 𝜕𝑦 2

Let 𝜂 = 𝛼𝑦𝑡 𝑎 (𝛼, 𝑎 𝑎𝑟𝑒 #; 𝑤ℎ𝑦 𝑛𝑜𝑡 𝑦 𝑎 𝑡 ? )

𝜕𝜙 𝜕𝜙 𝜕𝜂 𝜕𝜙
𝜕𝑡
= .
𝜕𝜂 𝜕𝑡
= 𝛼 𝜕𝜂 𝑎𝑦𝑡 𝑎−1 It will not work for any value of a!

𝜕𝜙 𝜕𝜙 𝜕𝜂 𝜕𝜙
𝜕𝑦
= .
𝜕𝜂 𝜕𝑦
= 𝜕𝜂
𝛼𝑡 𝑎

𝜕2 𝜙 𝜕 𝜕𝜙 𝜕 𝜕𝜙 𝜕𝜂 𝜕2 𝜙
𝜕𝑦 2
= 𝜕𝑦 ( 𝜕𝜂 ) 𝛼𝑡 𝑎 = 𝜕𝜂 ( 𝜕𝜂 ) 𝛼𝑡 𝑎 𝜕𝑦 = 𝛼 2 𝑡 2𝑎 𝜕𝜂2

Substitute,

𝜕𝜙 𝜕2 𝜙
𝛼 𝑎𝑦𝑡 𝑎−1 = 𝜈𝛼 2 𝑡 2𝑎
𝜕𝜂 𝜕𝜂2

𝜂
𝑦 = ⁄𝛼𝑡 𝑎

𝜕𝜙 𝜂 𝜕2 𝜙
Therefore, 𝛼 𝜕𝜂 𝑎 𝛼𝑡 𝑎 𝑡 𝑎−1 = 𝜈𝛼 2 𝑡 2𝑎 𝜕𝜂2

In order to have 𝑡 vanish from the equation to solve, collect all t-terms:

𝑡 𝑎−1
𝑡 𝑎 𝑡 2𝑎
= 𝑡 0 = 1 ⇒ 𝑎 = − 1⁄2

40
 𝑦
𝜂=𝛼 ⁄
√𝑡

Equation now becomes;

𝑑𝜙 1 𝑑2 𝜙
𝑑𝜂
(− 2) 𝜂 = 𝜈𝛼 2 𝑑𝜂2

𝑑2 𝜙 𝜂 𝑑𝜙
+ = 0 (𝑎 𝑙𝑖𝑛𝑒𝑎𝑟 𝑂𝐷𝐸)
𝑑𝜂2 2𝛼 2 𝜈 𝑑𝜂

1
Set 2𝛼 2 𝜈
= 2 (why? An arbitrary constant so that the method works!)

𝜂 = 0 (𝑦 = 0), 𝜙 = 1
𝑑2 𝜙 𝑑𝜙 𝑦→∞
𝑑𝜂2
+ 2𝜂 𝑑𝜂 =0 ⇒ →∞ ( ) , 𝜙 = 0} 𝑐𝑜𝑛𝑠𝑖𝑠𝑡𝑒𝑛𝑐𝑖𝑒𝑠 𝑏𝑒𝑓𝑜𝑟𝑒 & 𝑎𝑓𝑡𝑒𝑟 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚𝑎𝑖𝑡𝑜𝑛
𝑡=0
↑

(IC collapses with one of two BCs: this is how Combination of variables work and PDE is converted to
ODE)
𝑑𝜙 𝑑𝜓 2
Let 𝜓 = 𝑑𝜂
: 𝑑𝜂
+ 2𝜂𝜓 = 0 ⇒ 𝜓 = 𝐶1 𝑒 −𝜂

𝜂 2
Therefore, 𝜙 = 𝐶1 ∫0 𝑒 −𝜂 𝑑𝜂 + 𝐶2

1 = 𝐶1 (0) + 𝐶2 ⇒ 𝐶2 = 1 (𝐵𝐶 1)
𝜂 2
𝜙 = 𝐶1 ∫0 𝑒 −𝜂 𝑑𝜂 + 1

∞ 2 1
0 = 𝐶1 ∫0 𝑒 −𝜂 𝑑𝜂 + 1 𝐵𝐶(2) ⇒ 𝐶1 = − ∞ 2
∫0 𝑒 −𝜂 𝑑𝜂

𝜂 2
∫0 𝑒 −𝜂 𝑑𝜂 2 𝜂 −𝜂2
𝜙 =1− ∞ 2 =1− ∫ 𝑒 𝑑𝜂 = 1 − erf(𝜂)
∫0 𝑒 −𝜂 𝑑𝜂 √𝜋 0

𝐸𝑟𝑟𝑜𝑟 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛

1.0

erf(𝜂)

η
𝛼𝑦 𝑦
𝑣𝑥 = 𝑉𝑝 (1 − 𝑒𝑟𝑓( 𝑡 )) = 𝑉𝑝 (1 − 𝑒𝑟𝑓 4𝜈𝑡
)
√ √

𝑣𝑥
At what location, = 0.01 at a time t??
𝑉𝑝

𝛿 𝛿
0.01 = 1 − 𝑒𝑟𝑓 ( )⇒ = 2 𝑓𝑟𝑜𝑚 𝐸𝑟𝑟𝑜𝑟 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛
√4𝜈𝑡 √4𝜈𝑡

41
 𝟐
𝛿 = 4√𝜈𝑡 𝑓𝑜𝑟 𝑣𝑥 = 1% 𝑜𝑓 𝑉𝑝 (𝑜𝑟 𝒕 ∝ 𝜹 ⁄𝝂)

Edge of disturbance in 𝑦 −direction from plate due to plate movement: fluid above is nearly stationary

⇒ The penetration depth has a square root of time-dependence.

𝛿 ⇒ 𝑇ℎ𝑖𝑛𝑘 𝑜𝑓 𝑡𝑎𝑘𝑖𝑛𝑔 𝑠𝑛𝑎𝑝𝑠ℎ𝑜𝑡𝑠

𝑡 𝑎𝑡 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝑡𝑖𝑚𝑒𝑠 > 0
𝛿
𝛿
⇒ 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑠𝑖𝑚𝑖𝑙𝑎𝑟𝑖𝑡𝑦 𝑖𝑛 𝑡ℎ𝑒
𝑉𝑝 → 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑟𝑜𝑓𝑖𝑙𝑒𝑠

𝑙𝑎𝑠𝑡 𝑙𝑎𝑦𝑒𝑟 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑 𝑖𝑠 𝑎𝑙𝑤𝑎𝑦𝑠 𝑎𝑡 𝑉𝑝 𝑎𝑛𝑑

𝐵𝐶𝑠 { 𝑡ℎ𝑒 𝑙𝑎𝑦𝑒𝑟 𝑎𝑡 ′∞′ 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑖𝑠 𝑠𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑟𝑦
𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡𝑖𝑚𝑒𝑠.

𝑄𝑢𝑒𝑠𝑡𝑖𝑜𝑛 𝑐𝑎𝑛 𝑏𝑒 𝑝𝑜𝑠𝑒𝑑 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑙𝑦: 𝑎𝑡 𝑎 𝑓𝑖𝑥𝑒𝑑 ′𝛿 ′ , 𝑑𝑒𝑠𝑐𝑟𝑖𝑏𝑒 𝑣𝑥 (𝑡) ⇒

𝑡 ′𝛿′

𝑣𝑥
𝑉𝑝
Notes: (1) Method is called “Combination of variables” or “Similarity transform”: requires (a) a semi-
infinite domain in space and (b) IC & 1 BC should be the same (match)

(2) Physical intuition suggests profiles 𝑉𝑥 (𝑡, 𝑦) are similar.

(3) x − momentum 𝑑𝑖𝑓𝑓𝑢𝑠𝑒𝑠 in ′y′ direction (although no ∇p was imposed in x − direction)
𝜕𝑉𝑥
(𝜏𝑦𝑥 = 𝜇 ) How about 𝜏𝑥𝑦 , considering that shear stress is symmetric? Which momentum
𝜕𝑦
now diffuses in which direction? There is an ambiguity in such question. The definition of 𝜏𝑦𝑥
should remain the same as discussed in the previous lectures: x is the direction of force
(momentum) and y is the direction of the normal to the plane the force acts.
(4) Analogous examples in heat & mass transport (freezing of a water-lake or ice formation in a
refrigerator – tray or mass transfer in a falling film)
𝜕𝑇 ∗ 𝜕2 𝑇 ∗ 𝜕𝐶 ∗ 𝜕2 𝐶 ∗
𝜕𝑡 ∗
= 𝛼 𝜕𝑦∗2 𝑜𝑟 𝜕𝑡 ∗
= 𝐷 𝜕𝑦∗2 (Fourier’s or Fick’s 2nd law?)

See the analogous diffusion equations in momentum, heat and mass transport. But ensure that
the non-dimensionalized initial and boundary conditions are also identically the same to be able
to use the same non-dimensionalized solutions as derived above.

42
 Lecture #09
Prototype example 2: Unsteady-state laminar flow or start-up flow in a tube

𝑅 𝑙
𝑟
𝑧
𝑃𝑜 𝑃𝑙

t = 0- : fluid is stationary
∇𝑝
𝒕 = 𝟎 ∶ 𝑖𝑚𝑝𝑜𝑠𝑒 , 𝑓𝑙𝑜𝑤 𝑠𝑡𝑎𝑟𝑡𝑠 ⇒ 𝑣𝑧 (𝑡, 𝑟) = ? at t = 0+
𝑙

Restriction 𝜌 = 𝑐𝑜𝑛𝑠𝑡, 𝑁𝐹, 𝜇 = 𝑐, 𝑎𝑥𝑖𝑠 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐.

1 𝜕(𝑟𝑣𝑟 ) 𝜕𝑣
⃗⃗ = 0):
Solution: Continuity: (∇. 𝑉 + 𝜕𝑧𝑧 = 0; 𝑣Ө = 0 ∶ Ө − 𝑜𝑟 𝑎𝑥𝑖𝑠 − 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐
𝑟 𝜕𝑟

(cylindrical co-ordinate)

Also, vr = 0: r − symmetric. (Note that if a conical tube vr ≠ 0; flow is not r-symmetric.)

𝜕𝑣
Therefore, 𝜕𝑧𝑧 = 0 (from Continuity). Alternatively, assuming a fully developed flow in z-direction (not on
𝜕𝑣𝑧
time-domain) or neglecting the entrance-exit effects in the tube, 𝜕𝑧
= 0. This implies that (r𝑣𝑟 ) is a
constant (from Continuity). Since, 𝑣𝑟 𝑖𝑠 0 at the tube wall 𝑜𝑟 @ 𝑟 = 𝑅, it is 0 everywhere. In any case,
flow is unidirectional.

NS: (z momentum)

𝜕𝑣𝑧 𝜕𝑣𝑧 𝜕𝑣𝑧 𝜕𝑝 𝜕2 𝑣 1 𝜕 𝜕𝑣𝑧

𝜌 + 𝜌 (𝑣𝑧 + 𝑣𝑟 ) = − 𝜕𝑧 + 𝜇 [ 𝜕𝑧2𝑧 + 𝑟 𝜕𝑟 𝑟 ] + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑧 𝜕𝑟 𝜕𝑟

𝜕𝑣𝑧 𝜕p 1 𝜕 𝜕𝑣𝑧
𝜌 =− +𝜇 (𝑟 ) : Work on hydrodynamic pressure: ∇p = ∇𝑝 − 𝜌𝑔 so that the solution is
𝜕𝑡 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟

the same in a horizontal or vertical tube due to ∇𝑝, or without referring to the direction of gravity.

𝜕p 𝑑p Pl −Po Po −Pl
𝜕𝑧
= 𝑑𝑧 = 𝑙
=− 𝑙
(linear assumption for pressure gradient? )

(Considering that vz is a function of r only and the hydrodynamic pressure p is the function of z only, the
assumption is not required. Also, one can write r- and Ө − momentum equations with 𝑣𝑧 = 𝑣𝑟 = 0, and
consider p separate from actual pressure p to show that p = p(z) only )

Non-dimensionalize:

43
 𝑣𝑧 𝑣𝑧 One can solve SS flow:
𝜙= = Δ𝑝𝑅2
𝑣𝑚𝑎𝑥 ? ( ) ∂vz ΔpR2
𝑎𝑡 𝑡 → ∞ 4𝜇𝑙 (𝑃𝑢𝑡 ∂t
= 0 and integrate the above equation twice in r − direction: vmax = 4μl )
(𝑆𝑆) − 𝐇𝐚𝐠𝐞𝐧 − 𝐏𝐨𝐢𝐬𝐞𝐮𝐢𝐥𝐥𝐞′ 𝐬 𝐞𝐪𝐧 .

How about 𝑡𝑐 (𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝑡𝑖𝑚𝑒)?

𝑟
𝜉=𝑅 , 𝜏 = 𝑡⁄𝑡 (𝑡𝑐 ≠ 𝑙⁄𝑣𝑚𝑎𝑥 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑙 𝑖𝑠 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦 ? )
𝑐

𝑡𝑐 remains unknown; but remember that if all but one characteristic variables are known, the remaining
can be calculated/estimated:

𝜌 Δ𝑝𝑅 2 𝜕𝜙 Δ𝑝 μ Δ𝑝 1 ∂ 𝜕𝜙
Substitute, = + 𝑅2 4𝜇𝑙 𝑅 2 𝜉 𝜕𝜉 𝜉 𝜕𝜉
𝑡𝑐 4𝜇𝑙 𝜕𝜏 𝑙

𝜌 𝑅 2 𝜕𝜙 1 ∂ 𝜕𝜙
(𝑡 ) 𝜕𝜏 = 4 + 𝜉 𝜕𝜉 𝜉 𝜕𝜉
𝑐 𝜇
𝑚2
𝜌𝑅2 𝑅2
𝑡𝑐 = 𝜇
= 𝜈
: time to reach SS. Flow profile is bounded in 𝑟 direction; 𝑣 = 0 𝑎𝑡 𝑟 = 𝑅 (no slip-

𝑚 2 ⁄𝑠 condition). Δ𝑝 is applied axially (in 𝑧 direction), profiles develop in ‘𝑟’ direction

𝑅2
‘by momentum conduction/diffusion? 𝑡𝑐 is the time for the fluid at the center line to reach 𝑣𝑚𝑎𝑥, = 𝜈 .

𝑅2
Alternatively, if you are convinced that 𝑡𝑐 or characteristic time = 𝜈
because pressure is imposed in z-
𝜌𝑅2 𝑣𝑧
direction and radial velocity is zero, you can start with 𝜏 = 𝑡⁄𝑡 with 𝑡𝑐 = 𝜇
and 𝜙 = , you will get
𝑐 𝑣𝑐
Δ𝑝𝑅 2
the same result for 𝑣𝑐 as . To this end, the non-dimensionalized equation is
4𝜇𝑙

𝜕𝜙 1 ∂ 𝜕𝜙
= 4 + 𝜉 𝜕𝜉 (𝜉 𝜕𝜉 )
𝜕𝜏

𝐼𝐶: 𝜏 = 0 𝜙 = 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝜉

𝜉 = 1, 𝜙 = 0 (𝑛𝑜 − 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

𝐵𝐶𝑠 (𝑓𝑜𝑟 𝑎𝑙𝑙 𝜏 > 0): 𝜕𝜙
𝜉 = 0, 𝜙 = 𝑓𝑖𝑛𝑖𝑡𝑒 (𝑣𝑚𝑎𝑥 ) ⇒ 𝜕𝜉
= 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)

Notes: (1)Similarity Transform or Combination of variables– No good because we have a bounded domain.

(2) Separation of variables? No because the equation is non-homogeneous. You will not be able to
separate: 𝜙(𝜏, 𝜉) ≠ 𝜙(𝜉)𝑇(𝜏)

Therefore, remove non-homogeneity first:

𝜙(𝜏, 𝜉) = 𝜙𝑆𝑆 (𝜉) − 𝜙𝑡 (𝜏, 𝜉)

𝑆𝑆 𝑡𝑒𝑟𝑚 departure function from SS function? Or a residual

function?? Or, just a mathematical term??

44
In any case, it is clear that 𝜙(𝜏, 𝜉) has been defined such that 𝜙𝑡 gradually vanishes with increasing 𝜏 so
that the SS solution is 𝜙𝑆𝑆 and we have a fully developed SS velocity profile.

𝜕𝜙𝑡 1 d 𝑑𝜙𝑆𝑆 1 ∂ 𝜕𝜙𝑡

Plugging, − =4+ 𝜉 − 𝜉
𝜕𝜏 𝜉 𝑑𝜉 𝑑𝜉 𝜉 𝜕𝜉 𝜕𝜉

1 d 𝑑𝜙𝑆𝑆
𝝓𝑺𝑺 is the solution of 0 = 4 + 𝜉 ⇒ 𝜙𝑆𝑆 = 1 − 𝜉 2 (𝑃𝑜𝑖𝑠𝑒𝑢𝑙𝑙𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑟𝑜𝑓𝑖𝑙𝑒)
𝜉 𝑑𝜉 𝑑𝜉

𝜕𝜙𝑡 1 ∂ 𝜕𝜙𝑡
𝝓𝒕 is the solution of = 𝜉
𝜕𝜏 𝜉 𝜕𝜉 𝜕𝜉
𝑰𝑪: 𝜏 = 0 𝜙𝑡 = 𝜙𝑆𝑆 (𝑐ℎ𝑒𝑐𝑘 𝐼𝐶 𝑜𝑛 𝜙 )
𝜕𝜙𝑡
𝑩𝑪𝒔: 𝜉 = 0 𝜙𝑡 = 𝑓𝑖𝑛𝑖𝑡𝑒; 𝜕𝜉
=0
=1 𝜙𝑡 = 0

(Check 𝜙(𝜏, 𝜉) = 𝜙𝑆𝑆 (𝜉) − 𝜙𝑡 (𝜏, 𝜉). Also check IC and BCs on 𝜙.)

𝜙 = 1 − 𝜉 2 − 𝜙𝑡 (𝜏, 𝜉)

Now TRY 𝜙𝑡 (𝜏, 𝜉) = 𝑇(𝜏)(𝜉) (Separation of variables: the method works if the domain is bounded,
and the governing equations as well as initial/boundary conditions are homogenous).

𝑑𝑇 1 d 𝑑
 𝑑𝜏 = 𝑇 𝜉 𝑑𝜉 𝜉 𝑑𝜉

1 𝑑𝑇 1 1 d 𝑑 2𝜏 1 1 d 𝑑
𝑇 𝑑𝜏
=  (𝜉 𝑑𝜉 𝜉 𝑑𝜉) = −𝛼 2 ; ⤏ 𝑇 = 𝐶𝑜 𝑒 −𝛼 𝑎𝑛𝑑  (𝜉 𝑑𝜉 𝜉 𝑑𝜉) = −𝛼 2

𝑑2  𝑑
𝜉 2 𝑑𝜉2 + 𝜉 𝑑𝜉 + 𝜉 2 𝛼 2  = 0

- Homogeneous linear ODE

The equation is a special form of

𝑥 2 𝑦 ′′ + 𝑥𝑦 ′ + (𝑥 2 − 𝑛2 )𝑦 = 0 with n = 0

- For Bessel’s equation of order 𝑛 = 0. General solution is

𝑦 = 𝐶1 𝐽𝑛 (𝑥) + 𝐶2 𝐾𝑛 (𝑥)

BF of 1st kind BF of 2nd kind

𝑛 = 0:

𝐽𝑜

2.4 𝑥
5.5 8.654
𝑣𝑎𝑟𝑖𝑜𝑢𝑠 𝑟𝑜𝑜𝑡𝑠
𝐾𝑜

45
Here (on comparison), 𝑥 = 𝛼𝜉, 𝑦=

 = C1 Jo (𝛼𝜉) + C2 K o (𝛼𝜉)
2
Therefore, 𝜙𝑡 (𝜏, 𝜉) = 𝐶𝑜 𝑒 −𝛼 𝜏 [C1 Jo (𝛼𝜉) + C2 K o (𝛼𝜉)]
2
= 𝐶𝑜 𝑒 −𝛼 𝜏 [Jo (𝛼𝜉) + C2 K o (𝛼𝜉)] (C0, C2 , and 𝛼 are to be determined)

𝑩𝑪 𝟏. 𝜉 = 0 𝜙𝑡 is finite for all 𝜏 ⇒ 𝐶2 = 0 (Note Ko is infinite at 𝜉 = 0)

2
𝝓𝒕 = 𝐴𝑒 −𝛼 𝜏 Jo (𝛼𝜉)

𝑩𝑪 𝟐. 𝜉 = 1 𝜙𝑡 = 0 ⇒ Jo (𝛼) = 0

𝛼 has many roots ⇒ 𝛼𝑛

Sum over all solutions:

2
𝜙𝑡 = ∑∞
𝑛=1 𝐴𝑛 𝑒
−𝛼𝑛 𝜏 (𝛼
Jo 𝑛 𝜉)

𝑰𝑪 𝜏 = 0, 𝜙𝑡 = 𝜙𝑆𝑆

𝑇ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒, 1 − 𝜉 2 = ∑∞
𝑛=1 𝐴𝑛 Jo (𝛼𝑛 𝜉)

Use orthogonal property to calculate 𝐴𝑛

1
∫0 𝐽𝑛 (𝛼𝑖 𝑥)𝐽𝑛 (𝛼𝐽 𝑥 )𝑥𝑑𝑥 = 0 (𝑖 ≠ 𝑗)

1
= 2 [𝐽𝑛′ (𝛼)]2 (𝑖 = 𝑗)

1
∫0 (1 − 𝜉 2 )𝐽𝑜 (𝛼𝑚 𝜉) 𝜉𝑑𝜉 = ∑∞
𝑛=1 𝐴𝑛 Jo (𝛼𝑛 𝜉) Jo (𝛼𝑚 𝜉)𝜉𝑑𝜉

𝐴𝑚 4𝐽1 (𝛼𝑚 ) 𝐴𝑚 8
= [𝐽1 (𝛼𝑚 )]2 ⇒ 3 = [𝐽1 (𝛼𝑚 )]2 ⇒ 𝐴𝑚 = 3 𝐽 (𝛼 )
2 𝛼𝑚 2 𝛼𝑚 1 𝑚

Solution is,
8 2
𝜙𝑡 = ∑∞
𝑛=1 𝛼 3 𝐽 𝑒 −𝛼𝑛 𝜏 Jo (𝛼𝑛 𝜉)
𝑛 1 (𝛼𝑛 )

8 2𝜏
𝜙 = (1 − 𝜉 2 ) − ∑∞
𝑛=1 3 𝐽 (𝛼 ) 𝑒
−𝛼𝑛
Jo (𝛼𝑛 𝜉) 𝛼𝑛 = 2.4, 5.5 ….
𝛼𝑛 1 𝑛

*H.W. if 𝐾 ≠ 0 or flow in annulus:

𝑟1 𝑟1
𝐾= ⁄𝑟2 ≠ 0

𝑟2

46
Solved velocity profiles 𝜙(𝜏, 𝜉):

(See the old version of BSL book. There are table and figure to determine or estimate the time 𝜏 for the
centerline velocity to reach certain % SS velocity. The maximum velocity is within 10% of SS value when
𝜏 = 0.45, 𝑜𝑟 𝜙 = 0.9 𝑎𝑡 𝜏 = 0.45 and 𝜉 = 0. )

Reference for Bessel functions: Advanced Engineering Mathematics by Erwin Kreyszig

 Bessel’s equation of order n:

𝑥 2 𝑦 ′′ + 𝑥𝑦 ′ + (𝑥 2 − 𝑛2 )𝑦 = 0 (𝑛 ≥ 0)
Solution is,
(−1)𝑚 𝑥 2𝑚
𝐽𝑛 (𝑥) = 𝑥 𝑛 ∑∞
𝑚=0 22𝑚+𝑛 𝑚! 𝑚+𝑛!

− Bessel Function of the 1st Kind

(−1)𝑚 𝑥 2𝑚
𝐽−𝑛 (𝑥) = 𝑥 −𝑛 ∑∞
𝑚=0 22𝑚−𝑛 𝑚! 𝑚−𝑛!

But if n is an integer then

𝐽𝑛 (𝑥) = (−1)𝑛 𝐽−𝑛 (𝑥)

Therefore, if n is not an integer

𝑌(𝑥) = 𝐶1 𝐽𝑛 (𝑥) + 𝐶2 𝐽−𝑛 (𝑥) will be a general solution of the Bessel equation.

If n is an integer then;

𝑌(𝑥) = 𝐶1 𝐽𝑛 (𝑥) + 𝐶2 𝐾𝑛 (𝑥) forms the general solution, where 𝐾𝑛 (𝑥) is the Bessel function of the 2nd kind
and given by

2 𝑥 𝑥𝑛 ∞ (−1)𝑚−1 (ℎ𝑚 +ℎ𝑚+𝑛 )𝑥 2𝑚 𝑥 −𝑛 𝑛−1 𝑛−𝑚−1! 𝑥 2𝑚

𝑌(𝑥) = 𝜋 𝐽𝑛 (𝑥) (𝑙𝑛 2 + 𝑌) + 𝜋
∑𝑚=0
22𝑚+𝑛 𝑚! 𝑚+𝑛!
− 𝜋
∑𝑚=0 2𝑚−𝑛
2 𝑚!

1 1 1
𝑥 > 0, ℎ𝑜 = 0, ℎ𝑠 = 1 + 2 + 3 + ⋯ 𝑠
(𝑠 ≠ 0)

 Orthogonal Property of Bessel Function

47
 1
∫0 𝐽𝑛 (𝛼𝑖 𝑥)𝐽𝑛 (𝛼𝐽 𝑥 )𝑥𝑑𝑥 = 0 (𝑖 ≠ 𝑗)

1
= 2 [𝐽𝑛′ (𝛼)]2 (𝑖 = 𝑗)

 Some useful relationships

1) [𝑥 𝑛 𝐽𝑛 (𝑥)]′ = 𝑥 𝑛 𝐽𝑛−1 (𝑥)
[𝑥 −𝑛 𝐽−𝑛 (𝑥)]′ = −𝑥 −𝑛 𝐽𝑛+1 (𝑥)

2𝑛
2) 𝐽𝑛−1 (𝑥) + 𝐽𝑛+1 (𝑥) = 𝐽 (𝑥)
𝑥 𝑛
𝐽𝑛−1 (𝑥) − 𝐽𝑛+1 (𝑥) = 2 𝐽𝑛 ′(𝑥)

3) 𝐽𝑜′ (𝑥) = −𝐽1 (𝑥)

𝐽1′ (𝑥) = 𝐽𝑜 (𝑥) − 𝑥 −1 𝐽1 (𝑥)

4) ∫ 𝑥 𝑛 𝐽𝑛−1 (𝑥)𝑑𝑥 = 𝑥 𝑛 𝐽𝑛 (𝑥) + 𝐶

∫ 𝑥 −𝑛 𝐽𝑛+1 (𝑥)𝑑𝑥 = −𝑥 −𝑛 𝐽𝑛 (𝑥) + 𝐶
∫ 𝐽𝑛+1 (𝑥)𝑑𝑥 = ∫ 𝐽𝑛−1 (𝑥)𝑑𝑥 − 2𝐽𝑛 (𝑥)

For Homework problems:

5) If 𝑆𝑜 (𝛼𝑛 𝜉) = 𝐾𝑜 (𝛼𝑛 𝐾)𝐽𝑜 (𝛼𝑛 𝜉) − 𝐽𝑜 (𝛼𝑛 𝐾)𝐾𝑜 (𝛼𝑛 𝜉)

1 1
Then ∫𝐾 𝑆𝑜 (𝛼𝑛 𝜉)𝑆𝑜 (𝛼𝑚 ) 𝜉𝑑𝜉 = 2 [𝑆12 (𝛼𝑚 ) − 𝐾 2 𝑆12 (𝛼𝑚 𝐾)]
1 1
∫𝐾 𝜉 𝑆𝑜 (𝛼𝑚 𝜉) 𝑑𝜉 = 𝛼 [𝑆1 (𝛼𝑚 ) − 𝐾𝑆1 (𝛼𝑚 𝐾)]
𝑚

6) If 𝑍1 (𝛼𝑛 𝜉) = 𝐽1 (𝛼𝑛 𝜉)𝐾1 (𝛼𝑛 𝐾) − 𝐽1 (𝛼𝑛 𝐾)𝐾1 (𝛼𝑛 𝜉)

1 1
Then ∫𝐾 [𝑍1 (𝛼𝑛 𝜉)]2 𝜉𝑑𝜉 = 2 [𝑍𝑜2 (𝛼𝑛 ) − 𝐾 2 𝑍𝑜2 (𝛼𝑛 𝐾)]
1 1
∫𝐾 𝑍1 (𝛼𝑛 𝜉) 𝑑𝜉 = − 𝛼 [𝑍𝑜 (𝛼𝑛 ) − 𝑍𝑜 (𝛼𝑛 𝐾)]
𝑛

48
 Lecture #10
Prototype example 3: Flow in a permeable tube or bundles of hollow fibers
(Lubrication Approximation)
The lubrication approximations were originally applied to address the flow of a thin film of fluid between
two solids in relative motion, for example, flow of lubricating oils in slider bearing or sliding cylinder (See
Deen 6.6). However, the approximations are also applicable to the flow in a tapered channel (or
cylinders/tubes) with narrow gaps:

, and also, to the present example of flow in a permeable tube or hollow fiber of a membrane (Deen):

Assumptions: No entrance effect, SS, walls are permeable and vw is known as the permeate flux, 𝐽𝜔 =
𝑄𝜔
⁄(2𝜋𝑅𝐿)

𝑣𝜔
𝐿

𝑟 𝑅
𝑝1 𝑝2
𝑣𝑜
𝑥
𝑣𝜔
𝑝𝑒𝑟𝑚𝑒𝑎𝑏𝑙𝑒 (𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑒/ 𝑝𝑒𝑟𝑚𝑒𝑎𝑡𝑒)
𝑤𝑎𝑙𝑙𝑠

Velocities and pressure in the tube? (axisymmetric, 𝑣Ɵ = 0)

𝜕𝑝
𝑣𝑥 (𝑥, 𝑟) =?, 𝑣𝑟 (𝑥, 𝑟) =?, =?
𝜕𝑥

𝐍𝐨𝐭𝐞: In the present case, there are two non − zero velocities, unlike in the previous two prototype examples
where there was one non − zero velocity component (unidirectional flow) to solve for.
Recall the previous start − up flow example: (1) Inertial term, v. ∇v = 0 (no inertial effect) or
𝜕𝑣𝑥 𝜕𝑣𝑥 𝜕𝑣𝑥
𝑣𝑥 𝜕𝑥
+ 𝑣𝑟 𝜕𝑟
= 0 because we had a fully developed flow with 𝜕𝑥
= 0, implying
⇒ 𝑟𝑣𝑟 = constant (from Continuity) ⇒ Therefore, 𝑣𝑟 = 0 (because of no − slip conditon at walls)
∂p dp
(2) Also, = dx = constant
∂x
𝜕𝑣 2
⃗⃗ ] = − dp + 𝜇 [𝜕 𝑣2𝑥 + 1 𝜕 𝑟 𝜕𝑣𝑥 ]
⃗⃗ . 𝛻𝑉
Therefore, the governing NS equation was modified as ρ [ 𝜕𝑡𝑥 + 𝑉
{ dx 𝜕𝑥 𝑟 𝜕𝑟 𝜕𝑟

𝜕𝑣𝑥 1 𝜕(𝑟𝑣𝑟 )
In the present case, again start with Continuity: 𝜕𝑥
+𝑟 𝜕𝑟
=0

49
 𝜕𝑣𝑥 𝜕(𝑟𝑣𝑟 )
Considering 𝜕𝑥
≠ 0 as the permeate stream flows out of the tube, 𝜕𝑟
≠ 0. Therefore, 𝑣𝑟 ≠ 0 (total
flowrates in (x + r) directions are conserved). All it means is that the flow is not unidirectional and we have
to solve for both 𝑣𝑥 and 𝑣𝑟 .
𝜕𝑣𝑥 𝜕𝑣𝑥 𝜕p 𝜕2 𝑣 1 𝜕 𝜕𝑣𝑥
NS equations: 𝑥-momentum: 𝜌 (𝑣𝑥 𝜕𝑥
+ 𝑣𝑟 𝜕𝑟
) = − 𝜕𝑥 + 𝜇 ( 𝜕𝑥 2𝑥 + 𝑟 𝜕𝑟 (𝑟 𝜕𝑟
))

𝜕𝑣𝑟 𝜕𝑣𝑟 𝜕p 𝜕 2 𝑣𝑟 1 𝜕 𝜕(𝑟𝑣𝑟 )

𝑟 − momentum: 𝜌 (𝑣𝑥 + 𝑣𝑟 )= − +𝜇( 2 + ( ))
𝜕𝑥 𝜕𝑟 𝜕𝑟 𝜕𝑥 𝑟 𝜕𝑟 𝜕𝑟
And, 𝑣𝑥 = 𝑣𝑥 (𝑥, 𝑟) =? and 𝑣𝑟 = 𝑣𝑟 (𝑥, 𝑟) = ? (it is convenient to work on hydrodynamic pressure, p)

Approximations (or physical intuitions?)

⃗⃗ . ∇𝑉
1. Inertial effects are negligible 𝑉 ⃗⃗ ≈ 0 (flow is usually small)
𝜕2 𝑣𝑥 1 𝜕 𝜕𝑣
2. viscose − axial term is smaller than the corresponding radial term: ≪ (𝑟 𝑥 )
𝜕𝑥 2 𝑟 𝜕𝑟 𝜕𝑟
𝜕p 𝜕p
3. p = p(𝑥) 𝑜𝑟 𝜕𝑟
≪ 𝜕𝑥
𝜇 𝜕 𝜕𝑣 dp
x-momentum: 𝑟 𝜕𝑟
(𝑟 𝜕𝑟𝑥 ) = dx : Lubrication approximations

It should be mentioned that a physical intuition for the present flow-situation should also indicate the
same approximations. The majority or primary flow in a fiber or membrane tube is indeed in x-direction.
The permeate flowrates (flow in r-direction) are usually small, implying that the inertial term is negligible.
The situation is similar to creeping flow or flow around a sphere at low Reynolds number, when the inertial
term is neglected in comparison to viscose-term. With permeate flow being secondary or smaller than the
axial or primary flow, 𝑣𝑟 ≪ 𝑣𝑥 , and the flow may be considered to be 𝑎𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑒𝑙𝑦 uni −
directional or 1D. Similarly, pressure-drop occurs mostly in x-direction to overcome the viscous effects,
𝜕p 𝜕p
and that in r-direction is small, or 𝜕𝑟
≪ 𝜕𝑥
. Therefore, p = p(𝑥) only. Scaling analysis can also be used
to examine consistency in the above assumptions.

Revisit Continuity and NS equations:

𝑣𝑟 𝑅 𝑅 ≪ 𝐿𝑥 of the hollow fiber

~ ⇒ } From continuity
𝑣𝑥 𝐿𝑥 Therefore, 𝑣𝑟 ≪ 𝑣𝑥

main flow-direction (Note 𝒗𝒓 is small, but not zero; flow is nearly unidirectional)

Now, compare ′𝑥 ′ 𝑎𝑛𝑑 ′𝑟′ momentum terms (dominant terms) of the NS eqautions:

𝜕p 𝜕p
𝜕𝑟
≪ 𝜕𝑥
(𝑣𝑟 is a small quantity, 𝑣𝑟 ≪ 𝑣𝑥 )

With 𝑣𝑟 ≪ 𝑣𝑥 and 𝑅 ≪ 𝐿𝑥 (for thin films or channels) one can show that

𝜕2 𝑣𝑥 1 𝜕 𝜕𝑣 𝜕𝑣𝑥 𝜕𝑣
𝜕𝑥 2
≪ 𝑟 𝜕𝑟
(𝑟 𝜕𝑟𝑥 ) and 𝑣𝑥 𝜕𝑥
~ 𝑣𝑟 𝜕𝑟𝑥 (two inertial terms are of the same order of magnitude)

50
 𝑣 𝑅
, which means that inertial term can be neglected if 𝑅𝑒 (𝑅⁄𝐿 ) ≪ 1 where, 𝑅𝑒 = 𝑥𝜈 (This is not
𝑥
surprising considering the prevailing geometry and flow situations in hollow fibers of a membrane. This
condition is also similar to that for creeping flow: Re <<1, with the characteristic length chosen as “R” or
the radius of the sphere) Let us revert to the governing equation:

𝜇 𝜕 𝜕𝑣 dp
𝑟 𝜕𝑟
(𝑟 𝜕𝑟𝑥 ) = dx (1)
dp dp
Integrate twice 𝑣𝑥 (𝑥, 𝑟) in r direction with dx = dx (𝑥)

with BCs. 𝑣𝑥 (𝑥, 𝑅) = 0 (𝑛𝑜 − 𝑠𝑙𝑖𝑝), and

𝑣𝑥 (𝑥, 0) 𝑖𝑠 𝑓𝑖𝑛𝑖𝑡𝑒 (∇𝑣𝑥 (𝑥, 0) = 0): 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛
dp 8𝜇𝑣𝑥 (𝑥) 𝑟2
d𝑥
=− 𝑅2
and 𝑣𝑥 (𝑥, 𝑟) = 2𝑣𝑥 (𝑥) (1 − 𝑅2 ) (2a)
dp 𝑅2 𝑟2
Or simply, 𝑣𝑥 (𝑥, 𝑟) = − (dx ) 4𝜇 (1 − 𝑅2 ) (2𝑏) 𝑙𝑜𝑐𝑎𝑙 𝑚𝑒𝑎𝑛 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦(𝑟 − 𝑎𝑣𝑒𝑟𝑎𝑔𝑒𝑑)

1 𝜕 𝜕𝑣𝑥 𝜕𝑣𝑥 (𝑥) 𝑟2

Continuity: (𝑟𝑣𝑟 ) =− = −2 (1 − 𝑅2 ) (3)
𝑟 𝜕𝑟 𝜕𝑥 𝜕𝑥

Integrate eq. (3) along r-direction with condition @𝑟 = 0 ∇𝑣𝑟 = 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)

𝜕𝑣𝑥 (𝑥) 𝑟 1 𝑟 3
𝑣𝑟 (𝑟, 𝑥) = −𝑅 𝜕𝑥
[(𝑅) − 2 (𝑅) ] (4)

However, at 𝑟 = 𝑅, 𝑣𝑟 (𝑅, 𝑥) = 𝑣𝜔 (𝑘𝑛𝑜𝑤𝑛)

𝜕𝑣𝑥 (𝑥) 2𝑣𝜔
⇒ =− (5)
𝜕𝑥 𝑅

Also, @𝑥 = 0 𝑣 = 𝑣𝑜 (incoming velocity to tube). Integrate eq. (5) in x-direction to obtain

′ ′
𝑥 𝑐𝑎𝑛𝑛𝑜𝑡 𝑏𝑒 𝑙𝑜𝑛𝑔! 𝐼𝑡 𝑖𝑠 𝑎 𝑠ℎ𝑜𝑟𝑡 𝑡𝑢𝑏𝑒
2𝑣𝜔 𝑥
𝑣𝑥 (𝑥) = 𝑣𝑜 − 𝑅
( 𝑣 (𝑥) 𝑚𝑢𝑠𝑡 𝑏𝑒 𝑎 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑞𝑢𝑎𝑛𝑡𝑖𝑡𝑦 ) (6)

On substitutions
2𝑣𝜔 𝑥 𝑟2
𝑣𝑥 (𝑥, 𝑟) = 2 (𝑣𝑜 − 𝑅
) (1 − 𝑅2
) : from eq. (2a)

𝑟 1 𝑟 3
𝑣𝑟 (𝑥, 𝑟) = 2𝑣𝜔 (( ) − ( ) ) : from eqs. (4) and (5)
𝑅 2 𝑅

dp 8𝜇 2𝑣𝜔 𝑥
dx
= − 𝑅2 (𝑣𝑜 − 𝑅
) : from eqs. (2a) and (6)
𝟖𝝁𝒗𝒐 𝑳 𝒗 𝑳
Integrate the last equation to derive (𝒑𝟏 − 𝒑𝟐 ) = 𝑹𝟐
(𝟏 − 𝒗𝝎𝑹 ).
𝒐

Note that if 𝑣𝜔 = 0, the results expectedly reduce to those for an impermeable tube, i.e. Hagen-Poisueille
equation for pressure-drop in a straight cylindrical tube (with non-porous walls). Finally, you should also
note that 𝑣𝑟 (𝑥, 𝑟) is approximately zero but not exactly zero, and the flow is approximately 1D or
unidirectional but not exactly 1D or unidirectional. I leave it to you to plot the velocity and pressure
profiles, and compare them with those for a straight tube.

51
Prototype examples 4-5: Heat/ Mass transfer in laminar flow in a circular tube
- The Graetz problem and The Graetz-Nusselt problem

Physical situation:
𝑟

𝑇𝑜 𝑧

𝑟2
𝑣𝑧 = 𝑣𝑧 (𝑚𝑎𝑥) (1 − 𝑅2 ) ; fully developed flow, 𝑅𝑒 < 2100 𝑇 = 𝑇𝑜 (𝑢𝑛𝑖𝑓𝑜𝑟𝑚 𝑎𝑡 𝑡ℎ𝑒 𝑖𝑛𝑙𝑒𝑡)

a) Graetz-Nusselt prescribes heat flux at the tube-wall when 𝑧 > 0

∇𝑇𝑟=𝑅 is known
b) Graetz problem prescribes 𝑇𝑖 at wall when 𝑧 > 0

𝜇, 𝜌 =𝑐𝑜𝑛𝑠𝑡 (𝑎𝑣𝑔) 𝐶𝑝 , 𝑘 ≡ 𝑎𝑣𝑔. ⇒ 𝜇, 𝐶𝑝 , 𝑘

Restrictions: (1)
(2 ) Coupling between momentum and heat/mass transport is one-way: velocity profile is
superimposed on temp/concentration profiles & not vice-versa.
(3) Fully-developed laminar flow
(4) Negligible viscous dissipation (Brinkman number is small)
(5) Radial and axial symmetry: 𝑣𝑟 , 𝑣𝜃 = 0
(6) SS

Energy equation:

𝜕𝑇 𝑘 1 𝜕 𝜕𝑇 𝜕2 𝑇
𝑣𝑧 𝜕𝑧 = 𝜌 𝐶 ( 𝑟 𝜕𝑟 (𝑟 𝜕𝑟 ) + 𝜕𝑧 2
)
𝑝

𝑟2
Momentum/Continuity:⇒ 𝑣𝑧 (𝑟) = 𝑣𝑧 (𝑚𝑎𝑥) (1 − 𝑅2 )

(superimposed on temperature profiles or energy balance)

Prototype example 4: The Graetz-Nusselt problem (𝒒𝒘 𝒐𝒓 𝒘𝒂𝒍𝒍 𝒇𝒍𝒖𝒙 is prescribed)

BCs: 1. 𝑧 = 0, 𝑇 = 𝑇𝑜
𝜕𝑇
2. z = L, 𝜕𝑧
= 0; 𝑙𝑜𝑛𝑔 𝑡𝑢𝑏𝑒 𝑎𝑝𝑝𝑟𝑜𝑥? (yet to be determined)
𝜕𝑇
3. 𝑟 = 0, 𝑇 = 𝑓𝑖𝑛𝑖𝑡𝑒 → 𝜕𝑟
=0
𝜕𝑇
4. 𝑟 = 𝑅, −𝑘 𝜕𝑟 = 𝑞𝑤 (ℎ𝑒𝑎𝑡 𝑠𝑢𝑝𝑝𝑙𝑖𝑒𝑑 𝑜𝑟 𝑟𝑒𝑚𝑜𝑣𝑒𝑑: 𝑡ℎ𝑒 𝑛𝑢𝑚𝑒𝑟𝑖𝑐𝑎𝑙 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑞𝑤 𝑤𝑖𝑙𝑙 𝑏𝑒 ±)

Non-dimensionalize:

52
 𝑇−𝑇𝑜 𝑇−𝑇𝑜
𝜃 = (Δ𝑇) = 𝑞𝑤 𝑅⁄ (Why not θ =T/To? Because To is arbitrary; does not characterize heat-
𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 ( 𝑘)
𝜕𝑇 Δ𝑇
transport. On the other hand, 𝑞𝑤 = −𝑘 𝜕𝑟 ≈ −𝑘 Δ𝑟 , considering “R” is relatively smaller.)

𝑟 𝑧
𝜉=𝑅 , 𝜓=𝑧
𝑐

𝑧
convective length (𝑊ℎ𝑦 𝑛𝑜𝑡 𝜓 = 𝐿 ? 𝑓𝑙𝑢𝑥. Because "L" is arbitrary, fluid-temperature keeps on
increasing along the length no matter how small is the heat-flux at the wall)
𝑧 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑒𝑓𝑓𝑒𝑐𝑡
= 𝑅𝑃𝑒 ; 𝑧𝑐 = 𝑅 𝑃𝑒; 𝑃𝑒 = 𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑣𝑒 𝑒𝑓𝑓𝑒𝑐𝑡 = 𝑅𝑒 𝑃𝑟 (R is the conductive length) (Pr ~1 for gas and ~1000
for liquid)

𝑣𝑧 ⁄𝑣𝑧,𝑚𝑎𝑥 = (1 − 𝜉 2 )

𝜕𝜃 1 ∂ 𝜕𝜃 1 𝜕2 𝜃
(1 − 𝜉 2 ) = 𝜉 𝜕𝜉 𝜉 𝜕𝜉 + 𝑃2 𝜕𝜓2
𝜕𝜓 𝑒

If 𝑅𝑒 is large ≫ 1, the 3rd term is negligible compared (< 2100) to 1st and 2nd terms, i.e., conduction in z
direction is negligible in comparison to convective heat transport.

1 𝜕2 𝜃
That is, ≪ 1 ; On the other hand, radial conductive term is important & retained.
𝑃𝑒 2 𝜕𝜓2

𝜕𝜃 1 ∂ 𝜕𝜃
Now, (1 − 𝜉 2 ) = 𝜉 𝜕𝜉 𝜉 𝜕𝜉
𝜕𝜓

𝐵𝐶𝑠. 1 𝜓 = 0 𝜃=0
𝜕𝜃
2. 𝜉 = 0 𝜕𝜉
= 0
𝜕𝜃
3. 𝜉 = 1 − 𝜕𝜉
= 1
}

Note: Parabolic on 𝜓 (1st order)

-Similarity transform is no good- it has a bounded domain.

- Separation of variable should work (homogeneous PDE), but BC-3 is non-homogeneous ⇒ We should
first remove non-homogeneity in the boundary condition. Compare the situation with that of prototype
example 2 where we had to remove non-homogeneity in the equation in order to apply “Separation of
Variables” method.

𝑷𝒓𝒐𝒑𝒐𝒔𝒆: 𝜽 = 𝜽∞ (𝝍, 𝝃) − 𝜽𝒕 (𝝍, 𝝃)

Note that in the present case of constant heat flux at the wall, temperature will keep on increasing in the
tube no matter how small is the flux value. However, a distinct temperature profile can be expected far
(long distance) from the tube entrance. Also, similar to the previous prototype example, 𝜙(𝜏, 𝜉) has been
defined such that 𝜙𝑡 gradually vanishes with increasing 𝜓 or length, and

53
 Limiting form: 𝜃∞ (𝜓, 𝜉) = 𝐴(𝜓) + 𝐵(𝜉): (See BSL 10.8)

Note that ∇𝑇𝑟 ~ 𝑞𝑤 remains constant at the wall throughout the tube length. Actual temperature also
keeps on changing throughout the length. However, change in 𝜃∞ (𝜓, 𝜉) (temperature at long distances
in the tube) is similar. That is, a temperature profile is established in r-direction (shorter direction in the
𝜕𝜃∞
tube). Thus, although T changes along the length, 𝜕𝜓
remains constant (invariant along the length) in
the tube.

θ 𝜃∞
Plugging into equation:

𝑑𝐴 1 1 d 𝑑𝐵
𝑑𝜓
= (1−𝜉2 ) 𝜉 𝑑𝜉 𝜉 𝑑𝜉 = 𝐶𝑜

𝑓 𝑛 𝑜𝑓 𝜓 𝑜𝑛𝑙𝑦 𝑓 𝑛 𝑜𝑓 𝜉 𝑜𝑛𝑙𝑦

1 ∂ 𝑑𝐵 𝜉2 𝜉4
𝐴(𝜓) = 𝐶𝑜 𝜓 + 𝐶𝑜′ & 𝜉
𝜉 𝑑𝜉 𝑑𝜉
= 𝐶𝑜 (1 − 𝜉 2 )  𝐵(𝜉) = 𝐶𝑜 4
− 𝐶𝑜 16 + 𝐶1 𝑙𝑛𝜉 + 𝐶2 ′

It yields,

𝜉2 𝜉4
𝜃∞ (𝜓, 𝜉) = 𝐶𝑜 𝜓 + 𝐶𝑜 ( 4 − 16) + 𝐶1 𝑙𝑛𝜉 + 𝐶2

BCs. 𝜉 = 0, 𝜃∞ = 𝑓𝑖𝑛𝑖𝑡𝑒 ⇒ 𝐶1 = 0

∂𝜃∞
𝜉 = 1, − = 1 ⇒ 𝐶𝑜 = − 4
𝜕𝜉

(cannot use BC 𝜓 = 0 ⇒ 𝜃 = 0 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝜃∞ 𝑖𝑠 𝑣𝑎𝑙𝑖𝑑 𝑜𝑛𝑙𝑦 𝑓𝑜𝑟 𝑙𝑎𝑟𝑔𝑒 𝜓). C2 is yet to be determined.

Make use of integral heat balance over the tube length whose end is far from the inlet
𝑞𝑤
𝑧 ∞
𝑒𝑛𝑡ℎ𝑎𝑙𝑝𝑦
𝑏𝑎𝑙𝑎𝑛𝑐𝑒 ∶ 𝑧
𝜃 𝜃∞
(𝑖𝑛) (𝑜𝑢𝑡)

𝑞𝑤

𝑅
−2𝜋𝑅𝑧𝑞𝑤 = 2𝜋 ∫0 𝜌𝐶𝑝 (𝑇 − 𝑇𝑜 )𝑣𝑧 𝑟𝑑𝑟
𝑙𝑎𝑟𝑔𝑒
1
−𝜓 = ∫0 𝜃∞ (𝜓, 𝜉)(1 − 𝜉 2 )𝜉𝑑𝜉 → 𝑛𝑜𝑛 − 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑒
(what happened to 𝜌𝐶𝑝 ? It is there in 𝜓 = 𝑍⁄𝑍 𝑜𝑟 𝑍⁄𝑅𝑃𝑒)
𝑐

54
 1 𝜉2 𝜉4
= ∫0 (−4𝜓 − 4 ( 4 − 16) + 𝐶2 ) (1 − 𝜉 2 )𝜉𝑑𝜉
7
= −𝜓 − 96 + 𝐶2 1⁄4 ⇒ 𝐶2 = 7⁄24

𝜉4 7
𝜃∞ (𝜓, 𝜉) = −4𝜓 − (𝜉 2 − 4
) + 24

On 𝜓: linear 𝜉: parabolic temp profile

As a homework, determine the bulk fluid temperature or mixed-cup temperature: 𝜃𝑏 = −4𝜓 far from
the tube-length (see Dean)

Also, show that in the present case of constant wall heat flux for a laminar, Newtonian flow, far from the
tube entrance, the Nusselt number (based on the tube diameter), which is the dimensionless wall
𝝏𝜽
temperature gradient (𝝏𝝃 /(𝜽 − 𝜽𝒃 ) 𝐞𝐯𝐚𝐥𝐮𝐚𝐭𝐞𝐝 @ 𝝃 = 𝟏) = 48/11 or 4.364

Revert:

𝜉4 7
𝜃(𝜓, 𝜉) = −4𝜓 − (𝜉 2 − 4
) + 24 − 𝜃𝑡 (𝜓, 𝜉)

𝜕𝜃𝑡 1 ∂ 𝜕𝜃
𝜃𝑡 must satisfy (1 − 𝜉 2 ) = (𝜉 𝑡 ) (See BSL 12.2.1)
𝜕𝜓 𝜉 𝜕𝜉 𝜕𝜉

𝜕𝜃𝑡
𝐵𝐶𝑠: 𝜉 = 0 𝜕𝜉
=0
𝜕𝜃𝑡 𝜕𝜃 𝜕𝜃∞ 𝜕𝜃
𝜉=1 𝜕𝜉
=0 ( 𝜕𝜉𝑡 = 𝜕𝜉
− 𝜕𝜉 = 0)
𝜉4 7
𝜓 = 0 𝜃𝑡 (0, 𝜉) = 𝜃∞ (0, 𝜉) = − (𝜉 2 − 4 ) + 24
(𝜃𝑡 = 𝜃∞ − 𝜃: 𝑎𝑝𝑝𝑙𝑦 𝐵𝐶𝑠 𝑜𝑛 𝜃 )
}

Now, apply Separation of variables (equation and BCs are all homogenous for 𝜃𝑡 )

𝜃𝑡 (𝜓, 𝜉) = 𝑇(𝜉)𝑍(𝜓)
𝜕𝑍(𝜓) 2𝜓
𝜕𝜓
= −𝑐 2 𝑍 ⇒ 𝑍 = 𝐴𝑒 −𝑐 (𝑐 2 ≡ 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑛𝑠𝑡. )

𝐵𝐶𝑠
𝜕𝑇
And
1 d 𝑑𝑇
𝜉 + 𝑐 2 (1 − 𝜉 2 )𝑇 = 0 𝜉=0 𝜕𝜉
=0
𝜉 𝑑𝜉 𝜕𝜉
𝜕𝑇
=1 =0
𝜕𝜉 }

This equation is a special form of Sturm-Liouville equation (and not Bessel equation):
𝑑
(𝑟(𝑥)𝑦 ′ ) + [𝑞(𝑥) + 𝜆𝑃(𝑥)]𝑦 = 0 ∶ 𝑦(𝑥)
𝑑𝑥

BCs. 𝑎1 𝑦(𝑎) − 𝑎2 𝑦 ′ (𝑎) = 0

55
 𝑏1 𝑦(𝑏) − 𝑏2 𝑦 ′ (𝑏) = 0

Solution assumes the form of 𝑦𝑛 (𝑥) on orthogonal function with weight function 𝑝(𝑥) such that
𝑏
∫𝑎 𝑝(𝑥)𝑦𝑚 (𝑥)𝑦𝑛 (𝑥) = 0 𝑚≠𝑛
= 1⁄(𝑏 − 𝑎) 𝑚 = 𝑛

There exists an infinite set of eigenfunction 𝑇(𝜉) and eigenvalues 𝑐𝑖 , we can obtain by using the method
of Frobenius, and deriving a series solution about 𝜉 = 0. A recursive relationship allows us to get constants
in the series in terms of two unknown constants. We will do later. For now,
2𝜓
𝜙(𝜓, 𝜉) = 𝜃∞ (𝜓, 𝜉) − ∑∞
𝑖=1 𝐵𝑖 𝑒
−𝑐𝑖
𝑇(𝜉)
1
∫0 𝜃∞ (𝜉,0)𝑇(𝜉)(1−𝜉 2 )𝜉𝑑𝜉 𝑇 ≡ 𝐺𝑟𝑎𝑒𝑡𝑧 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛. 𝑆𝑒𝑒 𝑆𝑖𝑒𝑔𝑒𝑙, 𝑆𝑝𝑎𝑟𝑟𝑜𝑤, 𝐻𝑎𝑙𝑙𝑚𝑎𝑛
where, 𝐵𝑖 = 1 ;
∫0 [𝑇(𝜉)]2 (1−𝜉 2 )𝜉𝑑𝜉 𝐴𝑝𝑝 𝑆𝑐. 𝑅𝑒𝑠 𝐴7(1958)386 − 392

Asymptotic solution for small distances (BSL 12.2.2)

Explore temperature profiles near the walls; neglect curvature effects.

𝑆
𝑅
𝑍 ≡

𝑆 =𝑅−𝑟

𝐴ssume fluid extends from 𝑆 = 0 to 𝑆 → ∞ (semi- infinite)

Also, velocity gradient near the wall is linear: 𝑣𝑧 = 𝑣𝑜 𝑆⁄𝑅

𝑣𝑜 (𝑜𝑟 𝑣𝑚𝑎𝑥 )
Energy equation becomes 𝑅

𝑆 𝜕𝑇 𝜕2 𝑇 𝑘 𝜕2 𝑇
𝑣𝑜 (𝑅) 𝜕𝑍 = 𝛼 𝜕𝑆 2 = (𝜌𝐶 ) 𝜕𝑆 2 (Cartesian coordinate)
𝑝

𝑍 = 0 𝑇 = 𝑇𝑜 𝑎𝑡 𝑡ℎ𝑒 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑟ℎ𝑜𝑜𝑑
𝜕𝑇 }
𝑆=0 − 𝑘 𝜕𝑆 = 𝑞𝑤 ; 𝑆 → ∞ 𝑇 = 𝑇𝑜 𝑜𝑓 𝑡ℎ𝑒 𝑤𝑎𝑙𝑙𝑠

𝑣𝑜 𝜕𝑇 1 𝜕2 𝑇
or =𝛼
𝑅 𝜕𝑍 𝑆 𝜕𝑆 2

𝑣𝑜 𝜕2 𝑇 𝜕 1 𝜕2 𝑇
or 𝑅 𝜕𝑆𝜕𝑍
= 𝛼 𝜕𝑆 𝑆 𝜕𝑆 2

56
 𝑣𝑜 𝜕𝑞𝑠 𝜕 1 𝜕𝑞𝑠 𝜕𝑇
𝑅 𝜕𝑧
= 𝛼 𝜕𝑆 𝑆 𝜕𝑆
(define qs =−𝑘 𝜕𝑆
)

𝑞
Non-dimensionalize: 𝜓 = 𝑞𝑠 ; 𝜂 = 𝑆⁄𝑅 substitute to get
𝑤

2
𝜆 = 𝑍⁄ 2 = 𝑍⁄𝑍 (Note the characteristic length 𝑍𝑐 = 𝑣𝑜 𝑅 ⁄𝛼 )
(𝑣𝑜 𝑅 ⁄𝛼 ) 𝑐

𝜕𝜓 𝜕 1 𝜕𝜓
Therefore, 𝜕𝜆
= 𝜕𝜂 (𝜂 𝜕𝜂 )

𝜆 = 0, 𝜓 = 0 𝑠𝑒𝑚𝑖 − 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑑𝑜𝑚𝑎𝑖𝑛: 𝑇ℎ𝑒 𝐵𝐶 𝑜𝑛 𝜆

BCs }
𝜂 = 0, 𝜓 = 1; 𝜂 → ∞ 𝜓 = 0 𝑐𝑜𝑙𝑙𝑎𝑝𝑠𝑒𝑠 𝑤𝑖𝑡ℎ 𝑜𝑛𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑡𝑤𝑜 𝐵𝐶𝑠 𝑜𝑛 𝜂, 𝑖. 𝑒. 𝜂 → ∞

Similarity transform or Method of combination of variables:

𝜂
Use or derive 𝑍 = 𝑎 𝜂 𝜆𝑏 =
∛9𝜆

And 𝑍𝜓 ′′ + (3𝑍 3 − 1)𝜓 ′ = 0

𝑍=0 𝜓=1

𝑍→∞ 𝜓=0
∞ 3
𝑞𝑠 ∫ 𝑍𝑒 −𝑍 𝑑𝑍
Solution is 𝜓 = 𝑞 = 𝑍 ⁄ ∞ −𝑍3
𝑤 ∫0 𝑍𝑒 𝑑𝑍

3 ∞ 3
= Γ(2 ∫ 𝑍𝑒 −𝑍 𝑑𝑍
⁄3) 𝑍

𝑇 1 ∞
But ∫𝑇 𝑜 𝑑𝑇 = − ∫𝑆 𝑞𝑠 𝑑𝑆 (at the neighborhood of the walls)
𝑘

Therefore,

𝑇−𝑇𝑜 ∞ 𝑞
𝜃=𝑞 𝑅 = ∛9𝜆 ∫𝑍 𝜓𝑑𝑍 (𝜓 = 𝑞𝑠 )
𝑤 ⁄𝐾 𝑤

3
𝑒 −𝑍 Γ(2⁄3,𝑍3 )
𝜃(𝜂, 𝜆) = ∛9𝜆 [Γ(2 − 𝑍 {1 − Γ(2⁄3)
}] Incomplete function
⁄3)

Complete function

(There are tables to evaluate such mathematical functions including Graetz and gamma functions, for
example in the book by Kreyszig)

57
 Lecture #12
Prototype example 5: Graetz Problem (wall temperature is prescribed)

𝑇𝑤
𝑟 The restrictions are the
𝑇𝑜 𝑧 same as in example 4.

𝑇𝑤
𝑇−𝑇𝑤 𝑟 𝑧
𝜃= , 𝜉= , 𝜓= (characteristics variables are the same as in the previous case)
𝑇𝑜 −𝑇𝑤 𝑅 𝑅𝑃𝑒

The governing equation is also the same as before

𝜕𝜃 1 ∂ 𝜕𝜃
(1 − 𝜉 2 ) = 𝜉 𝜕𝜉 𝜉 𝜕𝜉 (note that the axial conduction term is again neglected; Pe is large)
𝜕𝜓

BCs: 𝜃(0, 𝜉) = 1;

𝜕𝜃
𝜃(𝜓, 0) = 𝑓𝑖𝑛𝑖𝑡𝑒 𝑜𝑟 𝜕𝜉
= 0;

𝜃(𝜓, 1) = 0 (compare to the previous case when the BC was non-homogenous)

In this case, ehe equation is homogenous and the BCs on 𝜓 are also homogenous. Separation of
variables will now work without any mathematical manipulation.

Separation of variables:

𝜃 = 𝑅(𝜉)𝐹(𝜓)
1 𝑑𝐹 1 1 d 𝑑𝑅
= 𝜉 = −𝐵2
𝐹 𝑑𝜓 𝑅(1−𝜉 2 ) 𝜉 𝑑𝜉 𝑑𝜉

2𝜓
⇒ 𝐹 = 𝐴𝑒 −𝐵 ;

d 𝑑𝑅
R: 𝑑𝜉
(𝜉 𝜕𝜉 ) + 𝐵2 (1 − 𝜉 2 )𝜉𝑅 = 0

𝑑𝑅
𝜉=0⇒ 𝑑𝜉
= 0 ; 𝜉 = 1 ⇒ 𝑅 = 0 (use BCs on θ to get BCs on R)

Check to see if Sturm-Liouville equation applies

𝑑
i. e. [𝑟(𝑥)𝑦′] + [𝑞(𝑥) + 𝜆𝑃(𝑥)]𝑦 = 0
𝑑𝑥
𝐵𝐶𝑠 𝐴1 𝑦(𝑎) − 𝐴2 𝑦 ′ (𝑎) = 0 }
𝐵1 𝑦(𝑏) − 𝐵2 𝑦 ′ (𝑏) = 0

58
 Comparing the equations, 𝑟(𝜉) = 𝜉 𝑞(𝜉) = 0 , 𝑃(𝜉) = 𝜉(1 − 𝜉 2 ) 𝑤𝑒𝑖𝑔ℎ𝑖𝑛𝑔 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛,
𝜆 = 𝐵2 ,

Comparing the BCs 𝑎 = 0 𝐴1 = 0 𝐴2 ≠ 0

𝑏 = 1 𝐵1 = 1 𝐵2 = 0

Hence, Sturm-Liouville equation form applies and

1
∫0 𝑅𝑛 𝑅𝑚 𝜉 (1 − 𝜉 2 )𝑑𝜉 = 0 ; 𝑚 ≠ 𝑛 (But types of function still unknown)
=1 ; 𝑚=𝑛

(This time, we will extensively discuss Graetz function):

Let 𝑈 = 𝜉𝐵 to get

𝑑2 𝑅 1 𝑑𝑅 𝑈2
+ + [1 − ]𝑅 =0
𝑑𝑈 2 𝑈 𝑑𝑈 𝐵2

⇒ There is a singularity at 𝑈 = 0

Propose solution of the form

𝑅 = ∑∞
𝑛=0 𝐶𝑛 𝑈
𝑛
(𝐹𝑟𝑜𝑏𝑒𝑛𝑖𝑢𝑠 𝑠𝑒𝑟𝑖𝑒𝑠)

Plug this to get

𝐶𝑛−2
∑∞
𝑛=2 [𝐶𝑛+2 (𝑛 + 2)(𝑛 + 1) + 𝐶𝑛+2 (𝑛 + 2) + 𝐶𝑛 − ] 𝑈 𝑛 + 𝐶1 𝑈−1 + 𝐶𝑜 2 + 𝐶3 3𝑈1 + 𝐶2 2 +
𝐵2
𝐶3 3.2. 𝑈 + 𝐶𝑜 + 𝐶1 𝑈1 = 0

Gather terms, 𝑅 = 𝐶1 ∑∞
𝑛=0 𝐵2𝑛 𝑈
2𝑛
+ 𝐶2 ∑∞
𝑛=0 𝐵2𝑛+1 𝑈
2𝑛+1

1 1
where, 𝐵2𝑛 = (2𝑛)2 [𝐵2 𝐵2𝑛−4 − 𝐵2𝑛−2 ]

𝐵2𝑛+1 = 0 identically

Therefore, to obtain complete solution identically, we use method of reduction of order where

exp(−∫ 𝑃1 (𝑥)𝑑𝑥)
𝑦2 (𝑥) = 𝑦1 (𝑥)∫ [𝑦1 (𝑥)]2
𝑑𝑥 where, 𝑦1 (𝑥), 𝑎𝑛𝑑 𝑦2 (𝑥) are solution of

1
𝑦" + 𝑃1 (𝑥)𝑦′ + 𝑃2 (𝑥)𝑦 = 0; 𝑃1 (𝑥) = − 𝑥 in the present case.

And, 𝑅(𝑈) = 𝐶1 𝑅1 (𝑈) + 𝐶2 𝑅2 (𝑈)

= 𝐶1 ∑∞
𝑛=0 𝐵2𝑛 𝑈
2𝑛
+ 𝐶2 ∑∞ 𝑛
𝑛=0{𝑅1 (𝑈)𝑙𝑛𝑈 − ∑ 𝑑𝑛𝑈 }

𝑅 should be finite as 𝑈 → ∞ ⇒ 𝐶2 = 0

59
Hence, 𝑅(𝑈) = 𝐶1 ∑∞
𝑛=0 𝐵2𝑛 𝑈
2𝑛

1 1 1
𝐵𝑜 = 1, 𝐵2 = − 22 , 𝐵2𝑛 = 2𝑛 [𝐵2 𝐵2𝑛−4 − 𝐵2𝑛−2 ]

2
𝜃 = ∑∞
𝑛=0 𝐴𝑛 𝑒
−𝐵𝑛 𝜓
𝑅(𝐵𝑛 𝜉) R is the Graetz function

To get 𝐴𝑛 , we use orthogonality conditions at 𝜓 = 0

1
∫0 𝜃(=1)𝜉(1−𝜉 2 )𝑅(𝐵𝑛 𝜉)𝑑𝜉 𝐼𝑈
𝐴𝑛 = 1 2 = 𝐼𝐿
∫0 (𝑅(𝐵𝑛 𝜉)) 𝜉(1−𝜉 2 )𝑑𝜉

𝐼𝑈 is evaluated using

1 𝑑 𝑑𝑅(𝐵 𝜉) 1
∫0𝑑𝜉
(𝜉 𝑑𝜉𝑛 ) 𝑑𝜉 = −𝐵𝑛2 ∫0 (1 − 𝜉 2 )𝜉𝑅(𝐵𝑛 𝜉) 𝑑𝜉 𝑓𝑟𝑜𝑚 𝑂𝐷𝐸

Hence,

𝑑𝑅 1
𝜉 𝑑𝜉 (𝐵𝑛 𝜉)| − 0 = 𝐵𝑛2 ∫0 (1 − 𝜉 2 )𝜉𝑅(𝐵𝑛 𝜉) 𝑑𝜉
𝜉=1

1 𝑑𝑅(𝐵𝑛 𝜉)
𝐼𝑢 = − ( )|
𝐵𝑛2 𝑑𝜉 𝜉=1

1 𝜕𝑅 𝜕𝑅(𝐵𝑛 𝜉)
Also, 𝐼𝐿 = ( (𝐵𝑛 𝜉) )|
2𝐵𝑛 𝜕𝐵 𝜕𝜉 𝜉=1

2
𝐴𝑛 = − [ 𝜕𝑅(𝐵𝑛 𝜉) ]
𝐵𝑛 |
𝜕𝐵𝑛 𝜉=1

2 2
Therefore, 𝜃(𝜓, 𝜉) = ∑∞
𝑛=0 − 𝜕𝑅(𝐵𝑛 𝜉) | 𝑅(𝐵𝑛 𝜉)𝑒 −𝐵𝑛 𝜓 (Check 𝜓 → ∞, Ɵ → 0 𝑜𝑟 𝑇 → 𝑇𝑤)
𝐵𝑛
𝜕𝐵 𝜉=1

where, 𝑅(𝑈) = ∑∞
𝑛=0 𝐵2𝑛 𝑈
2𝑛

1 1 1
𝐵𝑜 = 1, 𝐵2 = − 22 , 𝐵2𝑛 = (2𝑛)2 [𝐵2 𝐵2𝑛−4 − 𝐵2𝑛−2 ]

𝐵𝑛 are roots of 𝑅(𝐵) = 0 (𝑆𝑒𝑒 𝐽𝑎𝑐𝑜𝑏)

𝐵𝑜 = 2.705, 𝐵1 = 6.66, 𝐵2 = 10.3

𝜕𝑅 𝜕𝑅 𝜕𝑅
𝜕𝐵𝑜
= 0.50082, 𝜕𝐵1
== 0.371, 𝜕𝐵2
= −0.505

See ref. Graetz, L 1885, 25, #7, P22 Ann Dev Physik and Chem

Jacob, M 1953 V1, p 451 Heat Transfer

60
Gtraetz function is not easy to compute. Therefore, subsequent to the solutions provided by Graetz and
Graetz-Nusselt, many attempts were made to fully solve the two cases in heat transfer: constant wall flux
and constant wall temperature, especially for the extreme or far away from the entrance to the tube. The
books by Kays, Crawford, and Weigand (2005, 4th Edition) and Kaka, Shah, and Aung (1987) have
addressed/solved these cases slightly differently. The second case is indeed more computationally or
mathematically complex than the first case. In the second case, Nusselt number for the far end of the tube
is computed to be 3.657 (16% less than in the first case). Deen’s book has reported Nusselt numbers for
both cases for the entire tube length, using the study of Shah and London (1978). In both cases, Nusselt
numbers decrease along the tube length, and become constant at 4.364 and 3.657, respectively.
Kays, Crawford, and Weigand have used a general temperature profile for both cases (far from the tube
length) which is outlined here:
𝜕 𝑇 − 𝑇𝑤
( ) = 0, 𝑤ℎ𝑖𝑐ℎ 𝑜𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑡𝑖𝑜𝑛 𝑔𝑖𝑣𝑒𝑠:
𝜕𝑥 𝑇𝑏 − 𝑇𝑤

𝜕𝑇 𝑑𝑇𝑤 𝑇 − 𝑇𝑤 𝑑𝑇𝑤 𝑇 − 𝑇𝑤 𝑑𝑇𝑏

= −( ) +( )
𝜕𝑥 𝑑𝑥 𝑇𝑏 − 𝑇𝑤 𝑑𝑥 𝑇𝑏 − 𝑇𝑤 𝑑𝑥

Constant wall flux:

𝜕𝑇 𝑑𝑇𝑤 𝑑𝑇𝑏
= =
𝜕𝑥 𝑑𝑥 𝑑𝑥

𝜕𝑇 𝑇−𝑇𝑤 𝑑𝑇𝑏
Constant wall temperature: =( )
𝜕𝑥 𝑇𝑏 −𝑇𝑤 𝑑𝑥
A simple computation using the above different values for the temperature gradient gives Nu = 4.364 in
the first case. Some additional computation (again involving Graetz function!) gives Nu = 3.657. The
temperature variations with tube lengths for these two cases are as follows:

T Constant wall flux T Constant wall temperature

Tw Tw

Tb Tb

X X
A schematic representation of variation in T(z,r) in the constant wall temperature case is given below. Tw
remains constant along the length, but the wall flux or gradient varies. Far from the entrance, the entire
fluid is at Tw. You should compare the temperature profiles to the previous case in which case the gradient
at the tube walls remains constant along the length.
Tw Tw Tw Tw Tw

To

61
 Lecture #13
Prototype example 6: Brinkman problem (viscous heating)

𝑇𝑜
𝑟

𝑇𝑜 𝑧

𝑇𝑜
Solve 𝑇(𝑟, 𝑧) = ? ⇒ Fluid-temperature increases because of viscous heating. Assume SS, fully developed
laminar flow of a Newtonian, incompressible fluid. Note that the inlet and wall temperature are constant
at To.
𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑜𝑓 𝑣𝑖𝑠𝑐𝑜𝑢𝑠
Energy equation becomes 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛
𝜕𝑇 1 𝜕 𝜕𝑇 𝜕2 𝑇 𝜕𝑣 2
cal/s-cm3: 𝜌𝐶𝑣 𝑣𝑧 𝜕𝑧 = 𝑘 [𝑟 𝜕𝑟 𝑟 𝜕𝑟 + 𝜕𝑍2 ] + 𝜇 ( 𝜕𝑟𝑧) (𝑣𝑟, 𝑣Ɵ = 0)

𝐷𝑇 𝜕𝑃
Recall: 𝜌𝐶𝑣 ⃗⃗ ) − 𝜏 : ∇𝑉
= −∇. 𝑞 − 𝑇 (𝜕𝑇 ) (∇. 𝑉 ⃗⃗
𝐷𝑡 𝑉

⃗⃗ + ∇𝑉
𝜏 : ∇𝑉 = 𝜇[ ∇𝑉 ⃗⃗ 𝑇 ]: ∇𝑉
⃗⃗ = 𝜙𝑉 (+𝑣𝑒)

If 𝑃𝑒 ≫ 1 axial conduction term can be neglected (similar to the previous case):

𝜕2 𝑇 𝜕𝑇 1 𝜕2 𝑇
k ≪ 𝜌𝐶𝑣 𝑣𝑧 𝑜𝑟 ≪1
𝜕𝑍 2 𝜕𝑧 𝑃𝑒2 𝜕𝜓2

𝜕𝑇 1 𝜕 𝜕𝑇 𝜕𝑣 2
𝜌𝐶𝑣 𝑣𝑧 𝜕𝑧 = 𝑘 [𝑟 𝜕𝑟 𝑟 𝜕𝑟 ] + 𝜇 ( 𝜕𝑟𝑧) (1)

BCs: T(0, r) = To for all r ≥ 0

T(z, 0) = finite or ∇T = 0, and

T(z, R) = To for all z ≥ 0

Non-dimensionalize,
𝑇−𝑇𝑜 𝑟 𝑧 𝑘𝑧
𝜃= 𝑇𝑜
, 𝜉 = 𝑅 ; 𝜓 = 𝑅𝑃𝑒 = 𝜌𝐶 2
𝑝 𝑣𝑚𝑎𝑥 𝑅

𝑧𝑐
Substituting,

𝜕𝜃 1 ∂ 𝜕𝜃
(1 − 𝜉 2 ) = 𝜉 𝜕𝜉 (𝜉 𝜕𝜉 ) + 16𝐵𝑟𝜉 2 (2)
𝜕𝜓

62
 2
𝜇𝑣𝑚𝑎𝑥 𝑟𝑎𝑡𝑒 𝑜𝑓 ℎ𝑒𝑎𝑡 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑑 𝑏𝑦 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛
where, 𝐵𝑟 = ≡( )
4𝑘𝑇𝑜 𝑟𝑎𝑡𝑒 𝑜𝑓 ℎ𝑒𝑎𝑡−𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑏𝑦 𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑜𝑛

in radial direction

BCs: Ɵ(0, ξ) = 0 for all ξ ≥ 0

Ɵ(ψ, 0) = finite or ∇Ɵ = 0, and

Ɵ(ψ, 1) = 0 for all ψ ≥ 0

⇒ The equation is non-homogeneous, linear 2nd order on 𝜉. Non-homogeneity should be removed before
applying the separation of variable method.

Try 𝜃 = 𝜃∞ (𝜉) − 𝜃𝑡 (𝜓, 𝜉) (3)

a temperature profile is established in r-direction far from the tube-entrance, which is

constant along tube length.

⇒ Viscous-heat generated in the fluid is removed from the wall; ‘gradient’ adjusts to keep the wall
temperature constant at 𝑇𝑜

𝜕𝜃𝑡 1 ∂ 𝜕𝜃𝑡 1 ∂ 𝜕𝜃
−(1 − 𝜉 2 ) =− 𝜉 + (𝜉 ∞ ) + 16𝐵𝑟𝜉 2
𝜕𝜓 𝜉 𝜕𝜉 𝜕𝜉 𝜉 𝜕𝜉 𝜕𝜉

On 𝜽∞ :

1 d 𝑑𝜃
(𝜉 ∞ ) + 16𝐵𝑟𝜉 2 = 0
𝜉 𝑑𝜉 𝑑𝜉

𝜃∞ = −𝐵𝑟𝜉 4 − 𝐶1 𝑙𝑛𝜉 + 𝐶2

BCs: 𝜃∞ (0) = 𝑓𝑖𝑛𝑖𝑡𝑒 ⇒ 𝐶1 = 0

𝜃∞ (1) = 0 ⇒ 𝐶2 = 1

𝜃∞ = 𝐵𝑟(1 − 𝜉 4 );

𝜕𝜃 1 ∂ 𝜕𝜃𝑡
And on 𝜽𝒕 : (1 − 𝜉 2 ) 𝜕𝜓𝑡 = 𝜉 𝜕𝜉 𝜉 𝜕𝜉

𝜕𝜃𝑡 1 ∂ 𝜕𝜃𝑡
(1 − 𝜉 2 ) = 𝜉 𝜕𝜉 𝜉
𝜕𝜓 𝜕𝜉

BCs: 𝜃𝑡 (0, 𝜉) = 𝜃∞ see (3) above

𝜃𝑡 (𝜓, 0) = 𝑓𝑖𝑛𝑖𝑡𝑒 𝑜𝑟 ∇Ɵ = 0

𝜃𝑡 (𝜓, 1) = 0

Separation of Variables, 𝜃𝑡 (𝜓, 𝜉) = 𝑍(𝜓)𝐹(𝜉)

63
 ∂ ∂F
On 𝐹(𝜉): 𝜕𝜉
[𝜉 𝜕𝜉 ] = 𝛼 2 𝜉(1 − 𝜉 2 )𝐹

F is solved in same way as that for Graetz function! Z(ψ) is also evaluated the same way as before to solve
2
𝜃(𝜓, 𝜉) ≡ 𝐵𝑟(1 − 𝜉 4 ) − ∑∞
𝑛=0 𝐴𝑛 𝑅(𝐵𝑛 , 𝜉)𝑒
−𝐵𝑛 𝜓

where, 𝑅(𝐵𝑛 , 𝜉) = 𝐺𝑟𝑎𝑒𝑡𝑧 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛

1
2𝐵𝑟 1 ∑∞ 2𝑛
𝑛=0 𝐵2𝑛 𝐵𝑛 (2𝑛+6)(2𝑛+8)
𝑎𝑛𝑑 𝐴𝑛 = − −
𝐵𝑛 ( ∂R )| 𝐵𝑛 (
∂R dR
) |
𝜕𝐵𝑛 𝜉=1 𝜕𝐵𝑛 𝑑𝜉 𝜉=1

Ɵ Ɵ𝑤
𝜕𝜃 𝑔𝑟𝑎𝑑 𝑎𝑡 𝑤𝑎𝑙𝑙 ≠ 0
=0
𝜕𝜉 {𝑒𝑣𝑒𝑛 𝑖𝑓 𝑇𝑓𝑙𝑢𝑖𝑑 𝑎𝑡|
𝑟=𝑅
= 𝑇𝑜 (𝑤𝑎𝑙𝑙 𝑡𝑒𝑚𝑝. )
𝜉

0 1

1) Large ′∇𝑉 ′ 𝑛𝑒𝑎𝑟 𝑡ℎ𝑒 𝑤𝑎𝑙𝑙 → 𝑀𝑎𝑥𝑚𝑖𝑚𝑢𝑚 ℎ𝑒𝑎𝑡 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 (see eqs. 1 and 2)
2) Temperature grad changes at the tube walls along the tube length but 𝜃𝑤 remains const at 0. At
a far distance, gradient decreases to zero.
3) Temperature gradient is always zero at the center of the tube, as in the boundary condition.

You should consider solving another case when the wall heat flux is zero (adiabatic). In this case,
think of a solution in the form 𝜃∞ (𝜓, 𝜉) = 𝐴(𝜓) + 𝐵(𝜉). That is, the gradient at the wall is always
zero. But, at a far distance, the radial temperature profile is established, and the temperature
𝜕𝜃∞
gradient remains invariant along z-direction. See the problem 11B.2(c) of BSL.
𝜕𝜓

So far we solved three prototype problems in heat transport: constant wall flux, constant wall
temperature, and viscous heating in fluid flow. Note that there can be analogous mass transport
problems! For example, consider the constant concentration Co of a reactant in a fluid mixture at
the inlet of a tube. A zeroth order reaction occurs at the wall, which means concentration gradient
is constant at the wall. This is analogous to our prototype example 4. Follow the same procedure
including an integral mass balance to solve for concentrations at short and long distances in the
tube. Similarly, the analogous to viscous heating term is a chemical homogenous reaction in mass
transport. Again, follow the same procedure as above. It is clear that, the non-dimensional forms
of the conservation equations and BCs in heat and mass transport case should be the same or
similar to obtain the identical or similar solutions.

64
Prototype example: 7 (Natural Convection: Coupling between hydrodynamic and energy)
+∞

ℎ𝑜𝑡 Long vertical plate (−∞ 𝑡𝑜 + ∞)

𝑇ℎ { 𝑒𝑥𝑝𝑙𝑜𝑟𝑒 𝑠𝑡𝑒𝑎𝑑𝑦 − 𝑠𝑡𝑎𝑡𝑒
𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 & 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑟𝑜𝑓𝑖𝑙𝑒𝑠
g
(ΔT) 𝑇𝑐 (𝐶𝑜𝑙𝑑)

−∞
−𝑏 +𝑏

Energy balance equation:

𝜕𝑇 𝜕𝑇 𝜕𝑇 𝜕𝑇 𝜕2 𝑇 𝜕2 𝑇 𝜕2 𝑇 𝜕𝑃
𝜌𝐶𝑣 [ + 𝑣𝑥 + 𝑣𝑦 + 𝑣𝑧 ] = 𝑘[ + + ] − 𝑇 ( ) ∇. 𝑉 + 𝜇𝜙𝑉
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 2 𝜕𝑦 2 𝜕𝑧 2 𝜕𝑇 𝑉

SS (𝑖𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒, 𝑙𝑜𝑤 𝑀𝑎𝑐ℎ 𝑛𝑜, 𝑙𝑜𝑤 𝐵𝑟 #)

Note: ⇒ Plates are infinitely long: −∞ 𝑡𝑜 + ∞

𝜕𝑇 𝜕2 𝑇
𝜕𝑧
= 𝜕𝑧2 = 0

𝜕𝑣𝑧
⇒ 𝑣𝑦 = 0 𝑤ℎ𝑦? 𝐴𝑝𝑝𝑙𝑦 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦: = 0 , 𝑣𝑧 ≠ 0
𝜕𝑧

⃗⃗ = 0 𝑦𝑖𝑒𝑙𝑑𝑠 𝑣𝑦 = 0
Thus, ∇. 𝑉

𝜕𝑣𝑦
( 𝜕𝑦 = 0) (Also 𝑦 momentum balance)

Check for ∇𝑝 in 𝑦 direction

Energy balance equation is now reduced to:

𝑑2 𝑇
𝑑𝑦 2
= 0 ⇒ 𝑇 = 𝑇𝑐 @𝑦 = +𝑏

𝑇 = 𝑇ℎ @𝑦 = −𝑏
1 𝑦 𝑇ℎ +𝑇𝑐
𝑇 = 𝑇𝑚𝑒𝑎𝑛 − 2 Δ𝑇 (𝑏 ) 𝑤ℎ𝑒𝑟𝑒, 𝑇𝑚𝑒𝑎𝑛 = 2
𝑎𝑛𝑑 Δ𝑇 = (𝑇ℎ − 𝑇𝑐 ): 𝐼𝑡 𝑖𝑠 𝑎 𝑙𝑖𝑛𝑒𝑎𝑟 𝑝𝑟𝑜𝑓𝑖𝑙𝑒.

(Note that in this example, we have solved energy equation ahead of NS equation and temperature
profiles superimposed on momentum balance equation)

Velocity profile:

Z-momentum balance equation is simplified as:

65
 𝑑2 𝑣𝑧 𝑑𝑝
𝜇 − 𝑑𝑧 + 𝜌𝑔 = 0
𝑑𝑦 2

𝜌 = 𝜌 − 𝜌𝛽(𝑇 − 𝑇) : Taylor’s series neglecting the higher order term. 𝑇 is unknown. Revisit Lecture 6.

𝑑𝑝 𝑑p
= 𝜌𝑔 (𝑖𝑓 𝑓𝑙𝑢𝑖𝑑 𝑤𝑒𝑟𝑒 𝑠𝑡𝑎𝑡𝑖𝑐). Or, = 0, 𝑤ℎ𝑒𝑟𝑒 p is the hydrodynamic pressure.
𝑑𝑧 𝑑𝑧

𝑑 2 𝑣𝑧
𝜇 𝑑𝑦 2
= − 𝜌 𝑔 𝛽 (𝑇 − 𝑇) (𝜌@𝑇, 𝜌 @ 𝑇, 𝑣𝑖𝑧. a warm fluid pocket at T rises relatively to the cold
bulk fluid at a temperature 𝑇. 𝐼𝑠 𝑖𝑡 𝑎 𝑚𝑜𝑑𝑒𝑙? )
driving force for momentum in 𝑦 direction

𝑑 2 𝑣𝑧 1 𝑦
𝜇 𝑑𝑦 2
= − 𝜌 𝑔 𝛽 [(𝑇𝑚𝑒𝑎𝑛 − 𝑇) − 2 Δ𝑇 (𝑏 ) ] : substitute from the previous solution for T.

𝑩𝑪𝒔. 𝑦 = ±𝑏, 𝑣𝑧 = 0 (𝑛𝑜 − 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

𝜌 𝛽𝑔𝑏2 Δ𝑇 𝑦 3 𝑦 2 𝑦 𝑏(𝑇𝑚 −𝑇)

𝑣𝑧 = [(𝑏 ) − 𝐴 ( ) − ( ) + 𝐴] where, 𝐴 =
12𝜇 𝑏 𝑏 Δ𝑇

Note 𝑇 remains to be determined

Assume, Volume flow up = Volume flow down (it means that fluid is crossing over from left to right
𝑎𝑡 + ∞ and right to left at -∞; There cannot be an infinite source of fluid generation)
+1 𝑦
∫−1 𝑣𝑧 𝑑 (𝑏 ) = 0 (Let net volumetric flow in the 𝑧-direction be 0). Note it is not a Continuity equation.

2
− 3 𝐴 + 2𝐴 = 0 ⇒ 𝐴 = 0

𝑇ℎ +𝑇𝑐
Therefore, 𝑇 = 𝑇𝑚 = 2

Thus, 𝑇 is specific to the problem (temperature profile) 𝑇 = 𝑇𝑚

𝜌 𝛽𝑔𝑏2 Δ𝑇 𝑦 3 𝑦
𝑣𝑧 = [(𝑏 ) − (𝑏 )] 𝑇ℎ
12𝜇

𝑏𝑣𝑧 𝜌 𝑦
2
𝜌 𝛽𝑔𝑏3 Δ𝑇
𝑇𝑐
If 𝜙 = , 𝜂 = (𝑏 ) , 𝐺𝑟 = 𝑣𝑧
𝜇 𝜇2

1
𝜙 = 12 𝐺𝑟[𝜂 3 − 𝜂]
−𝑏 +𝑏
(Deen and BSL-latest edition) use 𝑇 = 𝑇𝑚𝑒𝑎𝑛 with the assumption that error is minimum in Taylor’s series
expansion for density changes if the expansion is around 𝑇𝑚𝑒𝑎𝑛 or around the density at
𝑑𝑝 𝑑p
𝑇𝑚𝑒𝑎𝑛 , and show that 𝑑𝑧 = 𝜌𝑔 𝑜𝑟 𝑑𝑧 = 0 where, p is the hydrodynamic pressure. The final answers are
the same. These books have also addressed natural convection in an unconfined domain, but that requires
boundary layer theory. Time permitting it will be solved later using Similarity Transform technique.

66
 Lecture #14
Prototype example 8: Steady-state diffusion through a stagnant gas film (Film Theory)

- Constant P and T; no chemical reaction.

𝑧
𝑃𝑢𝑟𝑒"𝐵"𝑔𝑎𝑠 𝑠𝑤𝑒𝑒𝑝𝑠 𝐴 𝑥𝐴 = 𝑥𝐴2 ~0 at 𝑧2
𝑓𝑖𝑙𝑚 𝑥𝐴 , 𝑥𝐵 = ?
(𝐴 + 𝐵) B is insoluble in A
𝐸𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑢𝑟𝑒 𝐴 (𝑙𝑖𝑞𝑢𝑖𝑑) 𝑥𝐴 = 𝑥𝐴1 at 𝑧1 𝑓𝑖𝑥𝑒𝑑 𝑏𝑦 𝑃𝑜 (𝑇)
0 /𝑅𝑜𝑢𝑙𝑡′𝑠 𝑙𝑎𝑤

Note: 𝑥𝐴 + 𝑥𝐵 = 1 at all points in the film

A evaporates, diffuses upward because of the concentration gradient, and is carried or swept away by B.
The species B diffuses in the opposite direction towards the liquid A also because of the concentration
gradient. But B is insoluble in A. Therefore, there is no “sink” for B, and therefore, there should not be any
concentration gradient of B at the liquid-vapor interface (𝐽𝐵 = −𝐷𝐴𝐵 ∇𝐶𝐵 )? However, this scenario (finite
concentration gradient) is possible if there is a ‘convective’ or “bulk- transport” of B induced in the positive
z-direction so that the net flux of B at any horizontal plane including g - l interface is zero. Note that there
is no pressure-driven (forced) convection or free (natural) convection in the film.

⇒ Recall the Stefan diffusion or Stefan tube experiment you may have performed in the UOP laboratory
at UG level to determine diffusion coefficient of an evaporating liquid (acetone) in air.

For such mass transfer problem when there is no pressure driven flow, it is often convenient to start up
from the first principle using the molar flux:

𝜕𝐶𝐴
𝜕𝑡
+ ∇. 𝑁𝐴 = 𝑅𝐴 (𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚3 )

dxA
where, 𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝐶𝐷𝐴𝐵 − (1) (𝑁 = 𝑁𝐴 + 𝑁𝐵 )
dz
𝑣𝐶𝐴 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑓𝑙𝑢𝑥 (𝑛𝑜𝑛 − 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑢𝑥)

similar to convective flux in a forced convection because of ∇p.

In this case, such flux may be termed “bulk transport”
(Recall the species conservation equation from the earlier lecture:

𝜕𝐶𝐴
𝜕𝑡
+ 𝒗. ∇𝐶𝐴 = 𝐷𝐴𝐵 ∇2 𝐶𝐴 + 𝑅𝐴 (𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚3 ): If the species conservation equation is written like
this, there is a clear understanding of fluid velocity v in the forced convection, but what is the “velocity”
in the present case? )

Revert to eq. (1). For 1-D mass transfer, SS, no reaction,

67
 𝑑
𝑁
𝑑𝑧 𝐴
=0

→ 𝑁𝐴 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 𝑁𝐴𝑜 (𝑧 = 𝑧1 )
𝑑 𝑑𝑥𝐴
{𝑥 (𝑁
𝑑𝑧 𝐴 𝐴
+ 𝑁𝐵 ) − 𝐶𝐷𝐴𝐵
𝑑𝑧
} =0 − (2)

𝑑
Also, 𝑁
𝑑𝑧 𝐵
= 0. Considering that B is insoluble in the liquid A, there should not be a finite or net flux of
B at the A-B interface -> 𝑁𝐵𝑜 = 0 𝑎𝑡 𝑧 = 𝑧1 .

Therefore, 𝑁𝐵 = 𝑁𝐵𝑜 = 0 everywhere in the gas film (𝑧1 ≤ 𝑧 ≤ 𝑧2 )

dxB
Or, 𝑁𝐵 = 0 = 𝑥𝐵 (𝑁𝐴 + 𝑁𝐵 ) − 𝐶𝐷𝐴𝐵
dz

𝑣𝐶𝐵

𝑑𝑥𝐵
0 = 𝑥𝐵 𝑁𝐴 − 𝐶𝐷𝐴𝐵 𝑑𝑧
-- (3)

𝑛𝑜𝑤, 𝑤𝑒 𝑐𝑎𝑛 𝑒𝑥𝑝𝑙𝑎𝑖𝑛 𝑡ℎ𝑒 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡

𝑛𝑒𝑡 𝑓𝑙𝑢𝑥 convective flux (? ) 𝑎𝑡 𝑧 = 0 (𝑖𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒)
of B of B

See the schematic below. JA + JB = 0 or JB = -JA from the Fick’s first law of equimolar binary diffusion. There
exists a concentration gradient of A from z = 𝑧1 to 𝑧2 . In such case, a concentration gradient of B develops
in the opposite direction from 𝑧2 to 𝑧1 . Thus, at any horizontal plane there is a diffusion flux of B in the
opposite direction of that of A. Also, note that there is a finite concentration gradient at the liquid-vapor
interface for both species A and B. But, B is stagnant (NB = 0) anywhere along the diffusion path. Therefore,
there is a bulk transport of B induced in the opposite direction of the diffusion flux to make the net flux or
transport of B to be zero across any horizontal plane in the film. As stated earlier, there is no forced
convection (no pump or external pressure applied) nor there is a free convection (constant temperature).
Thus, it can be said that there is a diffusion of ‘A’ through the stagnant film of ‘B’.

Revert. Either (1), (2) or (3) will lead to 𝑥𝐴 2 𝑥𝐵 2

𝑧
𝑑𝑥𝐴 𝑑 𝑑𝑥𝐴
𝑁𝐴𝑜 𝑑𝑧
= 𝑑𝑧 (𝐶𝐷𝐴𝐵 𝑑𝑧
) (2nd order ODE)
𝑥𝐴 + 𝑥𝐵 = 1
𝑑 𝐶𝐷𝐴𝐵 𝑑𝑥𝐴 𝑑𝑁
Or, (
𝑑𝑧 1−𝑥𝐴 𝑑𝑧
) =0 ( 𝑑𝑧𝐴 = 0)
𝑥𝐵 1 𝑥𝐴1
𝑧−𝑧1
1−𝑥𝐴 1−𝑥𝐴 𝑧2 −𝑧1
Result is, = [1−𝑥 2 ] Note that concentration gradients are non-linear.
1−𝑥𝐴1 𝐴1

Note: Slope of the concentration gradient for A is not constant but 𝑁𝐴 𝑖𝑠 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡. Slope of the
concentration gradient for B is also not constant, but NB is zero.

𝑑𝑥𝐴 dxA
𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝐶𝐷𝐴𝐵 𝑑𝑧
(Both xA and dz
vary along z)

68
 CD dxA
NA = − (1−xAB) dz
A

@ 𝑧 = 𝑧1 (𝑜𝑟 𝑖𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒)

CD dx CDAB (xA1 −xA2 )

NAo = − 1−xAB ( dzA ) = (z2 −z1 ) logmean (1−xA )
A1 z=z1

Defining k as the mass transfer coefficient: 𝐽 = 𝑘(𝐶𝐴𝑜 − 𝐶𝐴∞ ) ≡ 𝑘(𝐶𝐴1 − 𝐶𝐴2 )

DAB
𝑘 = (z −z )
which is valid within Film theory
2 1 logmax (xB )

− 𝑠𝑡𝑒𝑎𝑑𝑦 − 𝑠𝑡𝑎𝑡𝑒
− 1𝐷 𝑑𝑥
(𝑁𝐵 = 0 = 𝑥𝐵 𝑁𝐴 − 𝐶𝐷𝐴𝐵 𝑑𝑧𝐵 )
− 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑢𝑥 𝑜𝑓 𝐴 𝑟𝑒𝑐𝑜𝑔𝑛𝑖𝑧𝑒𝑑 𝑎𝑠 𝒙𝑨 𝑵 = 𝒙𝑨 𝑵𝑨
𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑓𝑙𝑢𝑥 𝑜𝑓 𝐵 𝑟𝑒𝑐𝑜𝑔𝑛𝑖𝑧𝑒𝑑 𝑎𝑠 𝒙𝑩 𝑵 = 𝒙𝑩 𝑵𝑨 }

Most important: 𝑘 ∝ 𝐷𝐴𝐵 or Sh ∝ 𝑆𝑐 by Film theory.

Also, film theory can simulate (model) a slow moving film (boundary layer) past a surface in turbulent
flow. Yet in such case, boundary layer theory is better to apply.
𝑤𝑒𝑙𝑙 𝑚𝑖𝑥𝑒𝑑
𝑈∞ diffusion of A across
𝐶𝐴2 the ′boundary layer′ film ⊥ to main
⇒
flow ∥ to the surface.
𝐶𝐴1
𝑛𝑜 − 𝑠𝑙𝑖𝑝 (𝑣 = 0)

See Do' s book on adsorption. While determining diffusion coefficient of a binary mixture using a 'Stefan-
dxA
tube', there is an error as high as 20% if NA = xA (NA + NB ) − CDAB dz
is approximated as NA =
dx
−CDAB dzA , i. e. ignoring ′convective − flow′. Also, see BSL (2nd ed) 18B. 2.

Before we conclude this lecture, let us look at the diffusion or transport in a spherical coordinate system
from the perspective of mathematically writing the correct corresponding species conservation equation,
for example in a spherical solid catalyst or a liquid droplet or an air bubble:
𝑑𝐶𝐴
𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝐷𝐴𝐵
𝑑𝑟
0
1 𝑑
What does Continuity tell? 2 (r 2 𝑁𝐴 ) = 0 (SS and no reaction)
r 𝑑𝑟

Therefore, r 2 𝑁𝐴 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡; It follows that 𝑁𝐴 ≠ 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, unlike in the previous Cartesian coordinate
system 𝑁𝐴 was 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡. All it means is that in the latter system both flux (moles/s-cm2) as well as the
rate (moles/s) are constant; in the spherical or cylindrical system the rate is constant but not flux. There
is a clear artifact of surface or interfacial area of the system under consideration:
r -- r+δr: CV = 4πr2δr, and NA = NA(r)

69
Prototype type example 9: Diffusion into a falling film (Surface renewal theory)

𝑝𝑢𝑟𝑒 𝑙𝑖𝑞𝑢𝑖𝑑
𝑥
𝐶𝑂2 /𝑆𝑂2 Velocity profile:
𝛿
𝑥2
𝑣𝑧 = 𝑣𝑚𝑎𝑥 (1 − )
𝛿2
𝑣∞
𝑆𝑆 (From NS equation)
𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤
𝑁𝐹 𝑥 = 𝛿 𝑣𝑧 = 0 (no-slip)
𝑓𝑢𝑙𝑙𝑦 𝑑𝑒𝑣𝑒𝑙𝑜𝑝𝑒𝑑 𝑓𝑙𝑜𝑤
} 𝑑𝑣𝑧
𝑥=0 = 0 (𝑣𝑧 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚)
𝑧 𝑑𝑥

𝐶𝐴 (𝑥, 𝑧) = ?

Species conservation equation:

𝜕𝐶𝐴 𝜕𝐶𝐴 𝜕𝐶𝐴 𝜕2 𝐶 𝜕2 𝐶𝐴

𝜕𝑡
+ 𝑣𝑥 𝜕𝑥
+ 𝑣𝑧 𝜕𝑧
= 𝐷𝐴𝐵 [ 𝜕𝑥 2𝐴 + 𝜕𝑧 2
] + 𝑅𝐴

SS

𝜕𝐶𝐴 𝜕2 𝐶𝐴
Note: 𝜕𝑧
≠ 0; however 𝜕𝑧 2
≈ 0 because Pe >> 1

Non-dimensionalize 𝑥 ∗ = 𝑥⁄𝛿 𝑎𝑛𝑑 𝑧 ∗ = 𝑧⁄𝐿 , where 𝐿𝑐 = 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑙𝑒𝑛𝑔𝑡ℎ = 𝑃𝑒 𝛿 (𝛿 ≪ 𝐿𝑐 )

𝑐

1 𝜕2 𝐶𝐴 𝜕𝐶𝐴 𝜕2 𝐶𝐴
to show that ≪ 1 𝑜𝑟 𝑣𝑧 ≫ 𝐷𝐴𝐵 𝑖𝑓 𝑃𝑒 ≫ 1
𝑃𝑒 2 𝜕𝑧 2 𝜕𝑧 𝜕𝑧 2

𝑥 2 𝜕𝐶𝐴 𝜕2 𝐶𝐴
Therefore, 𝑣𝑚𝑎𝑥 (1 − (𝛿 ) ) 𝜕𝑧
= 𝐷𝐴𝐵 𝜕𝑥 2
(1)

BCs. 𝑧 = 0 𝐶𝐴 = 0 (pure liquid)

𝑥=0 𝐶𝐴 = 𝐶𝐴𝑜 (solubility; in equilibrium with gas phase)

𝜕𝐶𝐴
𝑥 = 𝛿 − 𝐷𝐴𝐵 𝜕𝑥
= 0 (gas does not react with the solid surface; non-reactive surface)

Further simplifications – Slight penetration, short contact time for A in the liquid; A ‘sees’ liquid at velocity
𝑣𝑚𝑎𝑥 and does not sense the presence of the wall far away. Penetration depth 𝑥 ≪
the film thickness 𝛿. The equation is modified as

𝜕𝐶𝐴 𝜕2 𝐶𝐴
𝑣𝑚𝑎𝑥 𝜕𝑧
= 𝐷𝐴𝐵 𝜕𝑥 2
(1’)

𝜕𝐶𝐴
The boundary condition 𝑥 = 𝛿; −𝐷𝐴𝐵 𝜕𝑥
= 0 𝑖𝑠 𝑎𝑙𝑠𝑜 𝑠𝑖𝑚𝑝𝑙𝑖𝑒𝑑 𝑡𝑜 𝑥 → ∞; 𝐶𝐴 = 0 (Pigford (1941) has
solved with the original boundary condition)

70
 Liquid, B

𝑔𝑎𝑠 (𝐴)

Gas is confined within the top layer (hatched) near the gas-liquid interface in the liquid (characteristic
time of diffusion of A in the liquid, i.e., δ2/DAB is much greater than the residence time of the gas, i.e., L/V.)

Solution is sought using Similarity Transform method for a semi-infinite domain (Recall 1st prototype
plate-jerk problem)

𝐶𝐴 𝑥
= 1 − 𝑒𝑟𝑓 { } (2)
𝐶𝐴𝑜 4𝐷 𝑧
√ 𝐴𝐵
𝑣𝑚𝑎𝑥

(You must not re-solve the conservation eq. (1’). Just check that the non-dimensionalized conservation
momentum and mass species conservations equations and boundary conditions for both prototype
examples are identical)

You should also qualitatively plot CA vs z for increasing x and CA vs x for increasing z.

Mass flux from the concentration profiles (eq. 2):

𝜕𝐶𝐴 𝐷𝐴𝐵 𝑣𝑚𝑎𝑥

𝑁𝐴 (𝑧)|𝑥=0 = − 𝐷𝐴𝐵 |
𝜕𝑥 𝑥=0
= 𝐶𝐴𝑜 √ 𝜋𝑧
(You should also solve this as an exercise)

Note that the flux decreases along the plate length, which is expected as the liquid is increasingly absorbed
with the gas.

(Make use of Leibniz rule in differentiating eq. 2:

𝑣 (𝛼)
If 𝜙(𝛼) = ∫𝑣 2(𝛼) 𝑓(𝜂, 𝛼)𝑑𝜂
1

𝑑𝜙 𝑣 𝜕𝑓 𝑑𝑣2 𝑑𝑣1
𝑑𝛼
= ∫𝑣 2 𝜕𝛼 𝑑𝜂 + 𝑓( 𝑣2 , 𝛼) 𝑑𝛼
− 𝑓(𝑣1 , 𝛼) 𝑑𝛼
1

2 𝑥
In our case, 𝑓(𝜂, 𝛼) = 𝑒 −𝜂 , 𝑣1 = 0, 𝑣2 = 4𝐷 𝑧
√ 𝐴𝐵
𝑪𝒐𝒎𝒑𝒂𝒓𝒆: 𝑣𝑚𝑎𝑥

𝑑𝑣2 1 𝑑𝑣1 2 2
𝛼 = 𝑥; = ; = 0; 𝑓(𝑣2 , 𝛼) = 𝑒−𝜂 )
𝑑𝛼 4𝐷 𝑧 𝑑𝛼 √𝜋
√ 𝐴𝐵
𝑣𝑚𝑎𝑥

71
Total mass transfer per unit time,

𝑚𝑜𝑙𝑒𝑠 𝐿 𝑜 𝐷𝐴𝐵 𝑣𝑚𝑎𝑥 𝐿 −1⁄

𝑊𝐴 ( 𝑠
) = 𝑤 ∫0 𝑁𝐴 |𝑥=0 𝑑𝑧 = 𝑤𝐶𝐴𝑜 √
𝜋
∫0 𝑧 2 𝑑𝑧 (integrated over the contact length L)

4𝐷𝐴𝐵 𝑣𝑚𝑎𝑥
= 𝑤𝐿𝐶𝐴𝑜 √ 𝜋𝐿

Under SS condition, you should be able to show (as an exercise/see BSL 18C.3) that the total mass transfer
rate WA over the film height L is the same as that convected across the horizontal plane at L, considering
that a pure liquid enters the system from the top:
0 x
∞
= 𝑣𝑚𝑎𝑥 𝑤 ∫0 𝐶𝐴 |𝑧=𝐿 𝑑𝑥 δx CA(z,x) L

Again, 𝐽 = 𝑘(𝐶𝐴𝑜 − 0) , 𝑤ℎ𝑒𝑟𝑒 k is the mass transfer coefficient WA

4𝐷𝐴𝐵 𝑣𝑚𝑎𝑥
𝑘∝√ z
𝜋𝐿

𝑣𝑚𝑎𝑥 surface renewal rate

𝑎𝑛𝑑 𝑘 ∝ √ ~√ or √1/τ where τ is the contact time.
𝐿 different from film theory

𝑘 ∝ √𝑫𝑨𝑩 or Sh ∝ Sc1/2 ∶ Surface renewal theory

(Note: Film 𝒐𝒓 𝑻𝒘𝒐 − 𝒇𝒊𝒍𝒎 𝒐𝒓 𝒔𝒕𝒂𝒈𝒏𝒂𝒏𝒕 𝒇𝒊𝒍𝒎 𝒕𝒉𝒆𝒐𝒓𝒚 𝒑𝒓𝒆𝒅𝒊𝒄𝒕𝒔 𝒌 ∝ 𝑫𝑨𝑩 )

Ex: Consider the example of diffusion from an air bubble moving upward, in a liquid (or a stationary
bubble in a liquid flowing down past the air bubble). An equivalent example is also the absorption of a
gas into the falling liquid droplet.

𝐶𝐴 = 0
𝑣𝑡

𝑣∞ → 0
𝑔𝑎𝑠 − 𝑏𝑢𝑏𝑏𝑙𝑒
𝐶𝐴 𝑑𝑖𝑎𝑚𝑒𝑡𝑒𝑟 ~ 𝐷

𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛

Pure liquid flows down past the bubble. Liquid flow induces circulation of gas inside the bubble through
the interfacial shear force. The interfacial velocity is the same as vt, terminal velocity of rising bubble.

72
Levich has extensively discussed such mass transfer related problems, at least one of which will be later
discussed in this course. BSL (18.5.1) and Deen (10.1) have also addressed such problems. Solving the
problem requires first solving the velocity fields inside as well as outside the bubbles, and then solving the
concentration profiles with the velocity profiles superimposed. The velocity fields are solved using
streaming function. Concentration profiles have also been solved using stream functions.

In this lecture, we will restrict our discussion qualitatively. The first important thing to note is that this
problem is mechanistically analogous to the falling film over the vertical plate discussed earlier. Why?

⇒ Surface is renewed continuously by the pure liquid flowing down past the bubble, as the circulating gas
keeps the inside surface concentration constant. Thus, the boundary conditions are identical in two cases.
It is a clear that a complex-looking problem has been solved or addressed by using a simple analogy!

Thus, one can show that 𝑃𝑢𝑟𝑒 𝑙𝑖𝑞𝑢𝑖𝑑,

𝑘 ∝ √𝐷𝐴𝐵 , 𝑎𝑛𝑑

4𝐷𝐴𝐵 𝑣𝑡
𝑁𝐴𝑣𝑔 = √ 𝐶𝐴𝑜 𝐶𝑂2 L = πD
𝜋𝐷

𝑣
Surface is renewed @ 𝑟𝑎𝑡𝑒√ 𝐷𝑡 ; 𝑣𝑚𝑎𝑥
Here, surface is renewed @ 𝑟𝑎𝑡𝑒√
𝐿

Similarly, consider the trickle flow of a liquid over a packed bed of spherical catalysts:

𝑣
mass transfer rate ∝ √𝐷𝐴𝐵 and surface renewal rate ∝ √ 𝐷𝑡.

To this end, note that there is also “penetration theory” similar to the film theory, or surface renewal
theory, or boundary layer theory, to calculate mass transfer coefficient or mass transfer flux of a gas into
liquid. Each of these theories is applicable to different hydrodynamic conditions. The penetration theory
considers “eddies” randomly moving around in liquid flow under turbulent condition. Eddies reach the
gas-liquid interface, come in contact with the gas for different times, and return to the bulk liquid. During
this contact time, gas diffuses into eddies. In this case as well, the gas-liquid surface (interface) is renewed,
and 𝑘 ∝ √𝐷𝐴𝐵 or Sh ∝ Sc1/2

gas

liquid

(turbulent flow conditions)

73
 Lecture #15
Prototype example 10: Taylor Dispersion or Taylor – Aris Diffusion/Dispersion

- Taylor (1953, 1954) and Aris (1956) pioneered and extensively studied diffusion/dispersion in a
laminar tubular flow.
- Consider a fully developed laminar flow (𝑅𝑒 < 2100) in a long tube. A RTD or tracer experiment is
performed by injecting a dose (′𝛿 ′ 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛) of tracer at the inlet to the tube of radius R. We are
interested in studying concentration profiles in the tube, in particular response at the exit of the
tube.
𝐿
↑𝑟
𝐶(𝑡, 𝑟, 𝑧) → 𝐶̅ (𝑡, 𝑧) = ?
Z

𝛿(𝑡) 𝑟2
𝑉 = 𝑉𝑚𝑎𝑥 (1 − )
𝑅2
Species balance equation:

𝜕𝐶
𝜕𝑡
+ 𝑉. ∇𝐶 = 𝐷𝑚 ∇2 𝐶 + (−𝑟𝐴 ) (𝑅𝑒 < 2100), (𝐷𝑚 ≡ 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛)

𝜕𝐶 𝑟2 𝜕𝐶 𝜕2 𝐶 𝐷𝑚 𝜕 𝜕𝐶
Or 𝜕𝑡
+ 𝑉𝑚𝑎𝑥 (1 − 𝑅2 ) 𝜕𝑧 = 𝐷𝑚 𝜕𝑧 2
+ 𝑟 𝜕𝑟
(𝑟 𝜕𝑟 )

𝜕2 𝐶 𝑟2 𝜕𝐶
We have seen earlier that 𝐷𝑚 𝜕𝑧 2
≪ 𝑉𝑚𝑎𝑥 (1 − 𝑅2 ) 𝜕𝑧 if 𝑃𝑒 ≫ 1. Recall the equation in the non-
dimensionalized form when

1 𝜕2 𝐶 ∗ 𝜕𝐶 ∗
𝑃𝑒2 𝜕𝜓2
≪ (1 − 𝜉 2 ) 𝜕𝜓 (𝑤ℎ𝑒𝑟𝑒 𝑧𝑐 ≡ 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 𝑙𝑒𝑛𝑔𝑡ℎ = 𝑅𝑃𝑒)

(Diffusion in axial direction ≪ Convection)

𝑉 𝑡 𝑀
Introducing 𝜉 = 𝑟⁄𝑅 , 𝜓 = 𝑧⁄𝐿 , 𝐶 ∗ = 𝐶⁄𝐶 , 𝜏 = 𝑡⁄𝐿 = 𝑚𝑎𝑥
𝐿
; 𝐶𝐴𝑜 = 𝑄𝛥𝑡 (dose 𝑓𝑜𝑟 𝑠ℎ𝑜𝑟𝑡 𝑡𝑖𝑚𝑒)
𝐴𝑜 ⁄𝑉
𝑚𝑎𝑥

The time-dependent 2D (z-r) conservation equation is transformed as

𝜕2 𝐶 ∗ 1 𝜕𝐶 ∗ 𝑅2 𝑉𝑚𝑎𝑥 𝜕𝐶 ∗ 𝑅2 𝑉𝑚𝑎𝑥 𝜕𝐶 ∗
( 𝜕𝜉2 + 𝜉 𝜕𝜉
) =( 𝐷𝑚 𝐿
) 𝜕𝜏 +( 𝐷𝑚 𝐿
) (1 − 𝜉 2 ) 𝜕𝜓 − (1)
𝜕𝐶 ∗
IC: 𝐶 ∗ = 0, for all 𝜉 and 𝜓. BCs. 1. 𝜉 = 0 𝜕𝜉
=0
𝜕𝐶 ∗
2. 𝜉 = 1 − 𝜕𝜉 = 0
3. 𝜓 = 0 𝐶 ∗ = 1 for a short time 𝛥𝑡; 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒; 𝑜𝑟 𝜓 → ∞, 𝐶 ∗ = 0

74
Also note that 𝑧𝑐 = 𝑅𝑃𝑒 = 𝑅(𝑅𝑒𝑆𝑐)

𝑅𝑉𝑚𝑎𝑥 𝜈 𝑅2 𝑉𝑚𝑎𝑥
=𝑅 𝜈
∙𝐷 = ( 𝐷𝑚
)
𝑚

𝑧
, Or the term in the parenthesis of eq. (1) is nothing but ( 𝑐⁄𝐿). Taylor sought a solution valid for long
time, or when enough time has lapsed and pulse has travelled a distance 𝑙𝑝 such that

𝑙𝑝 𝑅2
⁄𝑉 ≫ (3.8)2 𝐷 (𝑠𝑒𝑒 𝐵𝑆𝐿)
𝑚𝑎𝑥 𝑚

Plawsky in Transport Phenomena Fundamentals has provided the solution in details:

𝑅2⁄
𝑅2 𝑉𝑚𝑎𝑥 𝐷𝑚 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 𝑅
𝛽𝐷 ≡ = 𝐿
= (𝑃𝑒 ) (A)
𝐷𝑚 𝐿 ( ⁄𝑉 ) 𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 𝐿
𝑚𝑎𝑥

If 𝛽𝐷 is large, then diffusion is not important and dispersion or spreading is determined by convection
alone and the tracer spreads in radial direction similar to the velocity profile:

𝐶𝐴𝑜 𝐶𝐴̅

𝑠ℎ𝑜𝑟𝑡 𝑡𝑖𝑚𝑒 𝑙𝑜𝑛𝑔 𝑡𝑖𝑚𝑒

𝑖𝑛𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡
𝑏𝑎𝑛𝑑 𝑜𝑓 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑛 𝑟𝑎𝑑𝑖𝑎𝑙
𝑑𝑜𝑠𝑒
𝑑𝑜𝑠𝑒 𝑠𝑝𝑟𝑒𝑎𝑑𝑠 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛, 𝑏𝑢𝑡 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑑
(𝑢𝑛𝑖𝑓𝑜𝑟𝑚 𝑎𝑐𝑟𝑜𝑠𝑠
𝑣𝑒𝑟𝑦 𝑚𝑢𝑐ℎ 𝑠𝑖𝑚𝑖𝑙𝑎𝑟 𝑡𝑜 𝑠𝑝𝑟𝑒𝑎𝑑𝑖𝑛𝑔 𝑖𝑛 𝑎𝑥𝑖𝑎𝑙
𝑟𝑎𝑑𝑖𝑢𝑠)
𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑟𝑜𝑓𝑖𝑙𝑒 { 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛.
𝑖𝑛 𝑡ℎ𝑒 𝑡𝑢𝑏𝑒 }

If 𝛽𝐷 is small, then diffusion will wipe out or remove radial gradient in concentration, in which case
diffusion is important. To determine when diffusion is important, revisit equation (1) but neglect
convection,

𝜕𝐶 ∗ 𝜕2 𝐶 ∗ 1 𝜕𝐶 ∗
𝛽𝐷 𝜕𝜏
= ( 𝜕𝜉2 + 𝜉 𝜕𝜉
)

Recall previous prototype equations for which the method of separation of variable were used to solve.

𝐶 ∗ = 𝑒 −𝛼𝜏 𝐽𝑜 (√𝛽𝐷 𝛼𝜉) (𝛼 ≡ 𝑐𝑜𝑛𝑠𝑡. 𝑜𝑓 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛)

Applying 𝐵𝐶 2, 𝐽1 (√𝛽𝐷 𝛼) = 0. There are multiple roots, but consider first root only as the Bessel function
decays fast.

√𝛽𝐷 𝛼 = 3.8 (Recall Bessel functions)

1
This solution represents the time it takes for any radial variation in concentration to decrease to of the
𝑒

75
original value, i.e., for longer times or larger distance in the tube. Therefore, for diffusion to be important
𝜷𝑫 ≪ (𝟑. 𝟖)𝟐 ;

Let us address now diffusion-controlled response or spreading.

The solution is sought by switching over to Lagrangian co-ordinate in which case the observer or co-
𝑉𝑚𝑎𝑥 𝜏
ordinate system moves with the average speed ( 2
) of the flow; 𝑧̅ = 𝑧 − 𝑣̅𝑧 (𝑡)𝑡 or 𝜁 = 𝜓 − 2 (𝜏 is the
dimensionless time (𝑡/ 𝐿⁄𝑉 ) and 𝜓 is the dimensionless distance (𝑧⁄𝐿) ).The modified equation is
𝑚𝑎𝑥

𝜕2 𝐶 ∗ 1 𝜕𝐶 ∗ 1 𝜕𝐶 ∗ 1 𝜕 𝜕𝐶 ∗ 1 𝜕𝐶 ∗
( 𝜕𝜉2 + 𝜉 𝜕𝜉
) = 𝛽𝐷 (2 − 𝜉 2 ) 𝜕𝜁
Or 𝜉 𝜕𝜉
(𝜉 𝜕𝜉
) = 𝛽𝐷 (2 − 𝜉 2 ) 𝜕𝜁

Under the present case, when diffusion controls the spreading, or the spreading is rapid after a sufficiently
long time (Eulerian approach), or after the pulse has travelled a large distance in the tube (Lagrangian
𝜕𝐶 ∗ ̅
𝜕𝐶
approach), may be assumed to be constant along z or 𝜓 direction, as :
𝜕𝜁 𝜕𝜁

1 𝜕 𝜕𝐶 ∗ 1 ̅
𝜕𝐶
𝜉 𝜕𝜉
(𝜉 𝜕𝜉
) = 𝛽𝐷 (2 − 𝜉 2 ) 𝜕𝜁 : (recall similar scenario in prototype example on constant wall heat flux)

Applying method of separation of variables and integrating

𝛽𝐷 1 ̅
𝜕𝐶
𝐶 ∗ = 𝐶𝑜 + 8
(𝜉 2 − 2 𝜉 4 ) 𝜕𝜁 : (𝐶 ∗ = 𝐶𝑜 𝜁 = 0 𝑜𝑟 𝜓 = 𝜏⁄2)

Thus, the average mass or molar flux across the tube cross-section as seen by the observer moving with
the mean velocity:

1 1 𝑅 𝑟2
𝑁𝐴 ~ − 2𝑉𝑚𝑎𝑥 ∫0 𝐶 ∗ ( − 𝜉 2 ) 𝜉𝑑𝜉 (integral term equivalent to ∫0 𝐶(𝑟, 𝑧) (1 − ) 2𝜋𝑟𝑑𝑟 )
2 𝑅2

Substituting 𝐶 ∗ in the above equation with 𝛽𝐷 from (𝐴) ,

𝑅2 𝑉 2 ̅
𝜕𝐶
𝑁𝐴 ~ − ( 192𝑚𝑎𝑥
𝐷
) 𝜕𝜁
𝑚

𝜕𝐶̅ 𝑅2 𝑉𝑚𝑎𝑥
2
~ −𝐾 𝜕𝜁 , where, 𝐾 = dispersion coefficient = 192 𝐷𝑚

(Axial change in concentration as seen by the observer is approximated by the radially averaged
concentration, as a sufficient spreading has occurred in the radial direction)

The above flux-equation implies that the 2D concentration distributions in a parabolic tubular flow can be
approximated as a 1D (axial) concentration distribution in a flow having a radially flat velocity profile with
𝑉
the mean average velocity 𝑉̅ = 𝑚𝑎𝑥, and axial dispersion coefficient as 𝐾. In other words, the general 2D
2
species balance equation (1) written for a tracer movement in a parabolic flow is replaced with

76
 𝜕𝐶̅ 𝜕𝐶̅ 𝜕 2 𝐶̅ 𝑅2 𝑉𝑚𝑎𝑥
2

𝜕𝑡
+ 𝑉̅ 𝜕𝑧 = 𝐾 𝜕𝑧2 − (2) 𝑤ℎ𝑒𝑟𝑒, 𝐾 = 192 𝐷𝑚

𝑅2 𝑉𝑚𝑎𝑥
2
Later, Aris modified the expression for 𝐾 for the full time domain: 𝐾 = 𝐷𝑚 + 192 𝐷𝑚

𝜕𝐶
Some notes: (1) 𝐾 can be expressed as 𝐽 = −𝐾 𝜕𝑧 where, 𝐾 is the Taylor-Aris dispersion coefficient
for Re < 2100. Note that ′𝐾′ is not a physical or thermodynamic property like ′𝐷𝑚 ′, and the value clearly
depends on 𝑉𝑚𝑎𝑥 & 𝐷𝑚 .

(2) If 𝐷𝑚 increases, ‘dispersion’ decreases, or vice-versa, which seems to be weird. But this is the way,
‘Taylor dispersion’ behaves. In general, “diffusion” is responsible for creating a concentration gradient; it
is the convection which wipes out the gradient. In this case, however, convection or the radially parabolic
velocity profile in the tube smears the radially flat concentration of the dose (input) at the tube-inlet to
the parabolic profile; diffusion wipes out the gradient later in the tube.
(3) Equation (2) represents the following scenario:

𝐴𝑥𝑖𝑎𝑙 𝑑𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 (𝐾) 𝐶(𝑡, 𝑧) = ?

𝛿(𝑡) 𝑉𝑚𝑎𝑥
𝑉̅ = (𝑝𝑙𝑢𝑔 𝑓𝑙𝑜𝑤! 𝑏𝑢𝑡 𝑅𝑒𝑦𝑛𝑜𝑙𝑑𝑠 𝑛𝑢𝑚𝑏𝑒𝑟 𝑖𝑠 𝑙𝑒𝑠𝑠 𝑡ℎ𝑎𝑛 2100‼ )
2

(4) 1𝑠𝑡 2𝑛𝑑 3𝑟𝑑 4𝑡ℎ

𝐾 𝑇𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑓𝑙𝑜𝑤
𝑜𝑟 𝐷𝐿 ′𝐾 ′
(𝑒𝑑𝑑𝑖𝑒𝑠)
𝐷𝑚
2100

𝑉𝑑
𝑅𝑒 =
𝜈
2 2
𝑅 𝑉𝑚𝑎𝑥
𝐾 = 𝐷𝑚 +
192 𝐷𝑚

For the region ‘2’, a commonly asked question is: why does axial dispersion coefficient (𝐷𝐿 𝑜𝑟 𝐾) exceed
molecular coefficient (𝐷𝑚 ) even if 𝑅𝑒 < 2100?

The answer is: We are referring to Taylor dispersion, when the parabolic velocity profile can be replaced
with a plug flow profile, with the radial concentration removed, but the axial ‘dispersion’ included. The
total mass is conserved with the radially distributed amount of the solute redistributed along ′𝑧 ′ or axial
direction; thus K or DL increases with increasing velocity or Reynolds number.

77
 Lecture #16
Prototype example 11: Strained diffusion-controlled reaction
⇒ Consider a strained diffusion-controlled reaction. Specifically, we are discussing here the unmixed
combustion of a fuel in a one-step irreversible reaction in a laminar diffusion flame formed by the
axisymmetric planer counter flows.

𝑦 +∞

−∞ 𝑥 ∞
0

𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑙𝑎𝑛𝑒
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑧𝑜𝑛𝑒 (ℎ𝑦𝑑𝑟𝑜𝑑𝑦𝑛𝑎𝑚𝑖𝑐𝑎𝑙𝑙𝑦)
𝑖𝑠 𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑
−∞

Reaction is, 𝐴 + 𝜈𝐵 → 𝐶 ⇒ 𝑟1 = 𝜈𝑟2

(2) (1) (3) 𝑟𝐵 = 𝜈𝑟𝐴

(𝑟1 − 𝜈𝑟2 = 0)

General case concentration profiles:

(1)
𝐵, 𝑓𝑢𝑒𝑙
(𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 + 𝑛𝑜 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

(3), 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

(𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 + 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)
(2)
𝐴, 𝑜𝑥𝑖𝑑𝑎𝑛𝑡 (𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 + 𝑛𝑜 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

Solution: All convection has done is to create a stagnation plane/zone (small velocities) in which diffusion
is important.

Velocity field (imposed from outside)

𝑣𝑥 = 𝜖 𝑥 Assume 𝜌 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

78
 𝑣∞ −0 𝑑𝑣𝑦
𝑣𝑦 = −𝜖 𝑦 𝜖 ≡ 𝑠𝑡𝑟𝑎𝑖𝑛𝑒𝑑 𝑟𝑎𝑡𝑒 ~ 𝐿
(~ 𝑑𝑥
)

Depending on 𝜖 𝑜𝑟 𝜖(𝑡), the ‘zone’ of small velocities will be different. Solve for mass fraction profiles:
𝜔1 (𝑦), 𝜔2 (𝑦) ≠ 𝑥 (𝑥 𝑖𝑠 𝑡ℎ𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑖𝑛 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒𝑙𝑦 𝑙𝑜𝑛𝑔 𝑝𝑙𝑎𝑡𝑒)

(If reaction is very fast zone will be of zero-thickness, viz. plane/sheet)

𝜕𝑣𝑥 𝜕𝑣𝑦 𝑣𝑥 = 𝜖 𝑥
⇒ Continuity ⇒ + =0 𝑐ℎ𝑒𝑐𝑘: 𝑣𝑦 = −𝜖 𝑦}
𝜕𝑥 𝜕𝑦

𝑦 − momentum balance is not required although pressure increases from 𝑃∞ 𝑡𝑜 𝑃𝑜 (𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑙𝑎𝑛𝑒)
2
as velocity 𝑣𝑦 decreases from 𝑣∞ 𝑡𝑜 0: 𝜌𝑣∞ ~Δ𝑝

Species:

𝜕𝜔1 𝜕𝜔1 𝜕𝜔 1 𝜕 𝜕𝜔
(1) 𝜕𝑡
+ 𝑣𝑥 𝜕𝑥
+ 𝑣𝑦 𝜕𝑦1 = 𝜌 𝜕𝑦 (𝜌𝐷 𝜕𝑦1 ) + 𝑟1
𝜕𝜔1 𝜕𝜔 𝜕𝜔 1 𝜕 𝜕𝜔
𝜕𝑡
+ 𝜖𝑥 𝜕𝑥1 − 𝜖𝑦 𝜕𝑦1 = 𝜌 𝜕𝑦 (𝜌𝐷 𝜕𝑦1 ) + 𝑟1
𝜕𝜔2 𝜕𝜔 𝜕𝜔 1 𝜕 𝜕𝜔
(2) 𝜕𝑡
+ 𝜖𝑥 𝜕𝑥2 − 𝜖𝑦 𝜕𝑦2 = 𝜌 𝜕𝑦 (𝜌𝐷 𝜕𝑦2 ) + 𝑟2

𝑡 = 0 reaction zone will be at stagnation zone, and 𝑡 > 0 reaction zone moves.

𝜔1 = 1 𝜔2 = 1
𝑡=0 }𝑦 > 0 } 𝑦<0
𝜔2 = 0 𝜔1 = 0

𝑦 → ∞ 𝜔1 = 1, 𝜔2 = 0
𝑡 > 0 𝑜𝑟 0+
𝑦 → −∞ 𝜔2 = 1, 𝜔1 = 0

From (1) & (2) with 𝑟1 − 𝜈𝑟2 = 0

𝜕(𝜔1 −𝜈𝜔2 ) 𝜕 𝜕 𝜕
𝜕𝑡
− 𝜖 𝑦 𝜕𝑦 (𝜔1 − 𝜈𝜔2 ) = 𝜕𝑦 𝐷 𝜕𝑦 (𝜔1 − 𝜈𝜔2 )

Let 𝑘 = (𝜔1 − 𝜈𝜔2 )

𝜕𝑘 𝜕𝑘 𝜕 𝜕𝑘
𝜕𝑡
− 𝜖 𝑦 𝜕𝑦 − 𝜕𝑦 (𝐷 𝜕𝑦) = 0 (reaction can be slow or fast)

Note: If the reaction is very fast, zone will be of zero thickness, and at the reaction zone/plane, 𝜔1 = 𝜔2 =
0 (species 𝐴, 𝐵 will be destroyed completely & instantaneously, leaving behind the product C only) ⇒ This
is what a flame (a torch of a burner) does!

𝑅𝑒𝑐𝑎𝑙𝑙 𝑡ℎ𝑒 𝑆𝑡𝑜𝑘𝑒 ′ 𝑠 1𝑠𝑡 𝑝𝑟𝑜𝑏𝑙𝑒𝑚.

Solution: Try 𝜂 = 𝑦𝑓(𝑡) {𝑆𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑏𝑦 𝑆𝑖𝑚𝑖𝑙𝑎𝑟𝑖𝑡𝑦 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚

(Combination of variables)

𝜕𝑘 𝑑𝑘 𝜕𝜂 𝑑𝑘 𝜂 𝑑𝑘
= = 𝑦𝑓 ′ = 𝑓′
𝜕𝑡 𝑑𝜂 𝜕𝑡 𝑑𝜂 𝑓 𝑑𝜂

79
𝜕𝑘 𝑑𝑘 𝜕2 𝑘 𝑑2 𝑘
𝜕𝑦
= 𝑓 𝑑𝜂 , 𝜕𝑦 2
= 𝑓 𝑑𝜂2

The equation becomes,

𝑓′ 𝑑𝑘 𝜂 𝜕𝑘 𝑑2 𝑘
(𝜂 𝑓 𝑑𝜂
− 𝜖 𝑓 𝑓 𝜕𝜂 − 𝐷𝑓 𝑑𝜂2 ) = 0

1 𝑓′ 𝑑𝑘 𝑑2 𝑘
(𝑓 ( 𝑓 − 𝜖) 𝜂 𝑑𝜂 − 𝐷 𝑑𝜂2 ) = 0 − (3)

𝑓(𝑡) must disappear for the method of combination to work!

1 𝑓′
(
𝑓 𝑓
− 𝜖) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 2 (𝑎𝑠𝑠𝑢𝑚𝑒) − (4)

𝑓 ′ − 𝜖𝑓 − 2𝑓 2 = 0

1 −1⁄2
𝑓(𝑡) ⇒ 𝑓 = { (1 − 𝑒 −2𝜖𝑡 )} (5)
2𝜖

1 −1⁄2
Therefore, 𝜂 = 𝑦𝑓(𝑡) = 𝑦 {2𝜖 (1 − 𝑒 −2𝜖𝑡 )} - (6)

(𝑦 𝑎𝑛𝑑 𝑡 ℎ𝑎𝑣𝑒 𝑏𝑒𝑒𝑛 𝑐𝑜𝑚𝑏𝑖𝑛𝑒𝑑 𝑡𝑜 𝜂)

Put (4) in (3); we are now solving (without ‘D’)

𝑑𝑘 𝑑2 𝑘
2𝜂 𝑑𝜂 − 𝑑𝜂2 = 0 where, 𝜂 = 𝜂(𝑦, 𝑡) from (6) - (7)

⇒ 2 BCs are required for 𝑘(𝜂)

𝑦>0 𝜔1 = 1, 𝜔2 = 0 𝜂>0 𝑘=1

BCs: 𝑡 = 0 for }≡
𝑦<0 𝜔2 = 1, 𝜔1 = 0 𝜂 < 0 𝑘 = −𝜈

𝑓𝑜𝑟 𝑎𝑙𝑙 𝑡 > 0 𝑦 → ∞ 𝜔1 → 1, 𝜔2 = 0 𝜂→∞ 𝑘=1

}≡
𝑜𝑟 𝑡 + 𝑦 → −∞ 𝜔2 → 1, 𝜔1 = 0 𝜂 → −∞ 𝑘 = −𝜈

𝑘 = 𝐴 + 𝐵 erf(𝜂)

1=𝐴+𝐵
} 𝐵𝐶𝑠 →
−𝜈 = 𝐴 − 𝐵
1 1
𝑘 = 𝜔1 − 𝜈𝜔2 = (1 − 𝜈) + 2 (1 + 𝜈)erf(𝜂) -- (8)
2

80
Now, let us ask where is the flame? Where is the reaction zone?

@ Reaction zone/flame: 𝑘 = 0 (fast kinetics)

𝑘 = 𝜔1 − 𝜈𝜔2
(𝑆𝑒𝑒 𝐵𝐶𝑠: − 𝜈 ≤ 𝑘 ≤ 1) ( )
𝜔1 = 𝜔2 = 0

Try locating the movement of flame.

𝜂 = 𝜂𝑓𝑙𝑎𝑚𝑒 (𝑦, 𝑡) 𝑎𝑡 𝑘 = 0

(1−𝜈)
Therefore, erf(𝜂𝐹 ) = − (1+𝜈)

(1−𝜈)
𝜂𝐹 = erf −1 {− (1+𝜈)} = 𝑦𝐹 𝑓(𝑡)

1⁄
1 −2𝜖𝑡 )} 2 (1−𝜈)
Time-consideration of flame movement: 𝑦𝐹 = {2𝜖 (1 −𝑒 erf −1 {− (1+𝜈)} (9) from eq. (5)

Check 𝑡 = 0, 𝑦 = 0 reaction zone is at stagnation plane

𝑡 > 0 , 𝑦 can move up or down depending upon sign of

(1−𝜈) ν > 1 y is + ve plane moves up
erf −1 {− (1+𝜈)} 𝑖. 𝑒. 𝑖𝑓
ν < 1 y is − ve plane moves down

𝑖. 𝑒. 𝜈 > 1 𝐵 is getting consumed faster, flame moves towards B (up in the present case)

𝜈 < 1 𝐴 is getting consumed faster, flame moves towards A (down in the present case)

𝑡=0 𝑡 = 0+ (𝑡1 ) 𝑡2 > 𝑡1

𝐹(𝐵) 𝐹(𝐵) 𝐹(𝐵)

𝑂𝑥𝑖𝑑𝑎𝑛𝑡 (𝐴)
𝑂𝑥𝑖𝑑𝑎𝑛𝑡 (𝐴) 𝑂𝑥𝑖𝑑𝑎𝑛𝑡 (𝐴)

0 𝑦1 0 𝑦2
0 𝑦

Plane moves from stagnation plane at 𝑦 = 0 𝑡𝑜 𝑦1 (𝑡1 ) 𝑡𝑜 𝑦2 (𝑡2 ) 𝑡𝑜 𝑎 𝑆𝑆 𝑙𝑒𝑣𝑒𝑙, given by equation (6).

For small 𝑡,

1 − 𝑒 −2𝜖𝑡 ≈ 2𝜖𝑡 + ⋯ 𝑒𝑥𝑝𝑎𝑛𝑑 𝑖𝑡

(1−𝜈) 𝑦𝐹
𝑇ℎ𝑒𝑟𝑒𝑓𝑜𝑟𝑒, 𝑦𝐹 ≈ √𝑡 erf −1 {− (1+𝜈)} 𝑎𝑛𝑑 𝜂𝐹 ≈ (10)
√𝑡

Location of flame

81
 𝑦 1 −1⁄2
With D incorporated, 𝜂 = 𝑦𝑓(𝑡) = { (1 − 𝑒 −2𝜖𝑡 )} (6′ )
√𝐷𝑡 2𝜖

1⁄
1 2 (1 − 𝜈)
𝑦𝐹 = 2√𝐷 { (1 − 𝑒 −2𝜖𝑡 )} erf −1 {− } − (9′)
2𝜖 (1 + 𝜈)

(1−𝜈) 𝑦𝐹
And approximated equations: 𝑦𝐹 ≈ 2√𝐷𝑡 erf −1 {− (1+𝜈)} and 𝜂𝐹 ≈ 2√𝐷𝑡
---- (10’)

Fuel consumption rate,

𝜕𝜔1 𝐷𝜖 1 2
𝑚̇ = 𝜌𝐷 (@ 𝑓𝑙𝑎𝑚𝑒) = 𝜌 √2𝜋 1 ∙ (1 + 𝜈)𝑒 −𝜂𝐹
𝜕𝑦 (1−𝑒 −2𝜖𝑡 ) ⁄2

𝑡 → ∞ 𝑦𝐹 → 𝑓𝑖𝑛𝑖𝑡𝑒
Note:
1⁄ 𝑆𝑆(𝑔𝑒𝑛𝑒𝑟𝑎𝑙𝑙𝑦) 𝑎𝑡 > 1⁄𝜖
𝑚̇ √𝑡

1⁄
𝜖
𝑡

(Fuel consumption is large to begin with; decreases to a SS value at SS or 𝑡 → ∞ or 𝑦𝐹 𝑎𝑡 𝑡 → ∞)

stagnation plane
reaction plane

- Convection has created diffusion- limitation condition

- The straining of the flame is physically significant because it increases the rate of reactant
consumption, not only by increasing the interfacial exposure of fuel to oxidant, but also by
convecting additional reactant to the flame. (SIAM J. APPL. MATH. VOI. 28, No. 2, 1975)
- Flame creates a reaction-plane away from the hydrodynamic-stagnation plane, where the
reaction is very fast (𝜔1 = 𝜔2 = 0)
- Similarity Method was used to solve ′𝑘′ or track the movement of flame.

82
 Lecture #17
Prototype example 12: Simultaneous momentum, heat, and mass transport
⇒ Condensation of steam in the presence of an inert gas (air) on a vertical surface (plate):
The example below will show that the effect of air mixed in steam is to reduce the condensation heat
transfer coefficient, possibly by one-order of magnitude. This considerable decrease in ′ℎ′ can occur at
the air quality as low as 0.1, implying that the condensers must be designed/operated with a proper
venting arrangement so that the non-condensable gas is instantly removed from the system. It is never
safe to assume that non-condensable gas will not be present. See In Boiling, Condensation and Gas-liquid
flow by Whalley, for details.
Before analysing the condensation of steam in the presence of air, let us begin with that of a saturated
steam without air:
𝑥=0 𝑦
𝑠𝑡𝑒𝑎𝑚 (vapour-phase side)
Coolant side
? 𝑇𝑠𝑎𝑡 (𝑃)
𝑇𝑤
𝛿(𝑥) 𝑎𝑠𝑠𝑢𝑚𝑖𝑛𝑔 𝜇𝑓 ≫ 𝜇𝑣
𝑔 𝜕𝑣𝑥 }
𝑣∞ = 0 = 0 𝑎𝑡 𝑦 = 𝛿
𝜕𝑦
x=𝐿 𝑣(𝑥) 𝑣𝑚𝑎𝑥

𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑎𝑡𝑒
/𝑙𝑖𝑞𝑢𝑖𝑑 𝑓𝑖𝑙𝑚

One can write down the continuity & NS equation or momentum balance equation in 𝑥-direction with
suitable BCs and approximation. And, the velocity profile of liquid in the film (condensate), assuming
laminar flow, can be derived as

𝑔(𝜌𝑓 −𝜌𝑣 )𝛿𝑓2 2

𝑦 𝑦
𝑣𝑥 (𝑦) = [2 (𝛿 ) − (𝛿 ) ] [𝛿𝑓 = 𝛿𝑓 (𝑥)] (This is another homework problem)
2𝜇𝑓 𝑓 𝑓

See Koh et.al (1961), Int 𝐽 Heat & Mass transfer for detailed analysis. You can also check the book by
McCabe-Smith or Kern. In general, acceleration term can be neglected assuming fully developed laminar
𝜕𝑣𝑥 𝑘𝑔
flow regime 𝜕𝑥
~0. From the velocity profile 𝑣𝑥 (𝑦), 𝑚̇(𝑥) or local mass flow rate ( 𝑠 ) can be calculated
as
𝛿 𝑔𝜌𝑓 (𝜌𝑓 −𝜌𝑣 )𝛿𝑓3 (𝑥)
𝑚̇(𝑥) = ∫0 𝑓 𝜌𝑓 𝑣𝑥 (𝑦)𝑑𝑦 = − (1) (0 ≤ 𝑦 ≤ 𝛿𝑓 )
3𝜇𝑓
Energy equation can also be written for the film to determine variation in the film thickness along the
length, viz. 𝛿𝑓 (𝑥), and define Nux or hx :
𝑑 2 𝑇𝑓 𝜕𝑇
Energy equation: 𝑑𝑦 2
= 0 (𝑆𝑆, 𝜕𝑥
~0)
With suitable BCs, on integration,

83
 𝑦
𝑇𝑓 = 𝑇𝑤 + (𝑇𝑠𝑎𝑡 − 𝑇𝑤 ) (𝛿 ) − (2)
𝑓 (𝑥)

𝑑𝑇 (𝑇𝑠𝑎𝑡 − 𝑇𝑤 )
Therefore, 𝑞𝑤 = −𝑘𝑓 𝑑𝑦| = 𝑘𝑓 𝛿𝑓
= ℎ(𝑇𝑠𝑎𝑡 − 𝑇𝑤 )}: definition for ′ℎ′ (not “truly” a thermal
𝑦=0
boundary layer; here, we have a physical film of condensate)

Still, we have one more conservation equation (macroscopic mass & energy balances) to use:

𝑚̇(𝑥) ̇
𝛿𝑚
∆𝑥 𝑑𝑚̇ 𝑑𝑇
⇒ 𝜆 ( 𝑑𝑥 ) = +𝑘𝑓 𝑑𝑦|
𝑚̇(𝑥) + 𝛿𝑚̇ 𝑦=0

The scenario depicted above shows that increase in the (liquid) condensate-mass over ∆𝑥 is because of
the condensation of steam at the liquid-vapor interface. The energy balance equation equates the heat
released by condensation to the heat conducted across the film (condensate) under SS.

𝑑𝑚̇ 𝑔𝜌𝑓 (𝜌𝑓 −𝜌𝑣 )𝛿𝑓2 (𝑥) 𝑑𝛿𝑓 𝑘𝑓 (𝑇𝑠𝑎𝑡 − 𝑇𝑤 )

From (1) ( ) = ≡ − (3)
𝑑𝑥 𝜇𝑓 𝑑𝑥 𝜆𝛿𝑓

Combining (1) & (2) & integrating (3) with BC, 𝑥 = 0, 𝛿𝑓 = 0

1/4
4𝑘𝑓 𝜇𝑓 (𝑇𝑠𝑎𝑡 −𝑇𝑤 )𝑥 𝑘𝑓
𝛿𝑓 (𝑥) = [ ] (ℎ𝑓 (𝑥) ~ 𝛿𝑓 (𝑥)
)
𝑔𝜌𝑓 (𝜌𝑓 −𝜌𝑣 )𝜆

1/4
𝑔𝜌𝑓 (𝜌𝑓 −𝜌𝑣 )𝜆𝑘𝑓3
From (2) ℎ𝑓 (𝑥) is calculated as [ 4𝜇 ]
𝑓 (𝑇𝑠𝑎𝑡 −𝑇𝑤 )𝑥

Local heat transfer coefficient

1/4
ℎ𝑓 𝑥 𝑔𝜌𝑓 (𝜌𝑓 −𝜌𝑣 )𝜆𝑥 3
𝑁𝑢 (𝑥) = =[ ] (It can be shown that the average heat transfer
𝑘𝑓 4𝜇𝑓 𝑘𝑓 (𝑇𝑠𝑎𝑡 −𝑇𝑤 )
coefficient @𝑥 = 𝐿 is 4/3ℎ𝑓 )
⇒ Revert to condensation in the presence of air or a non-condensing gas (mass transfer effects):
0 𝑦
𝑎𝑖𝑟 − 𝑙𝑖𝑞𝑢𝑖𝑑 𝑖𝑛𝑡𝑒𝑟𝑓𝑎𝑐𝑒
𝑁𝐴 , 𝑞𝑠 T1 (𝑁𝐴 + 𝑁𝐵 )
ℎ𝐶

𝑇𝑤 𝑇𝑠 ℎ , 𝑘
𝑔 𝑔 𝑠𝑡𝑒𝑎𝑚 𝑎𝑖𝑟
𝑇𝑐 ℎ𝑓 𝑎𝑖𝑟 + 𝑠𝑡𝑒𝑎𝑚 𝑝𝐴1 , 𝑦𝐴1 , 𝑇1
ℎ𝑔 , 𝑘𝑔 ≡ 𝑎𝑟𝑖𝑠𝑒𝑠 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑜𝑓
x 𝑠𝑒𝑛𝑠𝑖𝑏𝑙𝑒 ℎ𝑒𝑎𝑡 𝑎𝑛𝑑
𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑜𝑛/𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑒𝑓𝑓𝑒𝑐𝑡𝑠
𝑐𝑜𝑜𝑙𝑎𝑛𝑡 𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑎𝑡𝑒 𝑓𝑖𝑙𝑚

84
Notes: 1. There is no ′𝑘𝑔 ′ if pure component (no air)
2. ′ℎ𝑔 ′ must be corrected for the effect of diffusion flux onto the condensate-layer.
- Ackerman correction factor.
See Browers in AIChE J (1995) 41 (7) for details. Briefly the traditional stagnation film model is modified
by considering “convective velocity”, “bulk transport” or “Stephan flow” or “Induced velocity”. Recall the
prototype diffusion-in-stagnant film example/equation:
𝐶𝐴
𝑁𝐴 = (𝑁𝐴 + 𝑁𝐵 ) + 𝐽𝐴
𝐶

𝐶𝐴 𝑑𝐶𝐴
= (𝑁𝐴 + 𝑁𝐵 ) 𝐶
− 𝐷𝐴𝐵 𝑑𝑦

𝐴 𝑁
𝑁𝐴 𝐶𝐷𝐴𝐵 − 𝑥𝐴2
𝑁𝐴 +𝑁𝐵
𝑁𝐴 = 𝑁𝐴 +𝑁𝐵
( 𝛿𝑓
) 𝑙𝑛 [ 𝑁 𝐴
] with NB = 0 (air is stagnant)
− 𝑥𝐴 1
𝑁𝐴 +𝑁𝐵

= kg(𝑝𝐴1 − 𝑝𝐴2 ) or kx(xA1 - xA2 ): definition for kg or kx

Heat transfer coefficient (modified)

𝑑𝑇
𝑞𝑆 = ℎ𝑔 (− )𝑧 + 𝑁𝐴 𝐶𝑝 (𝑇 − 𝑇𝑠 ) (within air-vapor film)
𝑑𝑦 𝑓

diffusion length/air--vapor film

mass flux towards condensate film

𝑇𝑠 𝑑𝑇 1 0
∫𝑇 =ℎ ∫𝑧 𝑑𝑦 (integrated over diffusion length; not condensate-film)
1 𝑞𝑠 −𝑁𝐴 𝐶𝑝 (𝑇−𝑇𝑠 ) 𝑧
𝑔 𝑓 𝑓
𝑁 𝐶𝑝
ℎ𝑔 ( 𝐴 )
ℎ𝑔
𝑞𝑠 = ( 𝑁𝐴 𝐶 𝑝 ) (𝑇1 − 𝑇𝑠 );
1−exp(− )
ℎ𝑔
𝝐 𝑵𝑨 𝑪𝒑
= ℎ𝑔 𝐶(𝑇1 − 𝑇𝑠 ) ⇒ 𝑪 ≡ 𝑨𝒄𝒌𝒆𝒓𝒎𝒂𝒏 𝒄𝒐𝒓𝒓𝒆𝒄𝒕𝒊𝒐𝒏 𝒇𝒂𝒄𝒕𝒐𝒓 = ( ); 𝝐=
𝟏−𝒆−𝝐 𝒉𝒈
′ℎ𝑔 ′ may be considered as heat transfer coefficient 𝑤⁄𝑜 mass transfer similar to Dittus-Boelter equation;
1
ℎ𝑔 = 0.023 𝑅𝑒 0.8 𝑃𝑟 3
In summary, the effect of mass transfer (diffusion) is to modify ′ℎ𝑔 ′ through an additional resistance:
𝝐
𝒉′𝒈 = 𝒉𝒈 (𝟏−𝒆−𝝐 ). Note that if NA = 0, C = 1, h’g = hg (same as that for a pure or single component)
⇒ The remaining model equations for the design of a heat exchanger (condenser) are straight forward:
See Mazzarotta & Sebastian in Canadian J ChE (1995) 73 for details.
𝑜
𝑦

𝑥
𝑇𝑠 𝑇1 (𝑜𝑟 𝑇𝑔 )
𝑐𝑜𝑜𝑙𝑎𝑛𝑡 ℎ
𝑐 (𝑠𝑡𝑒𝑎𝑚 + 𝑎𝑖𝑟)
𝑇𝑤
𝑝𝑔
𝑇𝑐 ℎ𝑓 𝑁𝐴
ℎ𝑔 ′ 85
 (1) Air (gas) phase heat balance:
𝑑𝑇𝑔 𝜖 𝒄𝒂𝒍 𝜖
𝐺𝐶𝑝 = −ℎ𝑔 (𝑇𝑔 − 𝑇𝑠 ) ( ) (hg’ = ℎ𝑔 )
𝑑𝑥 1−𝑒 −𝜖 𝒔 1−𝑒 −𝜖

@ 𝑥 = 0 𝑇𝑔 = 𝑇𝑔,𝑖𝑛𝑙𝑒𝑡

(2) 𝑴𝒂𝒔𝒔 𝒕𝒓𝒂𝒏𝒔𝒇𝒆𝒓: 𝑛𝐴 = −𝑘𝑔,𝐴 (𝑝𝑔,𝐴 − 𝑝𝑠,𝐴 ); 𝑝𝑠,𝐴 = 𝑓(𝑇𝑠 ) (𝐴𝑛𝑡𝑜𝑖𝑛′ 𝑠 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛)

𝑛𝐴
𝑝𝑔,𝐴 = (𝑛 ) 𝑝𝑇𝑜𝑡𝑎𝑙
𝐴 +𝑛𝐵

𝑑𝑛𝐴
= −𝑘𝑔,𝐴 (𝑝𝑔,𝐴 − 𝑝𝑠,𝐴 )
𝑑𝑥
𝒎𝒐𝒍𝒆𝒔
@ 𝑥 = 0 𝑛𝐴 = 𝑛𝐴,𝑖𝑛𝑙𝑒𝑡 = 𝐺𝑖𝑛𝑙𝑒𝑡 × 𝑦𝐴,𝑖𝑛𝑙𝑒𝑡 ( )
𝒔

𝑁𝑜𝑡𝑒: 𝒏𝑩 = 𝒄𝒐𝒏𝒔𝒕 = 𝐺𝑖𝑛𝑙𝑒𝑡 × (1 − 𝑦𝐴,𝑖𝑛𝑙𝑒𝑡 )

𝑑𝑛𝐵
throughout the height 𝑜𝑟 𝑑𝑥
=0

(3) Heat Balance in coolant:

𝑑𝑇𝑐 𝑑𝑇𝑔 𝑑𝑛𝐴

𝑞𝑤 = 𝑤𝑐 𝑐𝑝 𝑑𝑥
= 𝐺𝐶𝑝 𝑑𝑥
+ 𝜆𝐴 𝑑𝑥
(total heat transferred across the film/wall)

sensible heat latent heat

𝜖 𝑑𝑛𝐴
= hg
1−𝑒 −𝜖
(𝑇𝑔 − 𝑇𝑠 ) + 𝜆𝐴 𝑑𝑥
(𝑎𝑠𝑠𝑢𝑚𝑒 𝑢𝑛𝑖𝑡 𝑤𝑖𝑑𝑡ℎ 𝑜𝑓 𝑝𝑙𝑎𝑡𝑒)

@𝑥 = 0 𝑇𝑐 = 𝑇𝑐,𝑖𝑛𝑙𝑒𝑡

There are three variables, namely, 𝑇𝑔 , 𝑇𝑐 and 𝑛𝐴 , and as many number of equations (ODEs) to solve for
the condenser length required for 𝑛𝐴 to decrease from the inlet value
(nA,inlet ) to some prescribed value at the bottom of the condenser. See Gupta and Verma (2002)
Chem Eng Sci 57 (14), 2679-2696, for detailed calculations.

Correlations:
1
0.8 3
a. hg = 0.023Re Pr ∶ gas − side overall ′hoc ′
1
can also be
b. hc = 0.36Re0.55 Pr 3
∶ coolant side defined as
2 1 1 1 1
kg pBM MA = h′ + h + h
c. G
Sc = 0.023 × Re−0.17
3
hoc g c f
}

86
 Lecture #18
Introduction to the 2nd part of the course (Advanced topics??)
(High Reynolds number-flow, Boundary layer theory, Turbulence, Flow at large Peclet number)

So far, we studied a total of 12 prototype examples: 3 on momentum transport, namely, Stoke’s

first problem, start-up flow, and flow in a permeable tube or bundles of hollow fibers (lubrication
approximation); 4 on heat transport, namely, Nusselt problem, Graetz-Nusselt problem,
Brinkman problem, and free convection in confined flow; and 4 on mass transport, namely,
diffusion in a stagnant fluid (film theory), diffusion in a flow past a vertical plate (surface renewal
theory), Talyor-Aris dispersion in a laminar tubular flow, and diffusion-strained reaction. The last
example was related to the simultaneous momentum, heat and mass transport; in particular we
addressed condensation of steam over a vertical plate, from a mixture of air and steam. If you
objectively re-look at these examples, you can say that we have indirectly studied 3 additional
examples on mass transport. How? By making an analogy with heat transport, each of the 4
prototype examples can have an analogous situation in mass transport. Thus, temperature
gradients at the walls can have an analogous concentration gradients at the reactive walls, viz.
heterogeneous reaction, and the homogenous source of heat generation (viscous heating) can
have an analogous homogeneous chemical reactions in fluid flow. Similarly, Grashoff number
based on temperature difference can also be defined based on concentration difference,
implying there can be free convection in a non-uniform concentration field under isothermal
condition.

Here, the main point to be considered is that, apart from the common assumptions of an
incompressible fluid, NF, and laminar flow (low Reynolds number), all examples had
unidirectional or approximately unidirectional flow. It also implies that these problems were
simple to solve, and basic mathematical methods or techniques (Combination of variables
and/or Separation of variables) sufficed to exactly solve the governing conservation equations.
Another common feature of these examples was that there was no reference to “boundary layer”
or “boundary layer theory”, which is rather a common feature of most of fluid flows in chemical
engineering applications. Also, note that we did not use heat transfer coefficient, or mass transfer
coefficient, or drag coefficient, or friction factor, nor we used Nusselt number or Sherwood
number to calculate heat/mass transfer coefficients. It was not necessary. Why not? Because the
velocity, temperature, or concentration profiles were all analytically solved from the
conservation equations. Once these profiles are solved, one can calculate the respective
gradients on a solid surface, and then calculate momentum flux or drag or pressure drop, as well
as heat conduction and diffusion flux. Of course, once the flux is calculated, heat or mass transfer
coefficient can be defined or calculated per say. Similarly, once drag or pressure drop is
determined, one can define/calculate drag coefficient or friction factor.

Situation is, however, completely different at high Reynolds numbers or flow under turbulent
conditions. Here, mathematical complexity is significantly high, and the problems are not easy to

87
 solve. In other words, velocity, temperature, and concentration fields are not easy to resolve,
and the respective wall gradients or flux cannot be analytically determined or calculated.
Therefore, there are transport models and theories to solve/address such problems. From
engineering calculation point of view, drag coefficients, friction factor, mass and heat transfer
coefficients must be determined empirically or experimentally, of course with assistance from
theory. In some cases, special mathematical techniques are required to solve the conservation
equations, even if the flow is at low Reynolds numbers (eg. creeping flow). In these contexts, we
start new topics (Advanced?) in this course.

Streamlines and Stream functions (incompressible fluid)

(laminar flow can be characterized by streamlines)

⃗⃗⃗⃗⃗ given by 𝑑𝑠
(1) Define streamlines direction 𝒅𝒔 ⃗⃗⃗⃗⃗ × 𝑉
⃗⃗ = 0 (i.e. parallel to velocity vector, or velocity
vector is tangent to streamline at all points in the flow field)

𝑖 𝑗 𝑘
If 2-dimensions, |𝑑𝑥𝑠 𝑑𝑦𝑠 ⃗⃗ )
𝑑𝑧𝑠 | = 0 ⇒ (𝑣𝑦 𝑑𝑥𝑠 − 𝑣𝑥 𝑑𝑦𝑠 )𝒌 = 0 (⊥𝑟 𝑡𝑜 𝑑𝑠⃗ & 𝑉
𝑣𝑥 𝑣𝑦 𝑣𝑧

𝑑𝑦𝑠 𝑣𝑦
𝑜𝑟, = ;
𝑑𝑥𝑠 𝑣𝑥

(note that particle path and streamlines are identical under SS; Streak lines ⇒ dye continuously put in &
tracing every lines. Particle lines ⇒ one dye)

(2) Stream function 𝝍(𝒙, 𝒚) defined as

𝜕𝜓 𝜕𝜓
𝑑𝜓 = | 𝑑𝑥
𝜕𝑥 𝑦
+ 𝜕𝑦 | 𝑑𝑦
𝑥
𝜕𝜓 𝜕𝜓
with |
𝜕𝑥 𝑦
= −𝑣𝑦 & |
𝜕𝑦 𝑥
= 𝑣𝑥 − (1)

⃗⃗ = 0 (𝑐ℎ𝑒𝑐𝑘).
Then, 𝑑𝜓 = −𝑣𝑦 𝑑𝑥 + 𝑣𝑥 𝑑𝑦, and when equation (1) holds good, ∇. 𝑉

Alternatively, for an incompressible fluid-flow, there exists ′𝜓(𝑥, 𝑦)′ with the constraints
𝜕𝜓 𝜕𝜓
𝑣𝑥 = |
𝜕𝑦 𝑥
𝑎𝑛𝑑 − 𝑣𝑦 = |
𝜕𝑥 𝑦
.

(3) Streamlines are given by 𝒅𝝍 = 𝟎 𝒐𝒓 𝝍(𝒙, 𝒚) = 𝒄𝒐𝒏𝒔𝒕𝒂𝒏𝒕, 𝒄 (from 1 and 2)

𝑑𝑦 𝑣𝑦
Or, = ⇒ 𝜓(𝑥, 𝑦) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 represents a streamline
𝑑𝑥 𝑣𝑥

Streamlines: 𝜓 = 𝑐1
𝑑𝑒𝑓𝑖𝑛𝑒𝑠 𝑎 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑜𝑓
𝜓 = 𝑐2
𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒𝑠 ℎ𝑎𝑣𝑖𝑛𝑔 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡
𝒏 𝜓 = 𝑐3
𝑣𝑎𝑙𝑢𝑒𝑠, 𝑐1 , 𝑐2 , 𝑐3 , 𝑒𝑡𝑐.
𝜓 𝑖𝑠 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑎𝑙𝑜𝑛𝑔 𝑎 𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒

88
 (4) M = mass flowrate between two streamlines ≡ Δ 𝜓 α (𝜓1 − 𝜓2 )
(c1, c2, c3 values are not arbitrary. They are scaled depending on "spread" or "distance" between the
streamlines. Numerically, their relative values are dependent on the mass flowrates between the
streamlines, or diameter of a stream tube)
Some physical significances of 𝜓:
(1) 𝑣𝑥 & 𝑣𝑦 = 0 at solid boundary because of no-slip condition, implying 𝜓(𝑥, 𝑦) = 0 represents a solid
boundary.

(2) Two streamlines cannot cross each other, otherwise there will be two velocities. Mass cannot transfer
across the streamlines.

(3) A class of flow fields can be solved using ′𝜓′ relatively easier rather than 𝑣𝑥 & 𝑣𝑦 . In other words, we
have ∇2 𝜓 = 0 in stead of ∇. 𝑉 ⃗⃗ = 0 for an incompressible 2D fluid flow. Similarly, the viscous ∇2 𝑉 term
𝜕4 𝜕4 𝜕4
in NS equation can be easily transformed to ∇4 𝜓 = (𝜕𝑥 4 + 2 𝜕𝑥 2 𝜕𝑦2 + 𝜕𝑦4 )𝜓. See BSL Table 4.2.1.

4) For creeping flow past a sphere (Re  0, viscous term >> inertial term) ∇4 𝜓 = 0….easy to solve 𝜓. Once
𝜓 is solved, determine 𝑣𝑟 & 𝑣Ɵ → ∇𝑉 (strain tensor) → τ and σ (stresses) at the solid surface → determine
forces along the flow − direction → integrate over the entire sphere to obtain the world famous Stokes
law (1851) = 6πµ𝒗∞ 𝑹. Re-solve Deen’s example 7.4.2 as a home work.

Inviscid fluid-flow and irrotational flow

Are they different? Can a flow of an inviscid fluid be rotational? Can a creeping flow (low Reynolds number
when viscose effect is dominant) around a sphere be irrotational? Does 𝜇 = 0 represent the same flow
field of an inviscid fluid as that of a fluid having 𝜔 ⃗⃗ = 0 ? See Deen’s book.
⃗⃗ = ∇ × 𝑉

Some definitions (and some occasional mix-up): Inviscid fluid: 𝜇 = 0

Ideal fluid: 𝜌 = 𝑐, 𝜇 = 0
⃗⃗ = 0
⃗⃗ (𝑣𝑜𝑟𝑡𝑖𝑐𝑖𝑡𝑦 𝑣𝑒𝑐𝑡𝑜𝑟) = ∇ × 𝑉
Irrotational flow: 𝜔
⃗⃗
Potential flow: 𝜌 = 𝑐, 𝜇 = 0 𝒂𝒏𝒅 ∇ × 𝑉 = 0

⇒ Let us start with the flow of an ideal fluid: (𝝆 = 𝒄, 𝝁 = 𝟎 𝒐𝒓 𝑹𝒆 → ∞)

⃗⃗
𝐷𝑉 𝜕𝑉⃗⃗
NS equation: 𝜌 𝐷𝑡 = 𝜌 ( 𝜕𝑡 + 𝑉 ⃗⃗ . ∇𝑉⃗⃗ ) = −∇𝒫 = −∇𝑃 + 𝜌𝑔⃗ (neglecting the viscous term; better known
as Euler equation)
𝜌𝑉 ⃗⃗ . ∇𝑉⃗⃗ = −∇𝒫 (𝑢𝑛𝑑𝑒𝑟 𝑆𝑆) − (1)
𝑉2
⃗⃗ . ∇𝑉
𝑉 ⃗⃗ = ∇ ( ) − 𝑉
⃗⃗ × (∇ × 𝑉 ⃗⃗ ) : vector algebra − (2)
2

⃗⃗ on both sides:
Take dot product with 𝑉
2 2
⃗⃗ . (𝑉
𝑉 ⃗⃗ . ∇𝑉⃗⃗ ) = 𝑉 ⃗⃗ . ∇ (𝑉 ) − 𝑉 ⃗⃗ . [𝑉
⃗⃗ × (∇ × 𝑉 ⃗⃗ . ∇ (𝑉 )
⃗⃗ )] = 𝑉
2 2
(Note 𝑉 ⃗⃗ . [𝑉
⃗⃗ × (∇ × 𝑉 ⃗⃗ )] = 0; 𝑢𝑠𝑒 𝐷𝑒𝑎𝑛 𝑇𝑎𝑏𝑙𝑒 𝐴 − 1 𝑡𝑜 𝑠ℎ𝑜𝑤 𝑡ℎ𝑖𝑠. The assumption of 𝛁 × ⃗𝑽⃗ = 𝟎 is
not required!)

89
 𝑉2 𝑉.∇𝒫 𝑃
⃗⃗ . ∇ ( ) = − (
Accordingly, 𝑉 ) = −𝑉. ∇ (𝜌 + 𝑔ℎ) ; here ℎ is the height above some horizontal
2 𝜌
reference plane
2
⃗⃗ . ∇ (𝑉 + 𝑃 + 𝑔ℎ) = 0
Or, 𝑉 2 𝜌

⃗⃗ is parallel to streamline ⃗⃗⃗⃗⃗

Recall that 𝑉 ⃗⃗ . ∇𝑏, where 𝑏 is some scalar quantity, represents
𝑑𝑠. Therefore, 𝑉
the rate of change of the scalar 𝑏 along a streamline, which is equivalent to writing

𝑉2 𝑃
Δ( + + 𝑔ℎ) = 0: 𝐵𝑒𝑟𝑛𝑜𝑢𝑙𝑙𝑖 ′ 𝑠 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛(𝑐𝑜𝑛𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝐾. 𝐸. +𝑃. 𝐸. +𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝐸𝑛𝑒𝑟𝑔𝑦)
2 𝜌

where, the term in the parenthesis is the scalar quantity, 𝑏, and ′Δ′ refers to any two points along the
same streamline of an ideal fluid-flow.

⇒ Let us now start with an irrotational flow 𝜔 ⃗⃗ = 0 of an ideal fluid 𝜌 = 𝑐, 𝜇 = 0:

⃗⃗ = ∇ × 𝑉

By taking the curl of each term in the NS equation, we get

⃗⃗⃗⃗
𝐷𝜔
=𝜔 ⃗⃗
⃗⃗. ∇𝑉 where, 𝜔 ⃗⃗
⃗⃗ = ∇ × 𝑉
𝐷𝑡

⃗⃗⃗⃗
𝐷𝜔
Therefore, if 𝜔
⃗⃗ = 0 anywhere in flow field, then = 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒. It means that if the vorticity is
𝐷𝑡
zero anywhere in a fluid (eg. uniform and unperturbed flow) which is frictionless or inviscid, that fluid
remains irrotational forever.

⃗⃗ = 0 to obtain
Now, revisit the previous equations (1) & (2) for 𝜇 = 0 and insert ∇ × 𝑉

𝑉2 𝑃
∇ ( 2 + 𝜌 + 𝑔ℎ) = 0

Thus, the same Bernoulli’s equation is obtained where the sum in the parenthesis is constant
everywhere in the fluid (i.e. the same for all streamlines)! Thus, Bernoulli’s equation is valid for
potential flow (𝝆 = 𝒄, 𝝁 = 𝟎 & 𝛁 × 𝑽 ⃗⃗ = 𝟎) everywhere in the flow field, not necessarily along a
streamline as in the previous case of the flow of an ideal fluid!! Deen also mentions that this contrasts
the creeping flow past a sphere of a viscous fluid (Re → 0), where 𝛁 × 𝑽 ⃗⃗ = 𝟎 (irrotational) only at far
distances from the sphere!!

⃗⃗ = 0, there is a potential function 𝜙 such that

If ∇ × 𝑉
𝜕𝜙 𝜕𝜙
𝑣𝑥 = 𝑎𝑛𝑑 𝑣𝑦 = (𝑐ℎ𝑒𝑐𝑘) 𝑜𝑟 𝑉 ⃗⃗ = ∇𝜙 𝑎𝑛𝑑 ∇2 𝜙 = 0. Thus, we have,
𝜕𝑥 𝜕𝑦

⃗⃗ = 0 ⇒ ∇2 𝜓 = 0 (𝜌 = 𝑐)
∇. 𝑉
{ These represent kinematic solutions to a class of
∇×𝑉 ⃗⃗ = 0 ⇒ ∇2 𝜙 = 0 (𝑖𝑟𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑓𝑙𝑜𝑤)
flow field problems, without reference to NS equation.

𝜕𝜓 𝜕𝜙
Also, 𝑣𝑥 = = ; and
𝜕𝑦 𝜕𝑥

90
 𝜕𝜓 𝜕𝜙
𝑣𝑦 = − 𝜕𝑥 = 𝜕𝑦
You should recall that these are Cauchy Riemann conditions

Consider an analytic function of a complex variable z: 𝑤(𝑧) = 𝜙 + 𝑖𝜓, 𝑤ℎ𝑒𝑟𝑒 𝑧 = 𝑥 + 𝑖𝑦. It can be
shown that the function has real & imaginary parts that are each solution to Laplace equations

∇2 𝜙 = ∇2 𝜓 = 0

Therefore, any analytic function 𝑤(𝑧) represents potential function & stream function, and thus
⃗⃗ = 0, μ = 0 ).
velocity field of an ideal irrotational flow or simply potential flow (ρ = c, ∇ × V

Ex. 𝑤(𝑧) = 𝑖𝑧 2 = 𝑖(𝑥 + 𝑖𝑦)2 = −2𝑥𝑦 + 𝑖(𝑥 2 − 𝑦 2 ) (in Cartesian coordinates)

𝜙 𝜓

Verify ∇2 𝜙 = ∇2 𝜓 = 0, i.e. the real part is potential function and imaginary part is stream function. In
other words, any complex analytic function 𝑤(𝑧) = 𝜙 + 𝑖𝜓 will satisfy ∇2 𝜓 = ∇2 𝜙 = 0, and describe
some potential flow, although it may not be the one you want!

Ex. Very often it is convenient to work on polar coordinates (r-Ɵ):

𝑤(𝑧) = 𝐴𝑧 𝑛 = 𝐴𝑟 𝑛 𝑒 𝑖𝑛𝜃 = 𝐴𝑟 𝑛 (𝑐𝑜𝑠𝑛𝜃 + 𝑖𝑠𝑖𝑛𝑛𝜃)
𝜙 = 𝐴𝑟 𝑛 𝑐𝑜𝑠𝑛𝜃 ; 𝜓 = 𝐴𝑟 𝑛 𝑠𝑖𝑛 𝑛𝜃
𝜕 1 𝜕 1 𝜕2
Verify ∇2 𝜙 = ∇2 𝜓 = 0 (Note that the polar representation of ∇2 ≡ + 𝑟 𝜕𝑟 + 𝑟 2 𝜕𝜃2 )
𝜕𝑟 2

𝜕𝜙 𝜕𝜓
Define, complex velocity 𝐶 = 𝑣𝑥 − 𝑖𝑣𝑦 in Cartesian coordinates, where 𝑣𝑥 = ; and 𝑣𝑦 = −
𝜕𝑥 𝜕𝑥
𝜕𝜙 𝜕𝜓 𝜕𝑤 𝑑𝑤 𝜕𝑧 𝑑𝑤 𝜕𝑧
Therefore, C = 𝜕𝑥
+ 𝑖 𝜕𝑥 = 𝜕𝑥
= 𝑑𝑧 𝜕𝑥
= 𝑑𝑧
(note 𝜕𝑥 = 1)

In polar coordiante system 𝐶 = |𝑣|𝑒 −𝑖𝛼 , where |𝑣| = √𝑣𝑥2 + 𝑣𝑦2 , 𝛼 = ∠𝑣𝑥 & 𝑣𝑦

And two coordinate systems are related as: 𝑣x = 𝑣𝑟 𝑐𝑜𝑠𝜃 − 𝑣𝜃 𝑠𝑖𝑛𝜃

𝑣𝑦 = 𝑣𝑟 𝑠𝑖𝑛𝜃 + 𝑣𝜃 𝑐𝑜𝑠𝜃
𝐶 = [𝑣𝑟 (𝑟, 𝜃) − 𝑖𝑣𝜃 (𝑟, 𝜃)]𝑒 −𝑖𝜃
Ex: Let 𝑤(𝑧) = 𝑢𝑒 −𝑖𝛼 z

𝑑𝑤
𝐶= 𝑑𝑧
= 𝑢𝑒 −𝑖𝛼 = 𝑢(𝑐𝑜𝑠𝛼 − 𝑖 𝑠𝑖𝑛𝛼). Therefore, 𝑢𝑥 = 𝑢 𝑐𝑜𝑠𝛼, 𝑢𝑦 = 𝑢 𝑠𝑖𝑛𝛼 and the flow field is

𝑖𝑦

𝑥
𝛼

𝑑𝑤
Ex. Revert to 𝑤 = 𝐴𝑧 𝑛 , 𝐶 = = 𝑛𝐴𝑧 𝑛−1 = 𝑛𝐴𝑟 𝑛−1 𝑒 𝑖𝑛𝜃 𝑒 −𝑖𝜃 = 𝑛𝐴𝑟 𝑛−1 (𝑐𝑜𝑠𝑛𝜃 + 𝑖 𝑠𝑖𝑛𝑛𝜃)𝑒 −𝑖𝜃
𝑑𝑧
𝑣𝑟 = 𝑛𝐴𝑟 𝑛−1 𝑐𝑜𝑠𝑛𝜃, 𝑣𝜃 = −𝑛𝐴𝑟 𝑛−1 𝑠𝑖𝑛𝑛𝜃; Also 𝜙 = 𝐴𝑟 𝑛 𝑐𝑜𝑠𝑛𝜃, 𝜓 = 𝐴𝑟 𝑛 𝑠𝑖𝑛𝑛𝜃

91
 𝑘𝜋
If 𝜃 = 𝑛
, 𝑘 = 1,2,3 … ; 𝑣𝑟 = (−1)𝑘 𝑛𝐴𝑟 𝑛−1 , 𝑣𝜃 = 0

Velocity fields are,

1
𝑛=2∶ 𝑛: 3:

1
2
< 𝑛 < 1: 𝑛: 2:

Source/Sink: 𝑤(𝑧) = 𝑘 𝑙𝑛𝑧 = 𝑘 𝑙𝑛 (𝑟𝑒𝑖𝜃 ) = 𝑘 𝑙𝑛𝑟 + 𝑖𝑘𝜃

𝜙 = 𝑘 𝑙𝑛𝑟, 𝜓 = 𝑘𝜃
𝐶=
𝑘 𝑘
= 𝑟 𝑒 −𝑖𝜃 𝑐𝑜𝑛𝑠𝑡. 𝑝𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 (𝑣𝑟 = 𝑘⁄𝑟 , 𝑣𝜃 = 0)
𝑧 𝑘 = −𝑣𝑒.
𝑣𝑟
𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒
2𝜋

2𝜋 2𝜋 𝑚
𝑚 = ∫0 𝑣𝑟 𝑟𝑑𝜃 = ∫0 𝑘𝑑𝜃 = 2𝜋𝑘 ⇒ 𝑘 = 2𝜋;

𝑚 𝑚
𝑤 = 2𝜋 𝑙𝑛𝑧 → 𝑠𝑜𝑢𝑟𝑐𝑒 𝑜𝑟 − 2𝜋 𝑙𝑛𝑧 ⇒ 𝑠𝑖𝑛𝑘

𝑖Γ
Vortex 𝒘(𝒛) = − 𝑙𝑛𝑧
2𝜋

Check and plot:

2𝜋
Γ = ∮ 𝑈. 𝑑𝑙 = ∫0 𝑣𝜃 𝑟𝑑𝜃
𝑣𝑟 = 0
𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒𝑠 𝑣𝜃 = Γ⁄2𝜋𝑟

𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑝𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙

Doublet (Source + Sink)

superimposition yields

𝜓𝑠

92
 𝜇 𝜇 𝑥 − 𝑖𝑦 𝜇𝑦
𝑤= (𝑐ℎ𝑒𝑐𝑘 𝑏𝑦 𝑎𝑑𝑑𝑖𝑛𝑔/𝑠𝑢𝑝𝑒𝑟𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛) = =( 2 2
)𝜇 ⇒ 𝜓 = − 2
𝑧 𝑥 + 𝑖𝑦 𝑥 +𝑦 𝑥 + 𝑦2

𝜇𝑦 𝜇 2 𝜇 2
⇒ 𝑥2 + 𝑦2 + = 0 ⇒ 𝑥 2 + (𝑦 + ) =( )
𝜓 2𝜓1 2𝜓1

𝜇 𝜇
Centre: (0, − 2𝜓 ) & 𝑟 = 2𝜓
1 1

𝑂𝑛 𝑡ℎ𝑒 𝑜𝑡ℎ𝑒𝑟 𝑝𝑙𝑎𝑛𝑒

𝜇
(0, + )
𝜓1 2𝜓1
𝜓3 𝜓2
𝑟

𝜇 𝜇 𝜇
𝐶(𝑧) = − 𝑧 2 = − 𝑅2 𝑒 −2𝑖𝜃 = − 𝑅2 (𝑐𝑜𝑠𝜃 − 𝑖𝑠𝑖𝑛𝜃) 𝑒 −𝑖𝜃
↑ ↑
𝑈𝑟 𝑈𝜃

93
 Lecture 19
Boundary layer theory
Before we introduce BLT, let us continue with one more example of potential flow:

Flow around cylinder without circulation:

𝜇 𝜇
𝑤(𝑧) = 𝑈𝑧 + = 𝑈𝑟𝑒 𝑖𝜃 + 𝑒 −𝑖𝜃
𝑧 𝑟
𝑠𝑢𝑝𝑒𝑟𝑖𝑚𝑝𝑜𝑠𝑒𝑑 𝜇 𝜇
≡ + 𝑓𝑙𝑜𝑤 𝑓𝑖𝑒𝑙𝑑𝑠 = (𝑈𝑟 + 𝑟 ) 𝑐𝑜𝑠𝜃 + 𝑖 (𝑈𝑟 − 𝑟 ) 𝑠𝑖𝑛𝜃
𝜇
𝜓 = (𝑈𝑟 − ) 𝑠𝑖𝑛𝜃; 𝑙𝑒𝑡 𝜇 = 𝑈𝑎2
𝑟
𝜓 = (𝑈𝑎 − 𝑈𝑎)𝑠𝑖𝑛𝜃 = 0 → cylinder (radius a)

𝜓=0

𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛
𝑝𝑜𝑖𝑛𝑡𝑠 𝑎

𝑎2 𝑎2
From stream function, calculate 𝑣𝑥 = 𝑣∞ (1 − 𝑐𝑜𝑠2𝜃), 𝑣𝑦 = −𝑣∞ ( 𝑠𝑖𝑛2𝜃) to show that @ r = a
𝑟2 𝑟2

𝑣 2 = 4𝑣∞ 2 𝑠𝑖𝑛2Ɵ. Where is the no-slip condition at the solid surface (except at the two stagnation
points)?. Apply Bernoulli equation between the points far away from the cylinder and on the cylinder to
1
show that 𝑝 − 𝑝∞ = 2 𝜌𝑣∞ 2 (1 − 4𝑠𝑖𝑛2Ɵ). See example 4.3.1 of BSL (Cartesian coordinate system) or 8.3-
1 of Dean (polar coordinate system). The pressure distribution is symmetric about x-axis (flow direction),
which means that there is no form drag on the cylinder! It is called d’Alembert’s paradox, with the
potential theory predicting zero drag on the cylinder.

So far we can solve

(1) Laminar viscous flow ⇒ low/moderate Reynolds nos including Re → 0 (creeping flow)
(2) Potential flow ⇒ high Reynolds nos.
↓

94
But, this gives rise to d’ Alembert’s paradox (predicts that drag on a plate in a flow field is 0, because µ
= 0 and/or ∇ × 𝑉 ⃗⃗ = 0) )

Potential flow is important because there exists a theorem which states that if the vorticity is zero
anywhere in a fluid which is frictionless then that fluid remains irrotational forever.
⃗⃗⃗⃗
𝐷𝜔
For 𝜌 = 𝑐 ⇒ = (𝜔 ⃗⃗
⃗⃗. ∇)𝑉 ⃗⃗ = ∇ × 𝑉
(𝜔 ⃗⃗ )
𝐷𝑡

⃗⃗⃗⃗
𝐷𝜔
If 𝜔
⃗⃗ = 0, 𝑡ℎ𝑒𝑛 𝐷𝑡
= 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒.

Supercooled 𝑙𝑖𝑞𝑢𝑖𝑑 𝐻𝑒 has zero 𝜇 ⇒ Helmholtz theorem suggests potential flow should be very common
in nature for low viscosity fluids. Reality is different: even low 𝝁-fluids have friction which is important.
You need a transfer pump to deliver or supply liquid He in a pipe.

𝑅𝑒 ↑ 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑡𝑒𝑟𝑚 ↓ 𝑖𝑟𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑓𝑙𝑜𝑤 ↑ 𝑑𝑟𝑎𝑔 ↑

𝑝𝑎𝑟𝑎𝑑𝑜𝑥 expt measurements/observations

Revisit NS equation: (𝑅𝑒 → ∞)

⃗⃗
𝜕𝑉
⃗⃗ . ∇)𝑉
𝜌 𝜕𝑡 + 𝜌(𝑉 ⃗⃗ = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗ + 𝜌𝑔⃗

⃗⃗ 𝑑𝑟𝑜𝑝𝑠 𝑜𝑢𝑡,
𝜇 → 0, 𝜇∇2 𝑉
(Highest order derivative was dropped out)
𝐷𝑉 ∗ 1 1 ∗
Non-dimensionalize: 𝐷𝑡 ∗
= −∇∗ 𝑝∗ + 𝑅𝑒 ∇∗ 2 𝑉 ∗ + 𝑔
𝐹𝑟 𝑟

term drops out as 𝑅𝑒 → ∞

One BC is removed! 𝑽𝒛 = 𝟎 at surface! ⇒ 𝑰𝒕 reduces the order of differential equation; removes the
cylinder (solid surface) from consideration!

− This is not feasible. All it implies is that non-dimensionalization procedure was not correct.
𝑥
→ 𝑥 ∗ = 𝐿 ⇒ 𝐿 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑧𝑒𝑠 ′𝑥′
→ 𝑦
𝑉∞ 𝑦∗ ≠ ⇒′ 𝑳′ 𝒄𝒂𝒏𝒏𝒐𝒕 𝒄𝒉𝒂𝒓𝒂𝒄𝒕𝒆𝒓𝒊𝒛𝒆 ′𝒚′,
→ 𝐿
→ 𝑖𝑓 𝑅𝑒 ↑
← 𝐿 →
One should realize that characteristic distance in 𝑦 direction is much-much smaller than 𝐿:

𝑦
𝑈∞ 𝑦
⇒ 𝛿 ≪ 𝐿 ⇒ 𝑦∗ ≠ 𝐿
𝛿 𝑦 𝑥
𝛿(𝑥) 𝑜𝑟 𝐿
≪ 1; 𝑦 ∗ = 𝛿
∗
& 𝑥 =𝐿
𝑂 𝐿 𝑥

95
 𝑘𝜋
If 𝜃 = 𝑛
, 𝑘 = 1,2,3 … ; 𝑣𝑟 = (−1)𝑘 𝑛𝐴𝑟 𝑛−1 , 𝑣𝜃 = 0

Velocity fields are,

1
𝑛=2∶ 𝑛: 3:

1
2
< 𝑛 < 1: 𝑛: 2:

Source/Sink: 𝑤(𝑧) = 𝑘 𝑙𝑛𝑧 = 𝑘 𝑙𝑛 (𝑟𝑒𝑖𝜃 ) = 𝑘 𝑙𝑛𝑟 + 𝑖𝑘𝜃

𝜙 = 𝑘 𝑙𝑛𝑟, 𝜓 = 𝑘𝜃
𝐶=
𝑘 𝑘
= 𝑟 𝑒 −𝑖𝜃 𝑐𝑜𝑛𝑠𝑡. 𝑝𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙 (𝑣𝑟 = 𝑘⁄𝑟 , 𝑣𝜃 = 0)
𝑧 𝑘 = −𝑣𝑒.
𝑣𝑟
𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒
2𝜋

2𝜋 2𝜋 𝑚
𝑚 = ∫0 𝑣𝑟 𝑟𝑑𝜃 = ∫0 𝑘𝑑𝜃 = 2𝜋𝑘 ⇒ 𝑘 = 2𝜋;

𝑚 𝑚
𝑤 = 2𝜋 𝑙𝑛𝑧 → 𝑠𝑜𝑢𝑟𝑐𝑒 𝑜𝑟 − 2𝜋 𝑙𝑛𝑧 ⇒ 𝑠𝑖𝑛𝑘

𝑖Γ
Vortex 𝒘(𝒛) = − 𝑙𝑛𝑧
2𝜋

Check and plot:

2𝜋
Γ = ∮ 𝑈. 𝑑𝑙 = ∫0 𝑣𝜃 𝑟𝑑𝜃
𝑣𝑟 = 0
𝑠𝑡𝑟𝑒𝑎𝑚𝑙𝑖𝑛𝑒𝑠 𝑣𝜃 = Γ⁄2𝜋𝑟

𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑝𝑜𝑡𝑒𝑛𝑡𝑖𝑎𝑙

Doublet (Source + Sink)

superimposition yields

𝜓𝑠

92
 𝜇 𝜇 𝑥 − 𝑖𝑦 𝜇𝑦
𝑤= (𝑐ℎ𝑒𝑐𝑘 𝑏𝑦 𝑎𝑑𝑑𝑖𝑛𝑔/𝑠𝑢𝑝𝑒𝑟𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛) = =( 2 2
)𝜇 ⇒ 𝜓 = − 2
𝑧 𝑥 + 𝑖𝑦 𝑥 +𝑦 𝑥 + 𝑦2

𝜇𝑦 𝜇 2 𝜇 2
⇒ 𝑥2 + 𝑦2 + = 0 ⇒ 𝑥 2 + (𝑦 + ) =( )
𝜓 2𝜓1 2𝜓1

𝜇 𝜇
Centre: (0, − 2𝜓 ) & 𝑟 = 2𝜓
1 1

𝑂𝑛 𝑡ℎ𝑒 𝑜𝑡ℎ𝑒𝑟 𝑝𝑙𝑎𝑛𝑒

𝜇
(0, + )
𝜓1 2𝜓1
𝜓3 𝜓2
𝑟

𝜇 𝜇 𝜇
𝐶(𝑧) = − 𝑧 2 = − 𝑅2 𝑒 −2𝑖𝜃 = − 𝑅2 (𝑐𝑜𝑠𝜃 − 𝑖𝑠𝑖𝑛𝜃) 𝑒 −𝑖𝜃
↑ ↑
𝑈𝑟 𝑈𝜃

93
 Lecture 19
Boundary layer theory
Before we introduce BLT, let us continue with one more example of potential flow:

Flow around cylinder without circulation:

𝜇 𝜇
𝑤(𝑧) = 𝑈𝑧 + = 𝑈𝑟𝑒 𝑖𝜃 + 𝑒 −𝑖𝜃
𝑧 𝑟
𝑠𝑢𝑝𝑒𝑟𝑖𝑚𝑝𝑜𝑠𝑒𝑑 𝜇 𝜇
≡ + 𝑓𝑙𝑜𝑤 𝑓𝑖𝑒𝑙𝑑𝑠 = (𝑈𝑟 + 𝑟 ) 𝑐𝑜𝑠𝜃 + 𝑖 (𝑈𝑟 − 𝑟 ) 𝑠𝑖𝑛𝜃
𝜇
𝜓 = (𝑈𝑟 − ) 𝑠𝑖𝑛𝜃; 𝑙𝑒𝑡 𝜇 = 𝑈𝑎2
𝑟
𝜓 = (𝑈𝑎 − 𝑈𝑎)𝑠𝑖𝑛𝜃 = 0 → cylinder (radius a)

𝜓=0

𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛
𝑝𝑜𝑖𝑛𝑡𝑠 𝑎

𝑎2 𝑎2
From stream function, calculate 𝑣𝑥 = 𝑣∞ (1 − 𝑐𝑜𝑠2𝜃), 𝑣𝑦 = −𝑣∞ ( 𝑠𝑖𝑛2𝜃) to show that @ r = a
𝑟2 𝑟2

𝑣 2 = 4𝑣∞ 2 𝑠𝑖𝑛2Ɵ. Where is the no-slip condition at the solid surface (except at the two stagnation
points)?. Apply Bernoulli equation between the points far away from the cylinder and on the cylinder to
1
show that 𝑝 − 𝑝∞ = 2 𝜌𝑣∞ 2 (1 − 4𝑠𝑖𝑛2Ɵ). See example 4.3.1 of BSL (Cartesian coordinate system) or 8.3-
1 of Dean (polar coordinate system). The pressure distribution is symmetric about x-axis (flow direction),
which means that there is no form drag on the cylinder! It is called d’Alembert’s paradox, with the
potential theory predicting zero drag on the cylinder.

So far we can solve

(1) Laminar viscous flow ⇒ low/moderate Reynolds nos including Re → 0 (creeping flow)
(2) Potential flow ⇒ high Reynolds nos.
↓

94
But, this gives rise to d’ Alembert’s paradox (predicts that drag on a plate in a flow field is 0, because µ
= 0 and/or ∇ × 𝑉 ⃗⃗ = 0) )

Potential flow is important because there exists a theorem which states that if the vorticity is zero
anywhere in a fluid which is frictionless then that fluid remains irrotational forever.
⃗⃗⃗⃗
𝐷𝜔
For 𝜌 = 𝑐 ⇒ = (𝜔 ⃗⃗
⃗⃗. ∇)𝑉 ⃗⃗ = ∇ × 𝑉
(𝜔 ⃗⃗ )
𝐷𝑡

⃗⃗⃗⃗
𝐷𝜔
If 𝜔
⃗⃗ = 0, 𝑡ℎ𝑒𝑛 𝐷𝑡
= 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒.

Supercooled 𝑙𝑖𝑞𝑢𝑖𝑑 𝐻𝑒 has zero 𝜇 ⇒ Helmholtz theorem suggests potential flow should be very common
in nature for low viscosity fluids. Reality is different: even low 𝝁-fluids have friction which is important.
You need a transfer pump to deliver or supply liquid He in a pipe.

𝑅𝑒 ↑ 𝑣𝑖𝑠𝑐𝑜𝑢𝑠 𝑡𝑒𝑟𝑚 ↓ 𝑖𝑟𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑓𝑙𝑜𝑤 ↑ 𝑑𝑟𝑎𝑔 ↑

𝑝𝑎𝑟𝑎𝑑𝑜𝑥 expt measurements/observations

Revisit NS equation: (𝑅𝑒 → ∞)

⃗⃗
𝜕𝑉
⃗⃗ . ∇)𝑉
𝜌 𝜕𝑡 + 𝜌(𝑉 ⃗⃗ = −∇𝑝 + 𝜇∇2 𝑉
⃗⃗ + 𝜌𝑔⃗

⃗⃗ 𝑑𝑟𝑜𝑝𝑠 𝑜𝑢𝑡,
𝜇 → 0, 𝜇∇2 𝑉
(Highest order derivative was dropped out)
𝐷𝑉 ∗ 1 1 ∗
Non-dimensionalize: 𝐷𝑡 ∗
= −∇∗ 𝑝∗ + 𝑅𝑒 ∇∗ 2 𝑉 ∗ + 𝑔
𝐹𝑟 𝑟

term drops out as 𝑅𝑒 → ∞

One BC is removed! 𝑽𝒛 = 𝟎 at surface! ⇒ 𝑰𝒕 reduces the order of differential equation; removes the
cylinder (solid surface) from consideration!

− This is not feasible. All it implies is that non-dimensionalization procedure was not correct.
𝑥
→ 𝑥 ∗ = 𝐿 ⇒ 𝐿 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑧𝑒𝑠 ′𝑥′
→ 𝑦
𝑉∞ 𝑦∗ ≠ ⇒′ 𝑳′ 𝒄𝒂𝒏𝒏𝒐𝒕 𝒄𝒉𝒂𝒓𝒂𝒄𝒕𝒆𝒓𝒊𝒛𝒆 ′𝒚′,
→ 𝐿
→ 𝑖𝑓 𝑅𝑒 ↑
← 𝐿 →
One should realize that characteristic distance in 𝑦 direction is much-much smaller than 𝐿:

𝑦
𝑈∞ 𝑦
⇒ 𝛿 ≪ 𝐿 ⇒ 𝑦∗ ≠ 𝐿
𝛿 𝑦 𝑥
𝛿(𝑥) 𝑜𝑟 𝐿
≪ 1; 𝑦 ∗ = 𝛿
∗
& 𝑥 =𝐿
𝑂 𝐿 𝑥

95
Therefore, 𝑓 ′ (𝜂) = 1, 𝑔(𝑥) = √𝑣∞ 𝜈𝑥 (𝑡𝑎𝑘𝑒 𝑖𝑡 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦)

𝑣𝑥 = 𝜓𝑦 = 𝑔𝑓′⁄𝛿 = 𝑣∞ 𝑓′

𝑔𝑓 ′ 𝑦 1 𝜈𝑣∞
−𝑣𝑦 = 𝜓𝑥 = 𝑔′ 𝑓 − = −2 √ (𝜂𝑓 ′ − 𝑓)
2𝑥𝛿 𝑥

𝑔′ 𝑓′ 𝑔𝑓′′𝑦 𝑔𝑓′ 𝑔𝑓′′ 𝑔𝑓′′′

𝜓𝑥𝑦 = − − 2𝑥𝛿 ; 𝜓𝑦𝑦 = ; 𝜓𝑦𝑦𝑦 = (check all these calculations)
𝛿 2𝑥𝛿 2 𝛿2 𝛿3

𝜕𝜓
BCs. on 𝜓: (1) 𝑦 = 0 𝑣𝑥 = 0 𝜕𝑦
=0

𝜕𝜓 𝜕𝜓
(2) 𝑣𝑦 = 0 =0 (3) 𝑦 → ∞ 𝑣𝑥 = 𝑣∞ = 𝑣∞
𝜕𝑥 𝜕𝑦

BCs: on 𝑓: (1) 𝜂 = 0 (𝑦 = 0) 𝑓′ = 0

(2) 𝜂 = 0 (𝑦 = 0) 𝑓=0 (3) 𝜂 → ∞ (𝑦 → ∞) 𝑓 ′ = 1

Plugging,

𝑔𝑓′ 𝑔′𝑓′ 𝑔𝑓′′𝑦 𝑔𝑓′ 𝑔𝑓′𝑦 𝑔𝑓′′ 𝑔𝑓′′′

( − − 2𝑥𝛿) − (𝑔′ 𝑓 − ) =𝜈
𝛿 𝛿 2𝑥𝛿 2 2𝑥𝛿 𝛿2 𝛿3

𝛿 𝜇𝑥
Recall 𝑔 = 𝑣∞ 𝛿 𝑎𝑛𝑑 𝑔′ = 𝑣∞ 2𝑥
(𝑅𝑒𝑐𝑎𝑙𝑙 𝛿~ √𝜌𝑣 )
∞

𝑓𝑓 ′′ + 2𝑓 ′′′ = 0 BCs. (1-2) 𝜂 = 0 𝑓 ′ = 𝑓 = 0 and (3) 𝜂 → ∞ 𝑓 ′ = 1

𝑑𝑓′′ 1
= − 𝑓𝑓′′
𝑑𝜂 2

𝑑𝑓′′ 1
∫ = − ∫ 𝑓𝑑𝜂
𝑓′′ 2

𝜂 1 𝜂
1 𝜂 ∞ 1 𝜂 𝜂 ∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂
𝑓 ′ ′ = exp (− 2 ∫0 𝑓𝑑𝜂)⁄∫0 𝑒𝑥𝑝 (− 2 ∫0 𝑓𝑑𝜂) ⇒ 𝑓′ = ∫0 𝑓 ′′ 𝑑𝜂 = ∞ 1 𝜂
∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂)

𝜂 𝜂 1 𝜂
𝜂 ∫0 ∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂𝑑𝜂
𝑓 = ∫0 𝑓 ′ 𝑑𝜂 = ∞ 1 𝜂
∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂

Recall 𝑣𝑥 = 𝑣∞ 𝑓 ′ (𝜂)
1 𝜈𝑣∞
𝑣𝑦 = 2 √ 𝑥
(𝜂𝑓 ′ − 𝑓)

∞ 𝑣
𝜂 = 𝑦√ 𝜈𝑥
}

Blasius Solution (1908 - 10):

For small values of 𝜂 (𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒):

100
 𝐴2 𝐴3
𝑓(𝜂) = 𝐴𝑜 + 𝐴1 𝜂 + 𝜂2 + 𝜂 3 + ⋯ (𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑐𝑒 𝑓𝑜𝑟 𝜂 < 1)
2! 3!

Plug in ODE:

1 𝑛 𝛼 𝑛+1 𝐶𝑛
𝑓 = ∑∞
𝑛=0 (− ) 𝜂 3𝑛+2
2 3𝑛+2!

where, 𝐶𝑜 = 1, 𝐶1 = 1, 𝐶2 = 11, 𝐶3 = 375, 𝐶4 = 27897, 𝐶5 = 3,817,137

What is 𝛼? ⇒ 𝛼 = 𝑓 ′′ (0); Blasius ‘patched’ a series of small 𝜂 to an asymptotic series for large 𝜂. Best
procedure is to use numerical integration to match 𝜂 = ∞ at BC. This yields 𝛼 = 0.332. 𝑓 ′′ (𝑥) is an
important variable to calculate. Check: it is nothing but the velocity gradient at the wall! Therefore, wall
drag can be calculated from 𝑓 ′′ (𝑥), shown later.

Once 𝑓(𝜂) is known & 𝑔(𝑥) = 𝑣∞ 𝛿(𝑥)

𝜓 = 𝑔(𝑥)𝑓(𝜂) = 𝑣∞ 𝛿(𝑥)𝑓(𝜂)
𝜕𝜓 𝜕𝜓
Therefore, 𝑣𝑥 = 𝜕𝑦
, 𝑣𝑦 = − 𝜕𝑥 (𝑠𝑜𝑙 𝑛 )

For small 𝜂 (close to plate)

1.33 𝑣∞ 𝑣∞ 𝑦 𝑦≪𝛿
𝑣𝑥 = 𝑣∞ √ 𝑦≈ ( )
4 𝜈𝑥 𝛿(𝑥) 𝜂≪1

𝜕𝑣 𝑣∞
( 𝜕𝑦𝑥 | =
𝛿(𝑥)
)
𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒

3/2
1.33𝑣∞ 𝜈𝑦 2
𝑣𝑦 = 𝑦2 ≈ (Note 𝑣𝑦 ≪ 𝑣𝑥 )
16𝜈1/2 𝑥 3/2 𝛿(𝑥)3

1.0 𝐵𝐿 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 𝑖𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑

0.99 𝑎𝑠 𝑡ℎ𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑤𝑎𝑙𝑙 𝑤ℎ𝑒𝑟𝑒
𝑣𝑥
𝑣𝑥 𝑣
= 𝑓 ′ (𝜂) = 0.99
= 𝑓’(𝜂) ∞
𝑣∞ 𝑇ℎ𝑖𝑠 𝑜𝑐𝑐𝑢𝑟𝑠 𝑎𝑡 𝜂 = 5
} 𝜈
𝑜𝑟 𝑦 = 5√𝑣 𝑥
5.0 ∞

𝑣∞
𝜂 = 𝑦√ 𝑦 1 𝛿(𝑥) 5
𝜈𝑥 = 5√ → =
𝑥 𝑅𝑒𝑥 𝑥 √𝑅𝑒𝑥

101
𝛿
𝐿
≪ 1 is the basic assumption of (Prandtl) Boundary layer Theory.

Consider flow around a cylinder:

𝛿≪𝑅 ≡ 𝑈∞ II

I
R

Flow field is divided in two regimes:

I. ⇒ Close to the surface: viscous term is important, no matter how high is 𝑅𝑒; inertial flow is also
important.
𝛿
𝐿
≪ 1 or characteristic distance in 𝑦 direction is much less than that in 𝑥 direction.

II. ⇒ Away from the surface: Potential flow; 𝑥, 𝑦 − characteristic lengths are ~ the same. If 𝑅𝑒 →
∞ then the viscous term → 0

NS equations in region I (See Schlichting chapter 3-5):

𝜕𝑉𝑥 𝜕𝑉 𝜕𝑉 𝜕𝑝 𝜕2 𝑉 𝜕2 𝑉
𝜌[ + 𝑉𝑥 𝑥 + 𝑉𝑦 𝑥 ] = − + 𝜇 [ 2𝑥 + 2𝑥 ]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦
𝜕𝑉𝑦 𝜕𝑉𝑦 𝜕𝑉𝑦 𝜕𝑝 𝜕2 𝑉𝑦 𝜕2 𝑉
𝜌 [ + 𝑉𝑥 + 𝑉𝑦 ] = − + 𝜇 [ 2 + 2𝑦 ]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑦

𝑉 𝑉𝑦 𝑥 𝒚
Note: 𝑉𝑥∗ = 𝑈𝑥 , 𝑉𝑦∗ = 𝑉𝑐
(𝑉 =? ), 𝑥 ∗ = , 𝒚∗ =
𝐿 𝜹
∞

𝜕𝑉𝑥 𝜕𝑉
Continuity: + 𝑦=0
𝜕𝑥 𝜕𝑦
𝑈∞ 𝜕𝑉𝑥∗ 𝑉𝑐 𝜕𝑉𝑦∗ 𝜕𝑉𝑥∗ 𝜕𝑉𝑦∗
⇒ + 𝛿 𝜕𝑦∗ = 0 𝐼𝑓 ≠0⇒ ≠0 𝛿(𝑥)
𝐿 𝜕𝑥 ∗ 𝜕𝑥 ∗ 𝜕𝑦 ∗
𝑈 𝑉 𝛿
or 𝐿∞ ≈ 𝛿𝑐 ⇒ 𝑉𝑐 ≈ 𝑈∞ (𝐿 ) 𝑥

𝛿 ≪ 𝐿 𝑉𝑐 ≪ 𝑈∞ but it is not zero (finite)

𝑉𝑦
𝑉𝑦∗ = 𝛿 , 𝑡 ∗ = 𝑡⁄𝐿/𝑈
𝑈∞ ( ) ∞
𝐿

𝜕𝑉𝑥∗
Boundary layer theory does predict 𝑉𝑦 ≠ 0 𝑎𝑠 𝜕𝑥 ∗
≠ 0 ⇒ There is deceleration in 𝑥 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛.

Non-dimensionalize:
𝑈∞2
𝜕𝑉𝑥∗ 𝑈∞2
∗
𝜕𝑉𝑥∗ 2
𝛿 𝑉𝑦∗ 𝜕𝑉𝑥∗ 𝑈∞2
𝜕𝑝∗ 𝑈∞ 𝜕 2 𝑉𝑥∗ 𝑈∞ 𝜕 2 𝑉𝑥∗
𝜌 + 𝜌 𝑉𝑥 + 𝜌𝑈∞ ( ) = −𝜌 + 𝜇 [ + ]
𝐿 𝜕𝑡 ∗ 𝐿 𝜕𝑥 ∗ 𝐿 𝛿 𝜕𝑦 ∗ 𝐿 𝜕𝑥 ∗ 𝐿2 𝜕𝑥 ∗ 2 𝛿 2 𝜕𝑦 ∗ 2

𝜕𝑉𝑥∗ 𝜕𝑉 ∗ 𝜕𝑉 ∗ 𝜕𝑝∗ 1 𝜕2 𝑉 ∗ 𝐿 2 𝜕2 𝑉𝑥∗

Or 𝜕𝑡 ∗
+ 𝑉𝑥∗ 𝜕𝑥𝑥∗ + 𝑉𝑦∗ 𝜕𝑦𝑥∗ = − 𝜕𝑥 ∗ + 𝑅𝑒 [𝜕𝑥 ∗𝑥2 + (𝛿 ) 𝜕𝑦 ∗ 2
]

0−1 (0 − 1) (0 − 1)
96
 Each term is of the order of one except (𝛿 ≪ 𝐿 𝑜𝑟 𝐿 ≫ 𝛿) and 𝑅𝑒 ↑↑
1 𝜕2 𝑉 ∗ 1 𝐿 2 𝜕2 𝑉𝑥∗
Therefore, 𝑅𝑒 𝜕𝑥 ∗𝑥2 → 0 𝑎𝑛𝑑 ( ) 𝜕𝑦∗ 2
𝑅𝑒 𝛿
→?

𝜕𝑝∗ 1 𝐿 2 𝜕2 𝑉𝑥∗ 𝐿 2
= − 𝜕𝑥 ∗ + 𝑅𝑒 (𝛿 ) 𝜕𝑦 ∗ 2
(𝑅𝑒 𝑖𝑠 ℎ𝑖𝑔ℎ, 𝑠𝑜 𝑖𝑠 (𝛿 ) )

1 𝐿 2
Among the three possibilities: ( )
𝑅𝑒 𝛿
≫ 1 𝑜𝑟 ≪ 1 𝑜𝑟 ≈ 1, the first two are ruled out because both
viscous and inertial terms must to be retained, or they are important, to avoid the paradox. Only last
condition holds good:

1 𝐿 2 𝛿 1
( ) ≈ 1 𝑜𝑟 ≈ (𝑅𝑒 ↑ 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑙𝑎𝑦𝑒𝑟 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 𝛿 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠)
𝑅𝑒 𝛿 𝐿 √𝑅𝑒

NS dimensionless equation in BL becomes:

𝜕𝑉𝑥∗ 𝜕𝑉 ∗ 𝜕𝑉 ∗ 𝜕𝑝∗ 𝜕2 𝑉 ∗
+ 𝑉𝑥∗ 𝜕𝑥𝑥∗ + 𝑉𝑦∗ 𝜕𝑦𝑥∗ = − 𝜕𝑥 ∗ + 𝜕𝑦 ∗𝑥2 ⇒ 𝑅𝑒 has disappeared. It is embedded in the equation.
𝜕𝑡 ∗

𝑦 ∗ = 𝑓(𝛿)
𝛿 1
≈ ≪ 1 𝑖𝑓 𝑅𝑒 → ∞ (~100)
𝐿 √𝑅𝑒

No matter how high is Re, one of the viscous terms, namely,

2
𝑩𝒐𝒖𝒏𝒅𝒂𝒓𝒚 𝑳𝒂𝒚𝒆𝒓 𝑻𝒉𝒆𝒐𝒓𝒚: ∂ Vx
μ
must be retained in NS equation. Mathematically, δ ≪ L or yc ≪ xc .
∂y 2
The theory will predict finite drag on a solid surface. On the other hand,
both terms can be dropped (or xc ≈ yc ) in potential flow (Re → ∞), 𝑎𝑤𝑎𝑦 𝑓𝑟𝑜𝑚
{ a solid surface.
What about NS in y-direction?

Non-dimensionalize the 𝑦 − direction momentum equation using the correct/appropriate characteristic

variables:
𝛿
𝑉𝑥 = 𝑉𝑥∗ 𝑈∞ , 𝑥 = 𝑥 ∗ 𝐿, 𝑦 = 𝑦 ∗ 𝛿, 𝑉𝑦 = 𝑉𝑦∗ 𝑈∞ (𝑓𝑟𝑜𝑚 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦):
𝐿
𝜕𝑉𝑥 𝜕𝑉𝑦 𝛿 ∗ 2
𝜕𝑥
+ 𝜕𝑦 = 0 ⇒ (𝑉𝑦 )𝑐 = (𝑉𝑥 )𝑐 𝐿 ; 𝑝 = 𝑝 𝜌𝑈∞

Substitute,
𝑈∞2 𝛿 𝜕𝑉𝑦∗ 𝑈2 𝛿 𝜕𝑉𝑦∗ 𝑈 2 𝐿 𝜕𝑝∗ 𝑈∞ 𝛿 𝜕2 𝑉𝑦∗ 𝑈∞ 𝛿 𝜕2 𝑉𝑦∗
𝜌 ( ) 𝑉∗ + 𝜌 ∞ ( ) 𝑉𝑦∗ ∗ = −𝜌 ∞ ( ) ∗ +𝜇[ ( ) + ( ) ]
𝐿 𝐿 𝑥 𝜕𝑥 ∗ 𝐿 𝐿 𝜕𝑦 𝐿 𝛿 𝜕𝑦 𝐿2 𝐿 𝜕𝑥 ∗ 2 𝛿2 𝐿 𝜕𝑦 ∗ 2
∗ ∗
𝜕𝑉 𝜕𝑉 𝐿2 𝜕𝑝∗ 𝜇𝐿2 𝑈 𝛿 𝜕2 𝑉𝑦∗
𝑉𝑥∗ 𝑦∗ + 𝑉𝑦∗ 𝑦∗ = − ( 2 ) ∗ + ( 2 ) ∞2 drops out
𝜕𝑥 𝜕𝑦 𝛿 𝜕𝑦 𝛿𝜌𝑈 𝛿 𝐿 𝜕𝑦 ∗ 2 ∞

1 𝐿 2 𝑝 = 𝑝(𝑥)
( )
𝑅𝑒 𝛿
≈1

𝜕𝑝∗
Therefore, 𝜕𝑦 ∗
= 0 𝑝 ≠ 𝑝(𝑦) → No pressure-drop in 𝑦 direction in BL, which implies that potential
flow/regime impresses its pressure all the way to the surface.

97
Summary: Governing equations in BL (Prandtl BL) :

𝜕𝑉𝑥 𝜕𝑉𝑦
Continuity: 𝜕𝑥
𝜕𝑦
+ =0
𝜕𝑉𝑥 𝜕𝑉 𝜕𝑉𝑥 1 𝑑𝑝 𝜕2 𝑉𝑥
𝑥 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜕𝑡 + 𝑉𝑥 𝜕𝑥𝑥 + 𝑉𝑦 𝜕𝑦
= − 𝜌 𝑑𝑥 + 𝜈 𝜕𝑦 2
𝜕𝑝
𝑦 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜕𝑦 = 0

BC. 1 𝑦 = 0 𝑉𝑥 = 𝑉𝑦 = 0 (𝑛𝑜 𝑠𝑙𝑖𝑝)

𝑦 → ∞ 𝑉𝑥 = 𝑈(𝑥, 𝑡) (𝑓𝑟𝑒𝑒 𝑠𝑡𝑟𝑒𝑎𝑚 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦)

2. }
(≠ 𝛿) = 𝑈∞

3. 𝑥 = 0, 𝑉𝑥 = 𝑈∞ (𝑎𝑝𝑝𝑟𝑜𝑎𝑐ℎ 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦)

𝑉𝑦 𝑈∞
II 𝑉𝑥
II

I
I 𝑥 𝑥

98
 Lecture #20
Blasius Solution to BL Equations
Flow over a flat plate:
𝑦
𝛿 1 𝜇
→ 𝑣∞ 𝑣∞ 𝑝∞ ~ = √𝜌𝑣
𝑝∞ 𝑥 √𝑅𝑒𝑥 𝑥𝑥
𝑣∞ →
𝜇𝑥
→ 𝑝∞ 𝛿~ √𝜌𝑣
∞
{

𝜕𝑣𝑥 𝜕𝑣𝑥 𝜕𝑣𝑥 1 𝑑𝑝 𝜕 2 𝑣𝑥

X: Momentum: + 𝑣𝑥 + 𝑣𝑦 = − 𝜌 𝑑𝑥 + 𝜈
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑦 2

𝜕𝑣𝑥 𝑑𝑝 𝜕𝑣𝑥
𝑆𝑆: 𝜕𝑡
= 0; 𝑑𝑥
= 0; 𝜕𝑥
≠ 0; semi − infinite domain;

𝑑𝑝
Why? Because = 0 and potential regime imposes its uniform pressure 𝑃∞ to BL.
𝑑𝑦

Using stream function:

𝜕𝜓
𝜕𝜓 𝜕2 𝜓 𝜕𝜓 𝜕2 𝜓 𝜕3 𝜓
𝑦=0 ⇒ 𝜕𝑥
=0 𝜓 = 0 (𝑛𝑜 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)
𝜕𝑦 𝜕𝑥𝜕𝑦
− 𝜕𝑥 𝜕𝑦 2
= 𝜈 𝜕𝑦3 (modified NS) BC 1. 𝜕𝜓 } ⇒
=0
𝜕𝑦

𝜕𝜓
2. 𝑦 → ∞, 𝑣𝑥 = 𝑣∞ ⇒ 𝜕𝑦
= 𝑣∞

BC 2: y ⤏ δ(x) is replaced with y ⤏ ∞. Why? This was Blasius’ approach to solving PBL equations.
Conceptually, two regimes (potential and BL) approach each other asymptotically. A person swimming in
the BL does not see the potential flow regime, or vice versa, no matter how high is Re. Two theories stand
apart. Also, potential flow is first solved, and the pressure and velocity solutions (conditions) are imposed
at the top edge of the BL.

𝑦 𝑣∞
Let 𝜂 = = 𝑦√ ; 𝜓 = 𝜓(𝑥, 𝑦)
𝛿(𝑥) 𝜈𝑥

Revert to solving the momentum equations:

𝜓 = 𝑓(𝜂)𝑔(𝑥) (replacing y with 𝜂 and 𝑥)

𝜕𝜓 𝜕𝜂 𝑣
𝜕𝑦
= 𝑓 ′ (𝜂)𝑔(𝑥) 𝜕𝑦 = 𝑓 ′ (𝜂)𝑔(𝑥)√ 𝜈𝑥
∞

𝑣
BC 2. 𝑣∞ = 𝑓 ′ (𝜂)𝑔(𝑥)√ 𝜈𝑥
∞

99
Therefore, 𝑓 ′ (𝜂) = 1, 𝑔(𝑥) = √𝑣∞ 𝜈𝑥 (𝑡𝑎𝑘𝑒 𝑖𝑡 𝑎𝑟𝑏𝑖𝑡𝑟𝑎𝑟𝑦)

𝑣𝑥 = 𝜓𝑦 = 𝑔𝑓′⁄𝛿 = 𝑣∞ 𝑓′

𝑔𝑓 ′ 𝑦 1 𝜈𝑣∞
−𝑣𝑦 = 𝜓𝑥 = 𝑔′ 𝑓 − = −2 √ (𝜂𝑓 ′ − 𝑓)
2𝑥𝛿 𝑥

𝑔′ 𝑓′ 𝑔𝑓′′𝑦 𝑔𝑓′ 𝑔𝑓′′ 𝑔𝑓′′′

𝜓𝑥𝑦 = − − 2𝑥𝛿 ; 𝜓𝑦𝑦 = ; 𝜓𝑦𝑦𝑦 = (check all these calculations)
𝛿 2𝑥𝛿 2 𝛿2 𝛿3

𝜕𝜓
BCs. on 𝜓: (1) 𝑦 = 0 𝑣𝑥 = 0 𝜕𝑦
=0

𝜕𝜓 𝜕𝜓
(2) 𝑣𝑦 = 0 =0 (3) 𝑦 → ∞ 𝑣𝑥 = 𝑣∞ = 𝑣∞
𝜕𝑥 𝜕𝑦

BCs: on 𝑓: (1) 𝜂 = 0 (𝑦 = 0) 𝑓′ = 0

(2) 𝜂 = 0 (𝑦 = 0) 𝑓=0 (3) 𝜂 → ∞ (𝑦 → ∞) 𝑓 ′ = 1

Plugging,

𝑔𝑓′ 𝑔′𝑓′ 𝑔𝑓′′𝑦 𝑔𝑓′ 𝑔𝑓′𝑦 𝑔𝑓′′ 𝑔𝑓′′′

( − − 2𝑥𝛿) − (𝑔′ 𝑓 − ) =𝜈
𝛿 𝛿 2𝑥𝛿 2 2𝑥𝛿 𝛿2 𝛿3

𝛿 𝜇𝑥
Recall 𝑔 = 𝑣∞ 𝛿 𝑎𝑛𝑑 𝑔′ = 𝑣∞ 2𝑥
(𝑅𝑒𝑐𝑎𝑙𝑙 𝛿~ √𝜌𝑣 )
∞

𝑓𝑓 ′′ + 2𝑓 ′′′ = 0 BCs. (1-2) 𝜂 = 0 𝑓 ′ = 𝑓 = 0 and (3) 𝜂 → ∞ 𝑓 ′ = 1

𝑑𝑓′′ 1
= − 𝑓𝑓′′
𝑑𝜂 2

𝑑𝑓′′ 1
∫ = − ∫ 𝑓𝑑𝜂
𝑓′′ 2

𝜂 1 𝜂
1 𝜂 ∞ 1 𝜂 𝜂 ∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂
𝑓 ′ ′ = exp (− 2 ∫0 𝑓𝑑𝜂)⁄∫0 𝑒𝑥𝑝 (− 2 ∫0 𝑓𝑑𝜂) ⇒ 𝑓′ = ∫0 𝑓 ′′ 𝑑𝜂 = ∞ 1 𝜂
∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂)

𝜂 𝜂 1 𝜂
𝜂 ∫0 ∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂𝑑𝜂
𝑓 = ∫0 𝑓 ′ 𝑑𝜂 = ∞ 1 𝜂
∫0 𝑒𝑥𝑝(−2 ∫0 𝑓𝑑𝜂 )𝑑𝜂

Recall 𝑣𝑥 = 𝑣∞ 𝑓 ′ (𝜂)
1 𝜈𝑣∞
𝑣𝑦 = 2 √ 𝑥
(𝜂𝑓 ′ − 𝑓)

∞ 𝑣
𝜂 = 𝑦√ 𝜈𝑥
}

Blasius Solution (1908 - 10):

For small values of 𝜂 (𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒):

100
Displacement Thickness: potential flow/regime
is displaced upward
𝑦
𝑣𝑥
@𝑦 = 𝛿, = 0.99
𝑣∞

𝛿 ∗ (𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑚𝑒𝑛𝑡 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠)
𝑥

Fluid gets slowed down as 𝑥 → 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠; Hence, it displaces the potential flow in 𝑦 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛. The
distance by which the external potential flow field is displaced outwards as a consequence of the decrease
in velocity in the boundary layer is determined by
∞
𝑣∞ 𝛿 ∗ = ∫0 (𝑣∞ − 𝑣𝑥 )𝑑𝑦

decrease in volume flow due to BL.

The book by Kayes et al. mentions that “there is some ambiguity in speaking of the boundary layer
thickness if what is meant is the region in which velocity changes from 𝑣∞ to 0”. 𝛿 ∗ is a measure of the
displacement of the main stream resulting from the presence of the flat plate and its BL. The book by
Schlitching attributes the “displacement” to the difference between the y-component of the velocities of
BL and potential regime at the outer edge of the BL, which was not considered in calculations, but was
considered for x-component: 𝑦 𝑎𝑡 𝛿 𝑜𝑟 𝑦 → ∞, 𝑣𝑥 = 𝑣∞ . Revert to calculation

∞ 𝑣𝑥
𝛿 ∗ = ∫0 (1 − ) 𝑑𝑦
𝑣∞

𝜈𝑥 ∞ 𝜈𝑥 𝜈𝑥 𝜈𝑥
= √𝑣 ∫0 (1 − 𝑓 ′ (𝜂))𝑑𝜂 = √𝑣 (𝜂1 − 𝑓(𝜂1 )) = 1.72√𝑣 ≪ 5√𝑣
∞ ∞ ∞ ∞

𝑎𝑡 𝑦 = 8
where, 𝜂1 denotes point/location outside BL. { 𝑜𝑟 𝜂 = 5
𝑉𝑥 → 0.99 𝑣∞

𝜈𝑥 𝜈𝑥
Thus, 𝛿 ∗ = 1.72 √𝑉 ; 𝑐𝑜𝑚𝑎𝑝𝑟𝑒 𝑡𝑜 𝛿 = 5√𝑉 → 𝛿 ∗ ≈ 1⁄3 𝛿
∞ ∞

𝑑𝑝 𝑑𝑝
‘𝛿 ∗ ’ means 𝑑𝑥 ≠ 0 which was zero in potential flow and superimposed on BL because 𝑑𝑦 = 0 , as assumed
𝑑𝑝
in Blausius solution. Therefore, because of ‘displacement’, a new 𝑑𝑥 in potential flow (Bernoulli’s theorem)
should be considered in BL theory/solution as a correction → 2nd order BL.

Skin Friction: (Potential theory cannot predict skin friction/drag)-

𝑙
𝐷 = 𝑏 ∫0 𝜏𝑜 𝑑𝑥 (𝑏 ≡ 𝑤𝑖𝑑𝑡ℎ 𝑜𝑓 𝑝𝑙𝑎𝑡𝑒) 1.0
0.99
𝑑𝑣𝑥
𝜏𝑜 (𝑥) = 𝜇 |
𝑑𝑦 𝑦=0
𝑤ℎ𝑒𝑟𝑒 𝑣𝑥 = 𝑣∞ 𝑓 ′ (𝜂)

𝑓 ′ (𝜂) 102
𝑠𝑙𝑜𝑝𝑒 = 0.332 5.0
𝜂 ∞
 𝑣 𝑑𝑣𝑥 𝑣
∞
= 𝜇𝑣∞ √ 𝜈𝑥 𝑓 ′′ (0) (𝜏 = +𝜇 𝑑𝑦
) ∞
on plate = 0.332𝜇𝑣∞ √ 𝜈𝑥

3/2
𝑣∞ 𝑙 1
𝐷 = 𝑏𝜇0.332 ∫0 𝑑𝑥
√𝜈 √𝑥

𝐷 = 2 × 0.332 × 𝑏𝑣∞ √𝜇𝜌𝑙𝑣∞ No adjustable constant

𝜏𝑜 (𝑥) 𝜈 0.332
𝐶𝐷 = 2 = 0.332√ = (compare 𝑖𝑡 𝑤𝑖𝑡ℎ 𝐶𝐷 𝑓𝑜𝑟 𝑐𝑟𝑒𝑒𝑝𝑖𝑛𝑔 𝑓𝑙𝑜𝑤, 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑎𝑛𝑎𝑙𝑦𝑡𝑖𝑐𝑎𝑙𝑙𝑦!)
𝜌𝑣∞ 𝑣∞ 𝑥 √𝑅𝑒𝑥

Momentum thickness is also defined by slowing down of fluid:

2 ∞
𝜌𝑣∞ 𝜃 = 𝜌 ∫0 𝑣𝑥 (𝑣∞ − 𝑣𝑥 )𝑑𝑦

∞ 𝑣𝑥 𝑣𝑥 𝜈𝑥 ∞ 𝜈𝑥
𝜃 = ∫0 𝑣∞
(1 − 𝑣∞
) 𝑑𝑦 = √𝑣 ∫0 𝑓′ (1 − 𝑓′)𝑑𝜂 = 0.664√𝑣
∞ ∞

Recall that Blasius solved BL equation using a similarity transform with

𝑑𝑝 𝜕𝑣𝑥 𝜕𝑣𝑥 𝑑𝑝 𝜕2 𝑣𝑥
𝑑𝑥
= 0: 𝑣𝑥 𝜕𝑥
+ 𝑣𝑦 𝜕𝑦
= − 𝑑𝑥 }𝑜𝑢𝑡𝑒𝑟𝑓𝑙𝑜𝑤 + 𝜈 𝜕𝑦 2
−𝑖𝑚𝑝𝑜𝑠𝑒𝑑

𝑑𝑝
In many cases, 𝑑𝑥 ≠ 0 curved solid surface

BL Separation:

Flow past a sphere or cylinder shows such features.

𝑑𝑝 𝜕𝑣𝑥
⇒ 𝐼𝑓 𝑑𝑥
≠ 0; it is difficult to solve NS in BL with the 𝑣𝑥 𝜕𝑥
non - linear term.
𝑑𝑝 𝑑𝑝
⇒ Same flow- problems with ≠ 0 have been solved assuming 𝑑𝑥 = 𝑒 𝑥 , but without similarity
𝑑𝑥
transform.
⇒ Falkner & Skin showed that if 𝑉∞ ≈ 𝑥 𝑚 , then similarity transform solution is possible (𝜂 𝑒𝑥𝑖𝑠𝑡𝑠).

→ Boundary layer theory has been extensively studied using numerous methods including more popular
von Karman momentum balance and Karman-Pohlhausen method. In the same context (BL), many
different types of flow situations including boundary layer separation, flow past bluff bodies and blunt
objects, positive pressure gradient, solid curvature effect, streamline bodies, air foil, movement of golf
and cricket balls, unsteady-state flow, and flow instabilities, etc, have been studied in depth. The books
authored by Schlitching, and also, Deen have a detailed coverage of some of these advanced topics related
to hydrodynamics. We will skip these topics and instead move to the concentration and thermal boundary
layers in the next lecture.

103
 Lecture #21
Thermal and Mass Boundary layer Theory
⇒ Consider diffusion only (Levich “Physicochemical Hydrodynamics”):
𝑦
𝑣𝑥
→ 𝛿𝑚 ( ≈ 0.99)
𝑣∞ → 𝑣∞
𝐶∞ →
𝐶
𝛿𝑐 ( ≈ 0.99)
𝐶∞
𝐶𝑠 𝑥
Consider two cases: 𝑆𝑐 ≈ 1000 (𝑙𝑖𝑞𝑢𝑖𝑑𝑠) & ≈ 1 (𝑔𝑎𝑠𝑒𝑠)
↓ 𝜈/𝐷
Case 1: 𝛿𝑚 ≫ 𝛿𝑐 𝑜𝑟 𝜈 ≫ 𝐷 (𝑙𝑖𝑞𝑢𝑖𝑑𝑠) (mass boundary layer is within momentum boundary layer)
Momentum transfer occurs over 𝛿𝑚 ≫ mass transfer diffusion occurs over 𝛿𝑐 .
BL equation for species:

𝜕𝐶 𝜕𝐶 𝜕2 𝐶
𝑣𝑥 𝜕𝑥 + 𝑣𝑦 𝜕𝑦 = 𝐷 𝜕𝑦 2 + (𝑡𝑒𝑟𝑚𝑠 𝑛𝑒𝑔𝑙𝑒𝑐𝑡𝑒𝑑)

𝜕𝐶 𝐶∞ 𝜕𝐶 𝐶∞ 𝜕2 𝐶 𝐶∞ 𝜕2 𝐶 𝐶∞
≈ ; ≈ , ≈ ; ≈
𝜕𝑥 𝑙 𝜕𝑦 𝛿𝑐 𝜕𝑥 2 𝑙2 𝜕𝑦 2 𝛿𝑐2

𝜕𝐶 𝐶∞
𝑣𝑦 𝜕𝑦 ≈ 𝑣𝑦 (𝐲 ≈ 𝛅𝐜 𝐚𝐧𝐝 𝐧𝐨𝐭 𝛅𝐦 )
𝛿𝑐

3/2 2
1.33 𝑣∞ 𝑦 𝜈𝑦 2
Recall 𝑣𝑦 (𝑛𝑒𝑎𝑟 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒) = ≈ 3 (see previous lectures)
16𝜈 1/2 𝑥 3/2 𝛿𝑚

𝜕𝐶 𝜈𝛿𝑐2 𝐶∞ 𝜈𝛿𝑐 𝐶∞ 𝜕2 𝐶 𝐶
Therefore, 𝑣𝑦 𝜕𝑦 ≈ 3 𝛿 = 3 and 𝐷 𝜕𝑦 2 ≈ 𝐷 𝛿∞2
𝛿𝑚 𝑐 𝛿𝑚 𝑐

𝛿 𝐶 𝛿 3 1
𝜈 𝛿3𝑐 𝐶∞ ≈ 𝐷 𝛿∞2 ⇒ (𝛿 𝑐 ) ≈ (𝜈 ⁄𝐷 )
𝑚 𝑐 𝑚

𝛿𝑚
Or 𝛿𝑐 ≈ 1 (𝑁𝑜𝑡𝑒: 𝑆𝑐 ↑↑ 𝑜𝑟 𝛿𝑚 ≫ 𝛿𝑐 )
𝑆𝑐 3
𝜹𝒎
𝑺𝒊𝒎𝒊𝒍𝒂𝒓𝒍𝒚, 𝜹𝑻 (𝒉𝒆𝒂𝒕 𝒕𝒓𝒂𝒏𝒔𝒇𝒆𝒓) ≈ 1 (analogy with mass transfer)
𝑃𝑟 3
𝛿𝑐 𝑃𝑟 1/3
𝛿𝑇
≈( )
𝑆𝑐
or 𝛿𝑐 ≈ 𝛿𝑇
𝛿𝑚 𝜈 1/2
See last lectures: ≈ (𝑣 )
𝑥 ∞ 𝑥
𝛿𝑐 𝜈 1/2 𝐷 1/3
≈ (𝑣 ) (𝜈 )
𝑥 ∞𝑥

104
 1 1
𝑥
⇒ 𝛿𝑐 ≈ 𝐷 3 𝜈 6 √𝑣
∞
𝜕𝐶 𝐷(𝐶∞ −𝐶𝑠 ) 𝐷(𝐶∞ −𝐶𝑠 )
If 𝑁 = −𝐷 , 𝑁≈ ≈ 1 1
𝜕𝑦 𝛿𝑐 (𝑥) 𝑥
𝐷3 𝜈6 √
𝑣∞

𝑚𝑜𝑙𝑒𝑠 𝐷 2/3 𝑣∞ 1/2

Or 𝑁 (𝑓𝑙𝑢𝑥, )=( 1 1 ) (𝐶∞ − 𝐶𝑠 )
𝑐𝑚2 −𝑠
𝜈6 𝑥 2
𝟐 𝟏 𝟏 𝟏
 𝒌𝒎 (𝒎𝒂𝒔𝒔 − 𝒕𝒓𝒂𝒏𝒔𝒇𝒆𝒓 𝒄𝒐𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒕) ~ 𝑫𝟑 𝝂−𝟔 𝒗∞ 𝟐 𝒙−𝟐 ⇒ 𝒍𝒊𝒒𝒖𝒊𝒅 − 𝒔𝒐𝒍𝒊𝒅
 Compare to film theory 𝑘𝑚 ~ 𝐷 (SS, 1D, stagnant film)
1 1 1
 Compare to surface renewal theory 𝑘𝑚 ~ 𝐷 2 𝑣𝑚𝑎𝑥 2 𝑥 −2 (gas-liquid surface is renewed)

Case 2: Consider 𝛿𝑐 ≈ 𝑜(1) ⇒ 𝑔𝑎𝑠

(momentum transfer ≈ molecular transfer: 𝜹𝒎 ≈ 𝜹𝒄 )

Now 𝑦 ≪ 𝛿𝑚 : 𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒 → 𝛿𝑚 ⁄𝛿𝑐

𝑣∞ →
Momentum equation: 𝐶∞ →
𝐶𝑠
𝜕𝑉 𝜕𝑉 𝜕2 𝑉
𝑣𝑥 𝜕𝑥𝑥 + 𝑣𝑦 𝜕𝑦𝑥
= 𝜈 𝜕𝑦2𝑥
𝜕𝐶 𝜕𝐶 𝜕2 𝐶
Species Balance: 𝑣𝑥 𝜕𝑥 + 𝑣𝑦 𝜕𝑦 = 𝐷 𝜕𝑦2
𝑦 𝑦
𝑦 ∗ = ⁄𝛿 𝑜𝑟 ⁄𝛿 (𝑠𝑎𝑚𝑒 𝑜𝑟𝑑𝑒𝑟)
𝑚 𝑐
𝑣𝑥 𝑣𝑦 𝛿
Non-dimensionalize: 𝑣𝑥 ∗ ∗
= ⁄𝑣∞ , 𝑣𝑦 = ⁄𝑣∞ 𝑚⁄𝑙 (𝑓𝑟𝑜𝑚 𝐶𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦)
(𝛿𝑚 ≈ 1⁄ )
{ √𝑅𝑒
𝜕𝑣 ∗ 𝜕𝑣 ∗ 𝜕2 𝑣 ∗
Recall: 𝑣𝑥∗ 𝜕𝑥𝑥∗ + 𝑣𝑦∗ 𝜕𝑦𝑥∗ = 𝜕𝑦∗𝑥2 (𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑏𝑎𝑙𝑎𝑛𝑐𝑒)
𝜕𝐶 ∗ 𝜕𝐶 ∗ 𝜕2 𝐶 ∗
𝑣𝑥∗ 𝜕𝑥 ∗ + 𝑣𝑦∗ 𝜕𝑦∗ = 𝜕𝑦∗ 2 (𝑠𝑝𝑒𝑐𝑖𝑒𝑠 𝑏𝑎𝑙𝑎𝑛𝑐𝑒)
(𝑃𝑒𝑐𝑙𝑒𝑡 𝑛𝑜 𝑜𝑟 𝑅𝑒𝑆𝑐 is hidden as 𝑅𝑒 in NS equation)
1
𝑅𝑒𝑆𝑐 = 𝑃𝑒 ≈ 𝑜 (𝛿 2 )
𝑐

𝛿𝑐 ≈ 1⁄
√𝑅𝑒𝑆𝑐
𝛿𝑐 𝑅𝑒
≈ √ ≈ 1⁄ ⇒ 𝑔𝑎𝑠𝑒𝑠 ~ 0(1)
𝛿𝑚 𝑅𝑒𝑆𝑐 √𝑆𝑐
Compare for liquids:

𝛿𝑐
𝛿𝑚
≈ 1⁄ 1
𝑆𝑐 3
𝑙𝑖𝑞𝑢𝑖𝑑𝑠 𝑃𝑟 ≈ 1000
Thermal boundary layer: replace 𝑆𝑐 𝑏𝑦 𝑃𝑟 { and 𝛿𝑐 by 𝛿ℎ (analogy between mass
𝑔𝑎𝑠𝑒𝑠 𝑃𝑟 ≈ 1
and heat transport). Similarly, km by h and Sh number by Nu number.

105
 Turbulence: A feature of fluid flow at high Reynolds number
(1) Structure of turbulence
(2) Reynold’s mathematical description of turbulent flow.

 Laminar flow is characterized by streamlines; dye-test with digital photographs can be used to
track the path of the particle, and streamlines can be plotted.
 Leave transience flow regime between laminar and turbulence, all flows which are not laminar
are turbulent, and vice – versa! Thus, fluid flow can be categorized in these two categories.
 Most of real world applications including industrial are based on turbulent flows.

Nature/characteristics of turbulence
(1) 𝑅𝑒 is large (𝑅𝑒 > 2100 𝑓𝑜𝑟 𝑡𝑢𝑏𝑢𝑙𝑎𝑟 𝑓𝑙𝑜𝑤)
(2) Irregularity in flow (dye test or visual or video can be used to check such feature)
(3) “Diffusivity” increases; thus mixing is high.
(4) Rotational
(5) Dissipation of heat/energy
(6) Inertial effects dominate the viscous effects
(7) Not a feature of fluids but of flow.
(8) Not amenable to mathematical analysis; rely heavily on models, theories, and experimental
measurements. Turbulent flow can be mathematically perceived as an instability in flow, arising
due to perturbation in laminar-turbulent flow regime. The prototype examples 1-12 all considered
laminar flow and could be solved relatively easily. Situation here is different.
(9) The most closest pictorial or visual description of a turbulent flow could be “violent swirling” flow,
termed “eddies" (not to be mixed up with streamlines having
angular velocity (vƟ ) in the regular or common laminar flow) .
(10) Yet, we are referring to continuum (and not molecular).

1 𝑡+𝑡𝑜
𝑣̅𝑧 = ∫ 𝑣𝑧 𝑑𝑡
𝑡𝑜 𝑡
𝑣𝑧 𝑣𝑧′ (𝑓𝑙𝑢𝑐𝑡𝑢𝑎𝑡𝑖𝑜𝑛)
time-average
𝑡

⇒ One approach is to model “smoothed” behaviour only ⇒ Reynold’s approach using time-averaged
quantities (to be discussed in the next lecture)
⇒ Another approach is to model the fluctuations, but non-linearity in 𝑁𝑆 theorem leads to “chaotic”
behaviour which is extremely sensitive to initial conditions.

Mathematically describe turbulence:

Consider velocity fluctuations

̅ = 𝟎 (𝒂𝒗𝒆𝒓𝒂𝒈𝒆𝒅 𝒘𝒓𝒕 𝒕𝒊𝒎𝒆)

𝒗′
⃗⃗ = ⃗𝒗⃗ + ⃗⃗⃗⃗
𝒗 𝒗′ ; where } If v̅′ ≠ 0, then it is no good for analysis.
̅̅̅̅̅̅̅
(𝒗′)𝟐 ≠ 𝟎

106
 - There are other theories: Taylor Vorticity Transport Theory, Von-Karman’s similarity Hypothesis

Example: “The law of the wall” (BSL P162, old edition)

𝑅
𝑠
→ 𝑟
→
→ 𝑠
⃗⃗
𝐷𝑉
𝜌 𝐷𝑡 = −∇𝑝̅ − ∇. (𝜏 𝑙 + 𝜏 𝑡 ) + 𝜌𝑔⃗
𝑑𝑝̅ 𝑙 1 𝑑 𝑡 )
0 = − 𝑑𝑧 − 𝑟 𝑑𝑟 𝑟 (𝜏𝑟𝑧 + 𝜏𝑟𝑧

𝑩𝑪𝒔: 𝜏 𝑙 = 0 𝑎𝑡 𝑟 = 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)
𝜏𝑡 = 0 𝑎𝑡 𝑟 = 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐) (𝑁𝑜𝑡𝑒 𝜏 𝑡 𝑖𝑠 𝑎𝑙𝑠𝑜 0 ; 𝑟 = 𝑅)
𝑑𝑝̅ 𝑟 ̅
𝑑𝑉 ̅
𝑑𝑉 2
Integrate, 𝜏 𝑙 + 𝜏 𝑡 = − 𝑑𝑧 2 (= −𝜇 ( 𝑑𝑟𝑧 ) − 𝜌𝑙 2 ( 𝑑𝑟𝑧 ) )
𝑑𝑝̅ 𝑅
At the wall 𝜏𝑡𝑜𝑡𝑎𝑙,𝑤 = 𝜏𝑜 = − 𝑑𝑧 2 laminar turbulent components

𝜏 𝑙 + 𝜏 𝑡 = 𝜏𝑜 𝑟⁄𝑅 − (1) (𝜏𝑜 𝑖𝑠 𝑡ℎ𝑒 𝑤𝑎𝑙𝑙 𝑠ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑠𝑠)

𝑡 ̅
𝑑𝑉 ̅𝑧
𝑑𝑉
𝜏𝑟𝑧 = −𝜌𝑙 2 | 𝑑𝑟𝑧 | (turbulent flow, away from the wall)
𝑑𝑟

𝑙𝑒𝑡 𝑙 = 𝑘. 𝑠 (An eddy cannot be larger than its distance from the walls in a thin BL)
𝐴𝑙𝑠𝑜, 𝑠 = 𝑅 − 𝑟 𝑜𝑟 𝑑𝑠 = −𝑑𝑟
̅
𝑑𝑉 ̅𝑧
𝑑𝑉 𝑅−𝑠
𝑡
Therefore, 𝜏𝑟𝑧 = 𝜌𝑘12 𝑠 2 | 𝑑𝑠𝑧 | = 𝜏𝑜 𝑓𝑟𝑜𝑚 (1) (neglecting ‘laminar’/viscous sub-layer;
𝑑𝑠 𝑅

𝜏 𝑡 ≫ 𝜏 𝑙 far from wall)

𝑠
Also, 𝑅 ≪ 1 close to wall ≈ 𝜏𝑜 (1 − 0) = 𝜏𝑜
(There is a contradiction in mathematical analysis; physically, Prandtl theory is non-defensible)
̅𝒛
𝒅𝑽 𝟏 𝝉 𝜏
Or = ± 𝒌 𝒔 √ 𝝆𝒐 − (2)
𝒅𝒔 𝟏

𝜏
Let us define friction velocity, 𝑉 ∗ = √ 𝜌𝑜

𝑉 ∗
Plug in outside buffer zone and integrate, 𝑉̅𝑧 = 𝑘 𝑙𝑛𝑠 + 𝑐1
1
∗
𝑉
𝑉̅𝑧 − 𝑉̅𝑧1 = 𝑙𝑛(𝑠⁄𝑠1 )
𝑘1
̅
𝑉 𝑠𝑉 ∗ 𝜌 𝟏 +
𝒔 ⁄ +)
𝐷𝑒𝑓𝑖𝑛𝑒, 𝑉 + = 𝑉𝑧∗ , 𝑠 + = (dimensionless #); Therefore, 𝑽+ − 𝑽+
𝟏 = 𝒌 𝒍𝒏 (
𝜇 𝟏 𝒔𝟏
(𝑛𝑜𝑡 𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑤𝑎𝑙𝑙/𝑛𝑜𝑡 𝑐𝑙𝑜𝑠𝑒 𝑡𝑜
𝑐𝑒𝑛𝑡𝑒𝑟 𝑒𝑖𝑡ℎ𝑒𝑟 𝑖𝑛 𝑡𝑢𝑟𝑏𝑙𝑒𝑛𝑡 𝑐𝑜𝑟𝑒)
112
Homogeneous Turbulence: √(𝑣𝑥′ )2 rms of fluctuation is constant over entire turbulence but √(𝑣𝑥′ )2 ≠
2
√(𝑣𝑦′ ) ≠ √(𝑣𝑧′ )2 .

2
⇒ Isotropic turbulence is homogeneous turbulence with the addition of √(𝑣𝑥′ )2 = √(𝑣𝑦′ ) = √(𝑣𝑧′ )2
(fluctuations in the direction of flow ≈ ⊥ 𝑟 to the flow )
Mathematic description of isotropic turbulence is possible. However, it exists only in the absence of mean
velocity gradients, far from walls – which is not much of application! We are interested in the flow near
the surface.
⇒ Shear turbulence – very common: 𝑣𝑥 ′(𝑥, 𝑡)
⇒ Scale of turbulence – is determined by distance over which correlated motion exists, eg. One might
look at the correlation tensor:
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
𝑄𝑖𝑗 (𝑟⃗) = 𝑣 ′ ′
𝑖 (𝑥)𝑣𝑗 (𝑥 + 𝑟) ≠ 0

√̅̅̅̅̅̅̅
(𝑣𝑖′ )2⁄
(𝑡)
⇒ Intensity of turbulence: 𝑣𝑖 = 𝑣̅𝑖 ;
Correlation:

ℎ𝑖𝑔ℎ𝑒𝑟 𝑑𝑒𝑔𝑟𝑒𝑒
𝐼𝑛𝑠𝑡𝑎𝑛𝑡𝑎𝑛𝑒𝑜𝑢𝑠
𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑜𝑓 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛
𝑁𝑜 − 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛

𝑡𝑖𝑚𝑒

⇒ What is the smallest eddy size possible? Smallest scale of turbulence?

Kolmogorov (1941) proposed Universal Equilibrium Theory. Suppose we have large eddies of length ′𝑙𝑜 ′
1
′2 )2 , with Reynolds number based on 𝑅𝑒 = 𝑣𝑜 𝑙𝑜 ≫ 1.
̅̅̅̅
and velocity 𝑣𝑜 = (𝑣 𝑜 𝜈
In general, these eddies are unstable and break down to small eddies when subjected to a disturbance:
Length scale 𝑙, 𝑣; 𝑙𝑜 ≫ 𝑙1 ≫ 𝑙𝑒 ≫ ⋯ 𝑙𝑛
↑
𝑅𝑒 is approaching 1 (viscous term is becoming important)
Assume:

1. Characteristic time for an eddy is one turn over time.

2. Kinetic energy transfer rate:
𝑑 1 ′2 𝑣2 𝑣2 𝑣3
𝑁𝑜𝑡𝑒: 𝑇𝑎𝑦𝑙𝑜𝑟 𝑚𝑖𝑐𝑟𝑜𝑠𝑐𝑎𝑙𝑒:
𝜖𝑜 = 𝑑𝑡 (2 ̅̅̅̅
𝑣 ) ≈ 𝑡𝑜 = 𝑙𝑜 𝑜 = 𝑙 𝑜 15𝜈𝑣𝑡2
⁄𝑣𝑜 𝑜 𝜖= ⁄2
𝑙𝑡
↑ 1
𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 𝑡𝑖𝑚𝑒 ̅̅̅̅
𝑣𝑡 = (𝑣 ′2 )2
( )

107
 3. 𝜖 is independent of eddy size
𝜖 𝜖
Large eddy → Small eddy → Kolmogorov eddy (smallest sized eddy)
𝑣𝑛3 𝑣𝑜3
Viscous dissipation is same everywhere (𝜖𝑛 = 𝑙𝑛
= 𝑙𝑜
= 𝜖𝑜 )
Ref. Book by Tennekes and Lumley

Dimensional analysis: Parameters determining small scale motion will depend on 𝜖 (𝑐𝑚2 /𝑠 3 ) and
𝜈 (𝑐𝑚2 /𝑠) (rate of dissipation & kinematic viscosity which is also a rate)

1/4
𝜈3
In (𝐾𝑜𝑙𝑚𝑜𝑔𝑜𝑟𝑜𝑣 𝑒𝑑𝑑𝑦) 𝑙𝑛 = 𝑓(𝜖, 𝜈) ≈ ( 𝜖 ) ; check dimensions

constant parameters
𝑙𝑛 𝑣𝑛
𝑣𝑛 = (𝜈𝜖)1/4 ; 𝜏𝑛 = (𝜈⁄𝜖 )1/2 ; 𝑅𝑒𝑛 = ≈ 1 (similarly, define the other variables)
𝜈

3 1/4
𝑙𝑛 𝑣𝑛 (𝜈 ⁄𝜖 ) (𝜈𝜖)1/4
Local Reynold number, 𝑅𝑒𝑛 = 𝜈
= 𝜈
= 1 (hence proven!)

1/4
1/4 1/4
3 3 𝜈 3 𝑙𝑜
𝑙𝑛 = (𝜈 ⁄𝜖 ) = (𝜈 ⁄𝑣 3 ) =( ⁄ 3)
𝑜 𝑣𝑜
𝑙𝑜
1/4 1
𝑙𝑛 𝑠𝑚𝑎𝑙𝑙𝑒𝑠𝑡 𝜈3𝑙 1 𝜈3 4 1
⇒ = ( ) = ( 𝑜⁄ 3 ) = {𝑣 3 𝑙3 } = 3/4
𝑙𝑜 𝑙𝑎𝑟𝑔𝑒𝑠𝑡 𝑣𝑜 𝑙𝑜 𝑜 𝑜 𝑅𝑒𝑜
𝑷𝒉𝒚𝒔𝒊𝒄𝒂𝒍 𝑺𝒄𝒂𝒍𝒆
𝑲𝒐𝒍𝒎𝒐𝒈𝒐𝒓𝒐𝒗 𝑺𝒄𝒂𝒍𝒆 = 𝟑/𝟒
𝑹𝒆

𝑙𝑛 1 𝑣𝑛 1 𝜏𝑛 1
𝑙𝑜
= 3/4 ; 𝑣𝑜
= 1/4 ; 𝜏𝑜
= 1/2 (check these equalities)
𝑅𝑒0 𝑅𝑒0 𝑅𝑒0

𝜌𝑣𝑛 𝑙𝑛 𝜇 𝜇
Ex. 𝑅𝑒𝑛 = = 1 ⇒ 𝑣𝑛 = ⁄𝜌𝑙 ; 𝑣𝑜 = ⁄𝜌𝑙 𝑅𝑒𝑜 (𝑅𝑒𝑜 ≠ 1)
𝜇 𝑛 𝑜
𝑣𝑛 𝑙𝑜 1 3/4 −1 −1/4
Therefore, =( ) = 𝑅𝑒0 𝑅𝑒0 = 𝑅𝑒0
𝑣𝑜 𝑙𝑛 𝑅𝑒𝑜
𝑣 1
⇒ 𝑣𝑛 = 1/4
𝑜 𝑅𝑒0

𝑆𝑡𝑎𝑐𝑘:

𝑙𝑛 = 𝑠𝑚𝑎𝑙𝑙𝑒𝑠𝑡 𝑒𝑑𝑑𝑦 𝑝𝑟𝑒𝑠𝑒𝑛𝑡

𝑙𝑜 𝑙𝑛 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒 𝑤ℎ𝑒𝑛 𝑙𝑎𝑟𝑔𝑒 𝑒𝑑𝑑𝑦
𝑏𝑟𝑒𝑎𝑘𝑠 𝑢𝑝.
𝐿
𝑙 𝑙𝑛 −3/4 𝑙𝑜 −1/2
= 𝑅𝑒𝑜 ~𝑅𝑒𝐿
𝑙𝑜 𝐿
Jet: 𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑤𝑖𝑛𝑑 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦

𝑙𝑜
𝑣𝑜
108

𝐿
 Lecture #22
Turbulent flow: Some mathematical analysis

𝑆𝑆 Unsteady state

𝑣′ 𝑣′

𝑣
𝑣̅ (𝑡) 𝑣 𝑣̅ (𝑡)
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝑡𝑖𝑚𝑒 𝑠𝑐𝑎𝑙𝑒
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝑡𝑖𝑚𝑒 𝑠𝑐𝑎𝑙𝑒

𝑡→ 𝑡→
̅ + 𝒗′ (𝑓𝑙𝑢𝑐𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠, 𝑠𝑚𝑎𝑙𝑙 𝑡𝑖𝑚𝑒 − 𝑠𝑐𝑎𝑙𝑒)
Reynold’s decomposition: 𝒗 = 𝒗

Time-average velocity/time-smoothed velocity

𝒗′ is defined such that 𝒗̅′ = 𝟎

Continuity: ∇. (𝑣̅ + 𝑣 ′ ) = 0 (all notions are vector) Take time-average ∇. 𝑣̅ = 0

𝜕
NS: 𝜌( (𝑣̅ + 𝑣 ′ ) + [(𝑣̅ + 𝑣 ′ ). ∇](𝑣̅ + 𝑣 ′ ) = −∇(p
̅ + 𝑝′ ) + 𝜇∇2 (𝑣̅ + 𝑣 ′ )
𝜕𝑡

Recall: ∇. (𝑣⃗ 𝑣⃗ ) = (𝑣⃗. ∇)𝑣⃗ + 𝑣⃗ (∇. 𝑣⃗)

Therefore, [(𝑣̅ + 𝑣 ′ ). ∇](𝑣̅ + 𝑣 ′ ) = ∇. [(𝑣̅ + 𝑣 ′ )(𝑣̅ + 𝑣 ′ )]

(𝑣̅𝑥 + 𝑣𝑥 ′)2 (𝑣̅𝑥 + 𝑣𝑥 ′)(𝑣̅𝑦 + 𝑣𝑦 ′) (𝑣̅𝑥 + 𝑣𝑥 ′)(𝑣̅𝑧 + 𝑣𝑧 ′)

2
= ∇. ((𝑣̅𝑥 + 𝑣𝑥 ′)(𝑣̅𝑦 + 𝑣𝑦 ′) (𝑣̅𝑦 + 𝑣𝑦 ′) (𝑣̅𝑦 + 𝑣𝑦 ′)(𝑣̅𝑧 + 𝑣𝑧 ′))
(𝑣̅𝑥 + 𝑣𝑥 ′)(𝑣̅𝑧 + 𝑣𝑧 ′) (𝑣̅𝑦 + 𝑣𝑦 ′)(𝑣̅𝑧 + 𝑣𝑧 ′) (𝑣̅𝑧 + 𝑣𝑧 ′)2

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Noting that 𝑣̅𝑥2 + 2𝑣̅𝑥 𝑣̅𝑥′ + ̅̅̅̅
𝑣𝑥′2 = 𝑣̅𝑥2 + ̅̅̅̅
𝑣𝑥′2 , and similarly, the time averaged 𝒑̅′ = 𝒗̅′ = 𝟎, take the time
average of the entire NS equation:

𝜕𝑣̅ ̅̅̅̅
𝜌( 𝜕𝑡 + ∇. 𝑣̅ 𝑣̅ + ∇𝑣 ′2
) = −∇𝑝̅ + 𝜇∇2 𝑣̅ (Note: ̅̅̅̅
𝑣 ′2 ≠ 0; ̅𝒗
̅̅̅′ = 0)

̅̅̅̅̅̅
𝑣 ′ ′ ̅̅̅̅̅̅
′ ′ ̅̅̅̅̅̅
𝑥 𝑣𝑥 𝑣𝑥 𝑣𝑦 𝑣𝑥 𝑣𝑧
′ ′

̅̅̅̅ ̅̅̅̅̅̅
𝑣 ′2 = (𝑣 ′ ′ ̅̅̅̅̅̅
′ ′ ̅̅̅̅̅̅
′ ′ ′2 𝑁
where, 𝑥 𝑣𝑦 𝑣𝑦 𝑣𝑦 𝑣𝑦 𝑣𝑧 ) 𝑜𝑟 𝜌𝑣̅1 ≡ 𝑹𝒆𝒚𝒏𝒐𝒍𝒅 𝒔𝒕𝒓𝒆𝒔𝒔𝒆𝒔 (𝑚2 )
̅̅̅̅̅̅
𝑣𝑥′ 𝑣𝑧′ ̅̅̅̅̅̅
𝑣𝑦′ 𝑣𝑧′ ̅̅̅̅̅̅
𝑣𝑧′ 𝑣𝑧′

𝐷𝑣̅ ̅̅̅̅
′2
𝜌 = −∇𝑝̅ − ∇. 𝜏 − ∇𝜌𝑣 (Note that an extra term has arisen on time-averaging)
𝐷𝑡

109
 ̅̅̅̅
Thus, ∇𝜌𝑣 ′2
= ∇. 𝜏 (𝑡) (define shear stress in turbulent flow)

𝐷𝑣̅ ̅̅̅̅̅̅̅
𝜌 = −∇𝑝̅ − ∇. 𝜏 𝑙 − ∇. 𝜏 𝑡 ( 𝝉𝒕 = −𝝆𝒗 ′ ′
𝟏 𝒗𝟏 ; 𝑎𝑛 𝑒𝑥𝑡𝑟𝑎 𝑠𝑜𝑢𝑟𝑐𝑒 𝑜𝑓 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛)
𝐷𝑡

Component – wise:

𝜕𝑣̅𝑥 𝜕𝑣̅𝑥 𝜕𝑣̅𝑥 𝜕𝑣̅𝑥

𝜌( + 𝑣̅𝑥 + 𝑣̅𝑦 + 𝑣̅𝑧 )
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝑝̅ 𝜕 2 𝑣̅𝑥 𝜕 2 𝑣̅𝑥 𝜕 2 𝑣̅𝑥 𝜕 ̅̅̅̅̅̅̅
′ ′
𝜕 ̅̅̅̅̅̅̅ 𝜕 ̅̅̅̅̅̅̅
=− +𝜇( 2 + + ) − 𝜌 ( (𝑣 𝑥 𝑣𝑥 ) + (𝑣𝑥′ 𝑣𝑦 ′ ) + (𝑣 ′𝑣 ′
))
𝜕𝑥 𝜕𝑥 𝜕𝑦 2 𝜕𝑧 2 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝑥 𝑧

⇒ See the difference between the NS equation for Laminar flow and turbulent flow – an extra term arising
on the RHS which describes an extra source (rate) of momentum generation or transport attributed to
‘eddies’ formation/generation, apart from the time-averaged quantities for all variables 𝑣⃗ 𝑎𝑛𝑑 𝑃 . The
mathematical analysis is consistent with the experimental (photographic) observation of the flow
structure of the boundary layer at the plate surface. Even in turbulent boundary layer, there is a region
close to the surface that resembles the ‘laminar’ boundary layer, hence called the laminar sub layer or
viscous layer, discussed later.

→ 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦
→
𝑣∞ 𝑙𝑎𝑦𝑒𝑟
→
→ → 𝑏𝑢𝑓𝑓𝑒𝑟 𝑙𝑎𝑦𝑒𝑟
𝑅
→ 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑠𝑢𝑏 − 𝑙𝑎𝑦𝑒𝑟

𝑝𝑙𝑎𝑡𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦
𝑇. 𝐼. 𝑖𝑠 𝑧𝑒𝑟𝑜
1 ⁄
̅̅̅′ 2 ) 2 ~
(𝑣 10%
𝑧
Turbulence intensity: 𝑣̅𝑧
𝑟 = 𝑅 (𝑤𝑎𝑙𝑙𝑙)
𝑅𝑀𝑆 𝑓𝑙𝑢𝑐𝑡𝑢𝑎𝑡𝑖𝑜𝑛 𝑟=0
= 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤
𝑀𝑒𝑎𝑛 𝑐𝑒𝑛𝑡𝑒𝑟
Recall: 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤 Turbulent flow (there are two components)
(𝑡𝑢𝑏𝑒)
̅𝜏𝑡𝑜𝑡𝑎𝑙
̅𝜏𝑡𝑜𝑡𝑎𝑙 ,
↑ ̅𝜏𝑡𝑜𝑡𝑎𝑙 𝜏̅ 𝑡 ,
𝜏 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐
𝜏̅ 𝑙 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡
(𝜏̅𝑡 + 𝜏̅𝑙 )

𝑟=0 𝑟=𝑅 𝑟=0 𝑟=𝑅 𝑟=0 𝑟=𝑅

𝑐𝑒𝑛𝑡𝑒𝑟 (𝑤𝑎𝑙𝑙) ̅𝜏𝑡𝑜𝑡𝑎𝑙 = ̅𝜏𝑙𝑎𝑚𝑖𝑛𝑎𝑟 + ̅𝜏𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝜏̅ 𝑡 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠 𝑎𝑛𝑑 𝜏̅ 𝑙 = 𝑙𝑎𝑚𝑖𝑛𝑎𝑟
𝑡ℎ𝑒𝑛 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒𝑠 𝑠𝑡𝑟𝑒𝑠𝑠 𝑖𝑛 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑐𝑒

110
(𝜌𝑣𝑧2′ + 𝜏̅𝑙 )
 ̅̅̅̅̅
Boussinesq model: 𝜏 𝑡 (= −𝜌𝑣 ′𝑣 ′ ) = −𝜇𝑡 𝜕𝑣̅𝑥 analogous to 𝜏 𝑙 = −𝜇 𝜕𝑣𝑥 (1𝐷)
𝜕𝑦 𝜕𝑦

(𝝁𝒕 𝒊𝒔 𝒆𝒅𝒅𝒚 𝒗𝒊𝒔𝒄𝒐𝒔𝒊𝒕𝒚; 𝒑𝒓𝒐𝒑𝒆𝒓𝒕𝒚 𝒐𝒇 𝒇𝒍𝒐𝒘. 𝑰𝒕 𝒊𝒔 𝒏𝒐𝒕 𝒂 𝒑𝒉𝒚𝒔𝒊𝒄𝒂𝒍 𝒑𝒓𝒐𝒑𝒆𝒓𝒕𝒚 𝒐𝒇 𝒇𝒍𝒖𝒊𝒅)

̅𝒙
𝝏𝝆𝒗
So that 𝝉𝒕𝒐𝒕𝒂𝒍 = −(𝝂𝒍 + 𝝂𝒕 ) 𝝏𝒚
(𝜈 𝑡 ≫ 𝜈 𝑙 )

All three eddy viscosity components in the 𝑥 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 are written as

̅̅̅̅̅̅̅
′ 𝑡 𝜕𝑣̅𝑥 ̅̅̅̅̅̅̅
′ 𝑡 𝜕𝑣̅𝑥 ̅̅̅̅̅̅̅
′ 𝑡 𝜕𝑣̅𝑥
𝜏̅𝑥𝑥 = −𝜌(𝑣 𝑥 𝑣𝑥 ′ ) = −𝜇𝑥𝑥 𝜕𝑥
, 𝜏̅𝑦𝑥 = −𝜌(𝑣 𝑦 𝑣𝑥 ′ ) = −𝜇𝑦𝑥 𝜕𝑦
, 𝜏̅𝑧𝑥 = −𝜌(𝑣 𝑧 𝑣𝑥 ′ ) = −𝜇𝑧𝑥 𝜕𝑧
, etc.
̅)
𝝏(𝝆𝑪𝒑 𝑻
Similarly, 𝑱𝒕 𝒉 = −𝜶𝒕 ̅̅̅̅̅̅
= −𝝆𝑪𝒑 (𝒗 ′ 𝑻′ ∶ 𝒆𝒏𝒆𝒓𝒈𝒚 (𝒓𝒂𝒕𝒆)𝒄𝒂𝒓𝒓𝒊𝒆𝒅 𝒃𝒚 𝒆𝒅𝒅𝒚 𝒊𝒏 𝒚 𝒅𝒊𝒓𝒆𝒄𝒕𝒊𝒐𝒏
𝒚 )
𝝏𝒚

̅
𝝏𝑪
𝑱𝒕 𝒎 = −𝑫𝒕 ̅̅̅̅̅̅
= −(𝒗 ′ 𝑪′ ∶ 𝒄𝒐𝒏𝒄𝒆𝒏𝒕𝒓𝒂𝒕𝒊𝒐𝒏 (𝒓𝒂𝒕𝒆)𝒄𝒂𝒓𝒓𝒊𝒆𝒅 𝒃𝒚 𝒆𝒅𝒅𝒚 𝒊𝒏 𝒚 − 𝒅𝒊𝒓𝒆𝒄𝒕𝒊𝒐𝒏
𝒚 )
𝝏𝒚
𝜈𝑡 𝜈𝑡
Also, 𝑃𝑟 𝑡 = 𝛼𝑡 ; 𝑆𝑐 𝑡 = 𝐷𝑡 (analogy with the laminar cases)

Thus, energy and species balance equations derived for laminar flow are also time-averaged for
turbulent flows, resulting in an extra source term for eddy transport for energy/species:
𝐷𝑇̅ ̅)
𝝏(𝝆𝑪𝒑 𝑻 𝐷𝐶̅ ̅)
𝝏(𝑪
𝜌𝐶𝑝 = ∇(𝜶𝒍 + 𝜶𝒕 ) ; = ∇(𝑫𝒍 + 𝑫𝒕 )
𝐷𝑡 𝝏𝒚 𝐷𝑡 𝝏𝒚

̅̅̅′ = 0, 𝐶
where, time-averaged quantities are defined: 𝑇 = 𝑇̅ + 𝑇 ′ and 𝐶 = 𝐶̅ + 𝐶 ′ , such that 𝑇 ̅̅̅′ = 0, and
there is an extra source term for energy and concentration in the conservation equations.
k-𝝐 model (two parameters model):
̅̅̅̅̅
𝑣 ′2 𝑘2
𝑘= ; 𝜖 = 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑘; by dimensional analysis 𝜈 𝑡 = 𝑐 (𝑐 = 0.009)
2 𝜖

Prandtl’s Mixing Layer Theory:

- Reynold stresses expressed in terms of the mean velocity by defining a characteristic length scale of
the turbulences, called as the Prandtl’s mixing length (𝑙), which is the length of the travel of an eddy
before mixing itself with a neighboring eddy; a concept/approach analogous to ′𝜆′ (mean free path)
used in the kinetic theory.
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑖𝑒𝑠:
𝑙{ 2. 𝑉̅𝑥 (𝑦 + 𝑙) 𝜕𝑉̅
Δ𝑉̅ = 𝑉̅𝑥 (𝑦) − 𝑉̅𝑥 (𝑦 − 𝑙) = 𝑙 ( 𝜕𝑦𝑥 )
𝑦 𝑙{ 𝑉̅𝑥 (𝑦) } 𝑥,1 1
1. 𝑉̅𝑥 (𝑦 − 𝑙) Δ𝑉̅𝑥,2 = 𝑉̅𝑥 (𝑦 + 𝑙) − 𝑉̅𝑥 (𝑦) = 𝑙 (𝜕𝑉̅𝑥 )
𝜕𝑦 2

1 ̅
𝜕𝑉
Therefore, |𝑉𝑥 ′| = 2 (|Δ𝑉̅𝑥 |1 + |Δ𝑉̅𝑥 |2 ) = 𝑙 ( 𝜕𝑦𝑥 ) (definition for fluctuating velocity component)
Assuming isotropic;
̅ ̅ 𝟐
𝜕𝑉 ′ 𝑽′ = −𝝆𝒍𝟐 (𝝏𝑽𝒙 )
̅̅̅̅̅̅̅
|𝑉𝑦 ′| = 𝑐|𝑉𝑥 ′| = 𝑐𝑙 ( 𝜕𝑦𝑥 ) 𝑹𝒆𝒚𝒏𝒐𝒍𝒅 𝒔𝒕𝒓𝒆𝒔𝒔𝒆𝒔 𝝆𝑽 𝒙 𝒚 𝝏𝒚
̅
𝑑𝑉 ̅𝑥
𝑑𝑉
𝜏𝑡 ≡ −𝜌𝑙 2 | 𝑑𝑦𝑥 | 𝑑𝑦
(easy to work, one parameter (𝑙) 𝑚𝑜𝑑𝑒𝑙)

111
 1
𝐹𝑜𝑟 𝑅𝑒 > 20,000 𝑉 + = 3.8 + 0.36 𝑙𝑛𝑠 + ; 𝑠 + > 26 (experimental measurements)

In laminar sub-layer 𝑽+ = 𝒔+ (velocity profile is linear, same as in the laminar flow case). Using these
equations, a universal velocity profile is plotted for tubular turbulent flow, including the buffer layer (see
BSL for the universal velocity profile plot).

Takeaway message: There is a viscous sub-layer near the tube wall even in the turbulent flow, which is
consistent with the BL postulate that at least one viscous term must be retained in the BL equation, no
matter how high is Reynolds number or how small the fluid viscosity is. As we move away from the surface,
the fluctuating velocity components increase rapidly and the Reynolds stresses are much larger than the
viscous stresses.

We conclude the 2nd part of the course and take up some prototype examples in the next lectures. It is
clear that apart from the first introductory lectures 1-7 covered on the basics of transport phenomena for
an incompressible Newtonian fluid, mainly setting up conservation equations, the next course materials
can be broadly categorized into two types of flows: one at low Reynolds numbers and the other at high
Reynolds numbers. A schematic representation of the different situations covered in this course can be
described as follows:
Transport Phenomena (momentum transport?)

Low Reynolds number High Reynolds number

 Laminar or streamline flows
 Viscose term in the conservation equation is retained.
 In some cases, inertial term may be neglected.
 Common examples are Stoke’s first problem
Start-up tubular flow, Poiseuille’s flow,
Nusselt and Graetz-Nusselt problems,
Free convection in confined flow, Mass transport considering
diffusion in a stagnant fluid, Surface renewal theory,
Creeping flow or Stoke’s law.
 Often exact (analytical) solutions

Potential theory Boundary layer theory Turbulent flow

(feature of fluid-flow)
complementary
Keywords (buzz):
⃗⃗ = 0, Euler equation, Bernoulli’s
Potential theory: potential flow, inviscid fluid, µ → 0, irrotational flow, ∇.τ = 0, ∇ × 𝑉
equation, source/sink, doublet, vortex, D’Alembert’s paradox, flow away from the solid surface.
Boundary layer theory: Prandtl’s boundary layer, drag, Blasius solution, boundary layer thickness, momentum
displacement, drag coefficient, skin friction, thermal and mass boundary layers, heat and mass transfer coefficients,
bluff bodies, boundary layer separation, higher order boundary layers, streamline bodies, air foil.
Turbulent flow (flow feature): fluctuations in velocity, irregularities in flow, eddies, wakes, large mixing rate, time-
averaged quantities, Reynolds stresses, eddy diffusivities, Prandtl’s mixing length, universal velocity profiles, Law of
the wall.

113
 Lecture #23
Prototype example 13: Reduction of BL convection – diffusion equation (for a
flow over solid surface) to heat conduction (Levich: Physicochemical Hydrodynamics)
Emphasis is on BL: Liquids ( 𝑆𝑐 𝑖𝑠 ℎ𝑖𝑔ℎ ~ 1000)

𝜕𝐶 𝜕𝐶 𝜕2 𝐶
𝑣𝑥 𝜕𝑥 + 𝑣𝑦 𝜕𝑦 = 𝐷 𝜕𝑦2 ∶ 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝐵𝐿

𝝏𝟐 𝑪 𝝏𝟐 𝑪 𝝏𝑪 𝝏𝑪
𝑵𝒐𝒕𝒆: 𝝏𝒙𝟐
≪ 𝝏𝒚𝟐 , and 𝒗𝒚 ≪ 𝒗𝒙 𝐛𝐮𝐭 𝝏𝒚
≫ 𝝏𝒙

𝜕𝑣𝑥 𝜕𝑣𝑦
+ = 0 ∶ 𝐶𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦
𝜕𝑥 𝜕𝑦

Replace independent variable (𝑥, 𝑦) by (𝑥, 𝜓), 𝑤ℎ𝑒𝑟𝑒 𝜓 = 𝜓(𝑥, 𝑦)

𝜕𝐶(𝑥,𝑦) 𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶 𝜕𝐶
⇒ | =( ) + | =( ) − | 𝑣
𝜕𝑥 𝑦 𝜕𝑥 𝜓 𝜕𝜓 𝑥 𝜕𝑥 𝜕𝑥 𝜓 𝜕𝜓 𝑥 𝑦

𝜕𝜓 𝜕𝜓
(Recall: 𝑣𝑥 = 𝜕𝑦
, 𝑣𝑦 = − 𝜕𝑥 )

𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶
| = | | = 𝑣𝑥 |
𝜕𝑦 𝑥 𝜕𝜓 𝑥 𝜕𝑦 𝑥 𝜕𝜓 𝑥

𝜕2 𝐶(𝑥,𝑦) 𝜕𝜓 𝜕 𝜕𝐶 𝜕 𝜕𝐶
𝜕𝑦 2
= ∙ (𝑣𝑥 𝜕𝜓)
𝜕𝑦 𝜕𝜓
= 𝑣𝑥 𝜕𝜓 (𝑣𝑥 𝜕𝜓)

Substitute in BL equation:

𝜕𝐶 𝜕𝐶 𝜕𝐶 𝜕 𝜕𝐶
𝑣𝑥 (𝜕𝑥 ) − 𝑣𝑥 𝑣𝑦 𝜕𝜓| + 𝑣𝑥 𝑣𝑦 𝜕𝜓| = 𝐷𝑣𝑥 𝜕𝜓 (𝑣𝑥 𝜕𝜓)
𝜓 𝑥 𝑥

𝐶 = 𝐶𝑜 𝜓 → ∞ (1)
𝜕𝐶 𝜕 𝜕𝐶
|
𝜕𝑥 𝜓
= 𝐷 𝜕𝜓 (𝑣𝑥 𝜕𝜓) 𝐵𝐶𝑠. {𝐶 = 0 𝜓 → 0 (2)
𝐶 = 𝑓𝑖𝑛𝑖𝑡𝑒 𝑥 → 0 (3)

Note: BC. 2 implies 𝜓 = 0 𝑜𝑟 𝑣𝑥 = 𝑣𝑦 = 0 𝑎𝑡 𝑦 = 0 (𝑁𝑜 − 𝑠𝑙𝑖𝑝 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

BC. 3 implies that ′𝐶′ has no singularity at the entrance of the fluid over the flat surface

𝑻𝒉𝒊𝒔 𝒆𝒒𝒖𝒂𝒕𝒊𝒐𝒏 𝒊𝒔 𝒂𝒏𝒂𝒍𝒐𝒈𝒐𝒖𝒔 𝒕𝒐 𝒕𝒉𝒆 𝒉𝒆𝒂𝒕 𝒄𝒐𝒏𝒅𝒖𝒄𝒕𝒊𝒐𝒏

{ 𝒆𝒒𝒖𝒂𝒕𝒊𝒐𝒏 𝒘𝒊𝒕𝒉 K 𝒂𝒔 𝒕𝒉𝒆𝒓𝒎𝒂𝒍 𝒄𝒐𝒏𝒅𝒖𝒄𝒕𝒊𝒗𝒊𝒕𝒚 𝒂𝒏𝒅 𝒂 𝒇𝒏 𝒐𝒇 𝒙 𝒂𝒏𝒅 𝒕:
𝝏𝑻 𝝏 𝝏𝑻
𝝏𝒕
= 𝝏𝒙 [𝑲(𝒙, 𝒕) 𝝏𝒙] (𝟐𝒏𝒅 𝒍𝒂𝒘 𝒐𝒇 𝑭𝒐𝒖𝒓𝒊𝒆𝒓′ 𝒔 𝒉𝒆𝒂𝒕 𝒆𝒒𝒖𝒂𝒕𝒊𝒐𝒏): 𝑺𝒆𝒆 𝒕𝒉𝒆 𝒃𝒐𝒐𝒌 𝒃𝒚 𝑪𝒂𝒓𝒔𝒍𝒂𝒘 𝒂𝒏𝒅 𝑱𝒂𝒆𝒈𝒆𝒓

Let us revert to diffusion equation for the flow over a surface, and pay attention to the region close to the
surface where concentration changes within the BL.

114
Recall the hydrodynamic BL solutions of Blasius:

𝑣𝑜 𝑦 𝜈𝑦 2
𝑣𝑥 = ; 𝑣𝑦 = 3 (close to plate)
𝛿𝑚 𝛿𝑚

𝟏
𝑣𝑥 = 𝟐 𝑣𝑜 𝑓′(𝜂)
Also, for small values of ψ ∶
𝜓 = √𝜈𝑣𝑜 𝑥 𝑓(𝜂)
𝑣
𝜂 = √𝜈𝑥𝑜 𝑦
}

𝜓 𝜂2
Also, Blasius series: 𝑓 = =𝛼 +⋯
√𝜈𝑣𝑜 𝑥 2

1 1 2𝜓
𝑓 ′ = 𝛼𝜂 ⇒ 𝑣𝑥 = 𝑣𝑜 𝛼𝜂 = 𝑣𝑜 𝛼√ (𝜈𝑣𝑜 𝑥)−1/4
2 2 𝛼

Diffusion equation:

𝜕𝐶 𝜕 1 2𝜓 𝜕𝐶
|
𝜕𝑥 𝜓
= 𝐷 𝜕𝜓 (2 𝑣𝑜 𝛼√ 𝛼 (𝜈𝑣𝑜 𝑥)−1/4 𝜕𝜓) (𝑣𝑥 𝑖𝑠 𝑟𝑒𝑝𝑎𝑙𝑐𝑒𝑑)

1 2 𝜕 𝜕𝐶
= 𝐷 2 𝑣𝑜 𝛼√𝛼 (𝜈𝑣𝑜 𝑥)−1/4 𝜕𝜓
(𝜓1/2 𝜕𝜓)

Therefore,
1/2 3/4
1
𝜕𝐶 𝜕 𝜕𝐶 𝐷(𝛼⁄2) 𝑣𝑜
𝑥 4 𝜕𝑥 | = 𝛽 𝜕𝜓 (√𝜓 𝜕𝜓) 𝑤ℎ𝑒𝑟𝑒, 𝛽 = 𝜈 1/4
𝜓

𝑣𝑜
𝑀𝑜𝑚𝑒𝑛𝑡𝑢𝑚
𝑣𝑥 𝛿𝑚
𝑣𝑦
𝐶𝑜 𝐶𝑜𝑛𝑐𝑒𝑛𝑟𝑎𝑡𝑖𝑜𝑛
𝛿𝑐 ≪ 𝛿𝑚 𝑆𝑐 ↑↑

𝐵𝐶𝑠. 𝐶 = 𝐶𝑜 𝜓 → ∞
𝐶=0 𝜓→0
𝐶 = 𝐶𝑜 𝑥 → 0

(Note: Semi-infinite domain & one 𝐵𝐶 𝑜𝑛 𝑥 collapses with the other 𝐵𝐶 𝑜𝑛 𝜓 )

- Similarity Transform ( 𝑜𝑟 𝑐𝑜𝑚𝑏𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒)

1
′ 𝜓2
𝜂 = (𝑎𝑛𝑜𝑡ℎ𝑒𝑟 𝜂 ? ) = 1 1 1 (But, BCs are different)
(𝛼⁄2)2 (𝜈𝑣𝑜 )4 𝑥 4
𝜂 ′ = 𝜂′(𝜓, 𝑥)

115
𝜕𝐶 𝜕𝐶 𝜕𝜂′ 𝜓1/2 1 𝑑𝐶
= 𝜕𝜂′ 𝜕𝑥 = − 4𝑥5/4 ∙ 1/2
𝜕𝑥 (𝛼⁄2) (𝜈𝑣𝑜 )1/4 𝑑𝜂′
𝜕𝐶 𝜕𝐶 𝜕𝜂′ 1 1 1 𝑑𝐶
= = ∙
𝜕𝜓 𝜕𝜂′ 𝜕𝜓 2 𝑥 1/4 𝜓1/2 (𝛼⁄ )1/2 (𝜈𝑣𝑜 )1/4 𝑑𝜂′
2
On substitution,
𝑑2 𝐶 𝛼𝜈 𝑑𝐶
2 + 2 𝐷 𝜂 ′2 =0
𝑑𝜂′ 𝑑𝜂′
Integration,
𝜂 𝛼 𝜈 𝑧3
𝐶 = 𝐶1 ∫0 𝑒𝑥𝑝 (− ) 𝑑𝑧 + 𝐶2
2𝐷 3

Some algebra and 𝐵𝐶 fitting:

𝑣 𝑦2
1/2√ 𝑜
𝜈𝑥 exp(−0.22𝑆𝑐 𝑧 3
𝐶𝑜 ∫0 )𝑑𝑧
𝐶(𝑥, 𝑦) = ∞
∫0 exp(−0.22𝑆𝑐 𝑧 3 )𝑑𝑧

∞ 3 1
∞ 3) ∫0 e−t 𝑑𝑡 3
Γ(1/3) 0.89
∫0 exp(−0.22 𝑆𝑐 𝑧 𝑑𝑧 = (0.22
𝑆𝑐)1/3
= (0.22 𝑆𝑐)1/3
≈ (0.22
𝑆𝑐)1/3

𝑣 𝑦2
(0.22 𝑆𝑐)1/3 𝐶𝑜 1/2√ 𝑜
𝐶(𝑥, 𝑦) = 0.89
∫0
𝜈𝑥
exp(−0.22 𝑆𝑐 𝑧 3 ) 𝑑𝑧

Graphical Solution:
𝑦 𝐶𝑜 𝑣𝑜

𝛿𝑚 𝛿𝑐 ≪ 𝛿𝑚
𝐶(𝑦) 𝑣𝑥 (𝑦)
𝐶
𝛿𝑐 𝑦 𝑣𝑥
𝑥

For a given value of 𝑥, the concentration increases rapidly with increasing 𝑦, and attains, at a
distance 𝛿𝑐 from the plate, a value that is almost (99%) equal to the concentration 𝐶𝑜 in the bulk
of the liquid, where
𝐷 1/3 𝜈𝑥
𝛿𝑐 = 3 ( 𝜈 ) √𝑣
𝑜
𝜕𝐶
𝐹𝑙𝑢𝑥 (𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟) 𝐽𝑚 = 𝐷 (𝜕𝑦)| (moles/s-cm2)
𝑦=0
𝐷𝐶𝑜 (0.22 𝑆𝑐)1/3 𝑣𝑜
𝐽𝑚 = 1.78
√𝜈𝑥
𝐷𝐶𝑜 √𝑣𝑜 𝜈 1/3 𝜈 𝑣𝑜 𝑥
𝐽𝑚 = 0.34 (𝐷 ) (𝑁𝑜𝑡𝑒 𝐷 = 𝑆𝑐, 𝑅𝑒𝑥 = )
√𝜈𝑥 𝜈

116
 𝐶
≈ 𝐷 𝛿𝑜 (𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑠 𝑙𝑖𝑛𝑒𝑎𝑟 𝑛𝑒𝑎𝑟 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒) : from here, mass transfer coefficient can also
𝑐
be estimated)
1 1
𝐷𝐶𝑜 𝐷 3 𝜈𝑥 𝐷 3
Therefore, 𝛿𝑐 = 𝐽𝑚
≈ 3 ( 𝜈 ) √𝑣 = 0.6 ( 𝜈 ) 𝛿𝑚 (same as before)
𝑜

(𝛿𝑚 = 5𝑥⁄ )
√𝑅𝑒𝑥
𝐽𝑚

⇒ 𝐴𝑠𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛𝑠 𝑖𝑛 𝑑𝑒𝑟𝑖𝑣𝑎𝑡𝑖𝑜𝑛: (1) 𝑅𝑒 ≫ 1 (2) 𝑆𝑐 ≫ 1 or 𝛿𝑐 ≪ 𝛿𝑚

⇒ Total amount diffusing to the plate surface can be calculated by integrating:

𝑊𝑡𝑜𝑡𝑎𝑙 (moles/s) = ∫ 𝐽𝑚 𝑑𝑥. 𝑏 (moles/s)

= 0.68 𝐷 𝐶𝑜 𝑏 (𝑆𝑐)1/3 (𝑅𝑒)1/2 (𝑏 ≡ 𝑤𝑖𝑑𝑡ℎ 𝑜𝑓 𝑝𝑙𝑎𝑡𝑒)

𝑣𝑜 𝐿
(𝑅𝑒 = 𝜈
) (𝐿 ≡ 𝑙𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒)

Prototype example 14: Flow over a plate with uniform suction

Consider the flow of an incompressible fluid over a plate with uniform suction, often used to prevent or
delay boundary layer separation, or simply, reduce drag. Such situation (negative y-component velocity)
may also be prevalent in a flow over a porous surface including membranes, at low Reynolds number. In
either case, the amount of fluid removed by suction is small compared to the free stream velocity 𝑈∞ . This
does not mean that it can be totally neglected. Write the boundary layer equations for this flow. What
are the boundary conditions? What is the difference between this problem and the flow over a flat plate
without suction? A simple exact solution can be obtained for the case in which the velocity in the direction
of flow is independent of length. The suction velocity must be of such a magnitude as to make this
assumption correct. Assume steady state and solve for the normally defined boundary layer thickness.

𝑈∞

𝑥
𝑣𝑜

117
 𝐶𝑜 (𝑏𝑢𝑙𝑘 𝑝ℎ𝑎𝑠𝑒)

𝜃=0
𝑠𝑡𝑎𝑔𝑛𝑎𝑛𝑡
𝑝𝑜𝑖𝑛𝑡

𝜃 𝑠𝑖𝑛2𝜃
Let 𝑡 = 𝐷𝑅 2 √3𝑣𝑜 ∫ 𝑠𝑖𝑛2 𝜃 𝑑𝜃 = 𝐷𝑅 2 √3𝑣𝑜 ( 2 − ) + 𝑎1 (? ) (now switch to 𝑪(𝝍, 𝒕))
4
Put in (6)

𝜕𝐶 𝜕 𝜕𝐶
= 𝜕𝜓 (√−𝜓 𝜕𝜓) − (7) : 𝑪(𝝍, 𝒕)
𝜕𝑡

You have a semi-infinite zone; apply again the method of combination of variables,

𝜓
𝜂 = − 𝑡 2/3 (See Levich for arriving at this combination)
𝜓 → 𝜆𝜓′
𝑡 → 𝜇𝑡′
2 𝑑𝐶 𝑑 𝑑𝐶 𝑛 (7)
− 3 𝜂 𝑑𝜂 = 𝑑𝜂 (√𝜂 𝑑𝜂 ) 𝑒𝑞 𝑟𝑒𝑚𝑎𝑖𝑛𝑠
𝑢𝑛𝑐ℎ𝑎𝑛𝑔𝑒𝑑 𝑖𝑓
{ 𝜇 = 𝜆3/2
Introduce 𝑧 = √𝜂
𝑑2 𝐶 4 𝑑𝐶
⇒ 𝑑𝑧 2 + 3 𝑧 2 𝑑𝑧 = 0

𝑧 4
Integrate to derive 𝐶(𝑧) = 𝐶1 ∫0 exp (− 9 𝑧 3 ) 𝑑𝑧 + 𝐶2
1/3
3𝑣𝑜 3𝑣 𝑠𝑖𝑛2𝜃
where, 𝑧 = √ 𝑦𝑠𝑖𝑛𝜃 ⁄[𝐷𝑅 2 √ 4𝑜 (𝜃 − ) + 𝑎1 ]
4 2

BC. 1. → 𝐶2 = 0
𝐶𝑜 𝐶𝑜 𝐶𝑜
2. → 𝐶1 = ∞ 4 = 1 =
∫0 exp(−9𝑧 3 )𝑑𝑧 9 1 1 1.15
( )3 ∙ Γ
4 3 3
(BC. 3 ? ⇒ 𝜃 = 0 𝑜𝑟 𝜃~0 𝑎1 = 0; 𝑖𝑡 𝑖𝑠 𝑛𝑜𝑡 𝑐𝑙𝑒𝑎𝑟 ℎ𝑜𝑤 𝐿𝑒𝑣𝑖𝑐ℎ ℎ𝑎𝑠 𝑑𝑜𝑛𝑒 𝑖𝑡)

3 3𝑣 𝑦 𝑠𝑖𝑛𝜃
Hence, 𝑧 = √4𝐷𝑅𝑜2 𝑠𝑖𝑛2𝜃 1/3 𝐶(𝑦, 𝜃)
(𝜃− )
2
𝐶 𝑧 4
𝑜𝑟 𝐶(𝑟, 𝜃)
∫0 𝑒𝑥𝑝 (− 9 𝑧 3 ) 𝑑𝑧}
𝑜
𝐶 = 1.15

123
  Free stream velocity 𝑈∞
 Due to suction at plate, the velocity in 𝑦 −direction is negative as fluid is being constantly
removed with the suction velocity, 𝑣𝑜 .

Momentum boundary layer equation:

𝜕𝑣𝑥 𝜕𝑣 𝜕𝑣𝑥 1 𝑑𝑝 𝜕2 𝑣𝑥 𝜕2 𝑣𝑥
1) 𝜕𝑡
+ 𝑣𝑥 𝜕𝑥𝑥 + 𝑣𝑦 𝜕𝑦
= − 𝜌 𝑑𝑥 + 𝜈 𝜕𝑥 2
+𝜈 𝜕𝑦 2
(𝑥 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚)
𝜕𝑝
2) 𝜕𝑦
=0 (𝑦 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚)
𝜕𝑣𝑥 𝜕𝑣𝑦
3) Continuity : 𝜕𝑥
+ 𝜕𝑦
=0

BCs: BCs for flow without suction

𝑣𝑥 = 𝑈∞ 𝑦→∞ 𝑣𝑥 = 𝑈∞ 𝑦→∞
𝑣𝑥 = 0 𝑦=0 𝑣𝑥 = 0 𝑦=0
𝑣𝑦 = 0 𝑦→∞ 𝑣𝑦 = 0 𝑦→∞
𝑣𝑦 = −𝑣𝑜 𝑦=0 𝑣𝑦 = 0 𝑦=0

Exact solution:
𝜕𝑣𝑥
Assumptions: Steady state: =0 (1)
𝜕𝑡
𝑑𝑝
Flow over flat plate (semi-infinite zone) 𝑑𝑥
=0 (2)
𝜕𝑣𝑥
Velocity in the direction of flow is independent of length: 𝜕𝑥
~ 0 (3)

(Note that the last term was expectedly non-zero in the Prandtl BL; here suction velocity mitigates the
slowing down of the fluid in x-direction)

𝜕𝑣𝑥 𝜕𝑣𝑦
Continuity: + =0
𝜕𝑥 𝜕𝑦

𝜕𝑣𝑦
= 0 ⇒ 𝑣𝑦 𝑖𝑠 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑎𝑙𝑜𝑛𝑔 𝑦 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛
𝜕𝑦

⇒ 𝑣𝑦 = −𝑣𝑜 (𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒)

𝜕𝑣𝑥 𝜕2 𝑣𝑥
Motion: 𝑣𝑦 𝜕𝑦
=𝜈 𝜕𝑦 2

𝜕𝑣𝑥 𝜕2 𝑣𝑥
⇒ −𝑣𝑜 𝜕𝑦
=𝜈 𝜕𝑦 2

𝜕𝑣𝑥
Put 𝜕𝑦
=𝑧

𝜈𝑜 𝑦
𝜕𝑧
−𝜈𝑜 𝑧 = 𝜈 ⇒ 𝑧 = 𝐴𝑒 − 𝜈
𝜕𝑦

𝜈𝑜 𝑦
𝜕𝑣
𝜕𝑦
= 𝐴𝑒 − 𝜈

118
 𝜈𝑜 𝑦
𝑣𝑥 = 𝐴′ 𝑒 − 𝜈 +𝐵

𝑦 = 0, 𝑣𝑥 = 0 ⇒ 𝐵 = −𝐴′

𝑦 = ∞, 𝑣𝑥 = 𝑈∞ ⇒ 𝐵 = 𝑈∞
𝜈𝑜
𝑣𝑥 = 𝑈∞ (1 − 𝑒 − 𝜈 𝑦 ) (𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑝𝑙𝑎𝑡𝑒)

Boundary Layer Theory: 𝑣𝑥 = 0.99 𝑈∞ at y = 𝛿𝑚

𝜈
−( 𝑜 )𝛿𝑚
⇒ 0.99 𝑈∞ = 𝑈∞ (1 − 𝑒 𝜈 )

𝜈
−( 𝑜 )𝛿𝑚
𝑒 𝜈 = 1 − 0.99 = 0.01
𝜈 4.6𝜈
Or ( 𝜈𝑜 ) 𝛿𝑚 = 4.6 ⇒ 𝛿𝑚 = 𝜈𝑜
: Boundary layer thickness is constant in the case of suction.

5𝑥 𝑥𝜈
Compare with 𝛿𝑚 (𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑠𝑢𝑐𝑡𝑖𝑜𝑛) = = 5√
√𝑅𝑒𝑥 𝑈∞

Diffusion boundary layer thickness (High 𝐨𝐫 𝐥𝐨𝐰 𝐒𝐜)

𝜕𝐶 𝜕𝐶 𝜕2 𝐶 𝜕2 𝐶
𝑣𝑥 + 𝑣𝑦 =𝐷 +𝐷
𝜕𝑥 𝜕𝑦 𝜕𝑦 2 𝜕𝑥 2

𝜕𝐶 𝜕𝐶 𝜕2 𝐶
𝑣𝑥 𝜕𝑥 + 𝑣𝑜 𝜕𝑦 = 𝐷 𝜕𝑦2

𝜕𝐶 𝜕2 𝐶 𝑐 𝑐
[𝑣𝑜 𝜕𝑦] = [𝐷 𝜕𝑦2 ] ⇒ 𝑣𝑜 𝛿𝑐∞ = 𝐷 𝛿𝑐∞2

Therefore, 𝛿𝑐 = 𝐷/𝑣𝑜
𝐷 4.6𝜈
Or, 𝛿𝑐 = 0.217 𝜈 𝛿𝑚 (𝛿𝑚 = 𝜈𝑜
)

= 0.217 𝑆𝑐 −1 𝛿𝑚

𝛿𝑚
𝛿𝑐 ≈ 3 𝑆𝑐 ↑↑ 𝑓𝑜𝑟 𝑙𝑖𝑞𝑢𝑖𝑑𝑠
√𝑆𝑐
𝛿𝑐 ≈ 𝛿𝑚 /𝑆𝑐 Compare { 𝛿𝑚
without suction
≈ 𝑆𝑐 ↓↓ 𝑓𝑜𝑟 𝑔𝑎𝑠𝑒𝑠
√𝑆𝑐

Since, 𝑣𝑦 is same everywhere in flow, net suction 𝛿𝑐 thickness is uniform at high or low Sc.

Similarly, 𝛿𝑡ℎ𝑒𝑟𝑚𝑎𝑙 ≈ 𝛿𝑚 /𝑃𝑟 . Once you have an estimate for boundary layer thicknesses, mass or heat
transfer coefficients can be calculated or estimated: km ≈ 𝐷/𝛿𝑚 and h ≈ 𝑘/𝛿ℎ .

119
 1 𝜕𝜓
𝑣𝜃 = − 𝑟 𝑠𝑖𝑛𝜃 𝜕𝑦
1 𝜕𝜓 𝜕𝜓 𝜕𝜓
≈ − 𝑅 𝑠𝑖𝑛𝜃 𝜕𝑦 close to the surface and ( 𝜕𝑟 = )
𝜕𝑦
3 𝑦
= 2 𝑣𝑜 𝑅 𝑠𝑖𝑛𝜃

√−3𝑣𝑜 𝜓
𝑣𝜃 = − (1) (What about 𝑣𝑟 ? 𝐼𝑡 𝑖𝑠 𝑐𝑎𝑛𝑐𝑒𝑙𝑙𝑒𝑑 𝑜𝑢𝑡. 𝑆𝑒𝑒 𝑙𝑎𝑡𝑒𝑟)
𝑅

Now, the diffusion equation is modified as

𝜕𝐶 𝑣𝜃 𝜕𝐶 𝜕2 𝐶 2 𝜕𝐶 𝜕2 𝐶 2 𝜕𝐶
𝑣𝑟 𝜕𝑟 + = 𝐷 (𝜕𝑟 2 + 𝑟 𝜕𝑟 ) − (2) (for relatively large R: ≫ 𝑟 𝜕𝑟 )
𝑟 𝜕𝜃 𝜕𝑟 2

Transform 𝑪(𝒓, 𝜽) → 𝑪(𝝍, 𝜽) (Recall: in many BL examples, switch from (x, y) to (x, 𝝍))

𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶
= 𝜕𝜓 = 𝜕𝜓 (𝑅𝑠𝑖𝑛𝜃 𝑣𝜃 ) (𝒓 ~𝑹, 𝒘𝒉𝒆𝒏 𝒚𝒐𝒖 𝒔𝒖𝒃𝒔𝒕𝒊𝒕𝒖𝒕𝒆 𝑪(𝒓, 𝜽) 𝒘𝒊𝒕𝒉 𝑪(𝝍, 𝜽))
𝜕𝑟 𝜕𝑟
1 𝜕𝜓 1 𝜕𝜓
as 𝑣𝑟 = − 𝑟 2 𝑠𝑖𝑛𝜃 𝜕𝜃 ; 𝑣𝜃 = + 𝑟𝑠𝑖𝑛𝜃 𝜕𝑟 − (3)

𝜕𝐶 𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶 𝜕𝐶
Therefore, | = 𝜕𝜃| + 𝜕𝜓 𝜕𝜃 = | − 𝑟 2 𝑠𝑖𝑛𝜃𝑣𝑟 𝜕𝜓| − (4)
𝜕𝜃 𝑟 𝜓 𝜕𝜃 𝜓 𝜃

𝜕2 𝐶 𝜕 𝜕𝜓 𝜕𝐶
= 𝜕𝜓 𝜕𝑟 (𝜕𝜓 (𝑅𝑠𝑖𝑛𝜃𝑣𝜃 ))
𝜕𝑟 2
𝜕 𝜕𝐶
= 𝑣𝜃 𝑅 𝑠𝑖𝑛𝜃 (𝑣𝜃 𝑅𝑠𝑖𝑛𝜃 ) − (5)
𝜕𝜓 𝜕𝜓

Plug (3), (4), & (5) in (2)

𝜕𝐶 𝑣𝜃 𝜕𝐶 𝜕𝐶 𝜕 𝜕𝐶
𝑣𝑟 (𝑅𝑠𝑖𝑛𝜃 𝑣𝜃 𝜕𝜓) + (𝜕𝜃 − 𝑅 2 𝑠𝑖𝑛𝜃𝑣𝑟 𝜕𝜓) = 𝐷𝑣𝜃 𝑅𝑠𝑖𝑛𝜃 𝜕𝜓 (𝑣𝜃 𝑅𝑠𝑖𝑛𝜃 𝜕𝜓)
𝑅

𝜕𝐶 𝜕 𝜕𝐶
= 𝐷𝑅 3 𝑠𝑖𝑛2 𝜃 𝜕𝜓 (𝑣𝜃 𝜕𝜓)
𝜕𝜃

Plug 𝑣𝜃 from (1)

𝜕𝐶 𝜕 𝜕𝐶
= 𝐷𝑅 2 𝑠𝑖𝑛2 𝜃√3𝑣𝑜 (√−𝜓 ) − (6) : 𝑪(𝝍, 𝜽)
𝜕𝜃 𝜕𝜓 𝜕𝜓

BC. 1. 𝜓 = 0 𝐶 = 0 ∶ 𝑠𝑢𝑟𝑓𝑎𝑐𝑒
2. 𝜓 → −∞ 𝐶 = 𝐶𝑜 (𝑦 → ∞)
3. 𝜃 = 0 𝜓 = 0 𝐶 = 𝐶𝑜 ∶ 𝑠𝑢𝑟𝑓𝑎𝑐𝑒

The last BC means: at 𝜃 = 0 , where there is the stagnation point, concentration at the surface
is the same as bulk phase concentration. The species have reached the surface by diffusion,
undisturbed by convection.

122
 𝐶𝑜 (𝑏𝑢𝑙𝑘 𝑝ℎ𝑎𝑠𝑒)

𝜃=0
𝑠𝑡𝑎𝑔𝑛𝑎𝑛𝑡
𝑝𝑜𝑖𝑛𝑡

𝜃 𝑠𝑖𝑛2𝜃
Let 𝑡 = 𝐷𝑅 2 √3𝑣𝑜 ∫ 𝑠𝑖𝑛2 𝜃 𝑑𝜃 = 𝐷𝑅 2 √3𝑣𝑜 ( 2 − ) + 𝑎1 (? ) (now switch to 𝑪(𝝍, 𝒕))
4
Put in (6)

𝜕𝐶 𝜕 𝜕𝐶
= 𝜕𝜓 (√−𝜓 𝜕𝜓) − (7) : 𝑪(𝝍, 𝒕)
𝜕𝑡

You have a semi-infinite zone; apply again the method of combination of variables,

𝜓
𝜂 = − 𝑡 2/3 (See Levich for arriving at this combination)
𝜓 → 𝜆𝜓′
𝑡 → 𝜇𝑡′
2 𝑑𝐶 𝑑 𝑑𝐶 𝑛 (7)
− 3 𝜂 𝑑𝜂 = 𝑑𝜂 (√𝜂 𝑑𝜂 ) 𝑒𝑞 𝑟𝑒𝑚𝑎𝑖𝑛𝑠
𝑢𝑛𝑐ℎ𝑎𝑛𝑔𝑒𝑑 𝑖𝑓
{ 𝜇 = 𝜆3/2
Introduce 𝑧 = √𝜂
𝑑2 𝐶 4 𝑑𝐶
⇒ 𝑑𝑧 2 + 3 𝑧 2 𝑑𝑧 = 0

𝑧 4
Integrate to derive 𝐶(𝑧) = 𝐶1 ∫0 exp (− 9 𝑧 3 ) 𝑑𝑧 + 𝐶2
1/3
3𝑣𝑜 3𝑣 𝑠𝑖𝑛2𝜃
where, 𝑧 = √ 𝑦𝑠𝑖𝑛𝜃 ⁄[𝐷𝑅 2 √ 4𝑜 (𝜃 − ) + 𝑎1 ]
4 2

BC. 1. → 𝐶2 = 0
𝐶𝑜 𝐶𝑜 𝐶𝑜
2. → 𝐶1 = ∞ 4 = 1 =
∫0 exp(−9𝑧 3 )𝑑𝑧 9 1 1 1.15
( )3 ∙ Γ
4 3 3
(BC. 3 ? ⇒ 𝜃 = 0 𝑜𝑟 𝜃~0 𝑎1 = 0; 𝑖𝑡 𝑖𝑠 𝑛𝑜𝑡 𝑐𝑙𝑒𝑎𝑟 ℎ𝑜𝑤 𝐿𝑒𝑣𝑖𝑐ℎ ℎ𝑎𝑠 𝑑𝑜𝑛𝑒 𝑖𝑡)

3 3𝑣 𝑦 𝑠𝑖𝑛𝜃
Hence, 𝑧 = √4𝐷𝑅𝑜2 𝑠𝑖𝑛2𝜃 1/3 𝐶(𝑦, 𝜃)
(𝜃− )
2
𝐶 𝑧 4
𝑜𝑟 𝐶(𝑟, 𝜃)
∫0 𝑒𝑥𝑝 (− 9 𝑧 3 ) 𝑑𝑧}
𝑜
𝐶 = 1.15

123
 𝜕𝐶 𝐷𝐶𝑜 3 3𝑣𝑜 𝑠𝑖𝑛𝜃
Flux, 𝐽 = D | = √ 1
𝜕𝑦 𝑦=0 1.15 4𝐷𝑅 2 𝑠𝑖𝑛2𝜃 3
(𝜃− )
2
1 2 2
It is clear that 𝐽 ∝ 𝑣𝑜3 , 𝐷3 , 𝑅 −3 , and 𝑓(𝜃)

At 𝜃 = 0 (𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡 𝑝𝑜𝑖𝑛𝑡) 𝑓𝑙𝑢𝑥 is maximum. At 𝜃 = 𝜋 (𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑝𝑜𝑖𝑛𝑡) 𝑓𝑙𝑢𝑥 is zero.

1
𝑠𝑖𝑛2𝜃 3
𝐷𝐶𝑜 1.15(𝜃− ) 3 4𝐷𝑅 2
2
BL 𝛿𝑐 from 𝐽 ~ ⇒ 𝛿𝑐 = √
𝛿𝑐 𝑠𝑖𝑛𝜃 3𝑣𝑜

Diffusion BL:

𝛿𝑐
𝛿𝑐 = 0 𝑎𝑡 𝜃 = 0 𝑎𝑛𝑑 𝛿𝑐 ~𝑅 𝑎𝑡 𝜃 = 180𝑜
𝜋
𝑇𝑜𝑡𝑎𝑙 𝑚𝑜𝑙𝑎𝑟 𝑓𝑙𝑜𝑤𝑟𝑎𝑡𝑒 = ∫ ∫ 𝐽𝑑𝐴 = 2𝜋𝑅 2 ∫0 𝐽 𝑠𝑖𝑛𝜃𝑑𝜃
4
𝐷𝐶𝑜 𝑅 3 3 3𝑣𝑜 𝜋 𝑠𝑖𝑛2 𝜃𝑑𝜃
= √ 2
∙ 2𝜋 ∫0 1
1.15 4𝐷𝑅 𝑠𝑖𝑛2𝜃 3
(𝜃− 2 )
1
= 7.98𝐷2/3 𝑣𝑜3 𝑅 4/3 𝐶𝑜
1
3
= 4𝜋𝐷𝐶𝑜 𝑅 (0.64𝑃𝑒 ) 𝐟𝐨𝐫 𝑷𝒆 ≫ 𝟏 (𝐹𝑟𝑜𝑚 𝐿𝑒𝑣𝑖𝑐ℎ)

For Pe → 0 (stagnant liquid), start from ∇2 𝐶 = 0 to calculate 𝑊 as:

𝑆𝑒𝑒 𝐵𝑆𝐿. 𝑆𝑖𝑚𝑝𝑙𝑦 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒 𝑢𝑠𝑖𝑛𝑔
𝑊 = 4𝜋𝐷𝐶𝑜 𝑅 𝐵𝐶𝑠 𝐶(0) = 0, 𝑎𝑛𝑑
𝑘𝑚 𝑑𝑝 𝐶(∞) = 𝐶𝑜
(𝑆ℎ = = 2)
𝐷
𝑅
{𝑡𝑜 𝑠𝑜𝑙𝑣𝑒 𝐶(𝑟) = 𝐶𝑜 (1 − 𝑟 ) and then calculate 𝐽 = −𝐷∇𝐶 𝑎𝑡 𝑟 = 𝑅

𝑭𝒐𝒓 𝑷𝒆 ~ 𝟏 (𝒂 𝒎𝒂𝒕𝒉𝒆𝒎𝒂𝒕𝒊𝒄𝒂𝒍 𝒏𝒖𝒎𝒃𝒆𝒓 𝒇𝒐𝒓 𝒈𝒂𝒔 − 𝒇𝒍𝒐𝒘? )

1
3
𝑊 = 4𝜋𝐷𝐶𝑜 𝑅 (1 + 0.64𝑃𝑒 ) : Amazing! Compare it to the previous case Pe ≫ 1

Analogous to mass transfer (diffusion) for 𝑆𝑐 ≫ 1 (liquid) from/to spherical particle in creeping
flow, the heat transfer rate (𝑃𝑟 ≫ 1) from a heated sphere of radius R, with a viscous fluid in
creeping flow past the sphere has been calculated in BSL. Therein, the velocity profiles were
calculated as before. However, the solution was sought differently, viz. using Taylor-series
expansion for the velocity fields expressed in terms of interfacial velocity gradient, and then using
the Von Karman boundary layer integral expression. The final expressions for
𝛿𝑇 (𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑙𝑎𝑦𝑒𝑟) and the total heat flow rate are as follows:

124
 1
1 3
1 (𝜋 − 𝜃 + 𝑠𝑖𝑛2𝜃)
𝛿𝑇 = 0.91(2𝑅)(𝑅𝑒𝑃𝑟) 3− 2
𝑠𝑖𝑛𝜃
𝑘
𝜃 = 0.991 (𝜋𝐷2 )(𝑇𝑜 − 𝑇∞ ) (𝐷) (𝑅𝑒𝑃𝑟)1/3

(Here, 𝐷 = 2𝑅, & 𝜃 has been taken from the rear end of the sphere, which is the same as (𝜋 −
𝜃) of the Levich’s solution). Interestingly, but for small differences in the coefficients, and the
assumption 𝑅𝑒 ~1, these solutions are similar to those of the previous solutions for mass transfer
derived by Levich:
1
1
(𝜃− 𝑠𝑖𝑛2𝜃)3 3 4𝐷𝑅 2
2
𝛿𝑐 = 1.15 √ (ℎ𝑒𝑟𝑒, 𝐷 ≡ 𝐷𝑚 )
𝑠𝑖𝑛𝜃 3𝑣𝑜
1
3
𝑊 = 4𝜋𝐷𝐶𝑜 𝑅 (0.64 (𝑅𝑒𝑆𝑐) )

v (2R) ν 2(v R)
(Note: Pe = ReSc = o ν ∙ D = Do ;
𝜃=0 most of the heat transfer occurs in the
𝛿𝑇 incident region where δT is small.)
𝜃=𝜋

𝑣∞ 𝑇∞

125
 𝐹 2 − 𝐺 2 + 𝐹 ′ 𝐻 = 𝐹 ′′ ∶ 𝑟
2𝐹𝐺 + 𝐺 ′ 𝐻 = 𝐺 ′′ ∶ 𝜃
{ ′ ′ ′′
𝐻𝐻 = 𝑃 + 𝐻 ∶ 𝑦
2𝐹 + 𝐻 ′ = 0 ∶ 𝐶𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦

𝜉 = 0 𝐹 = 0, 𝐺 = 1, 𝐻 = 0 𝑣𝑜
BCs: Here, 𝛼 = 𝜈𝜔 (to be determined; 𝒗𝒐 𝐢𝐬 𝐮𝐧𝐤𝐧𝐨𝐰𝐧)
𝜉 → ∞ 𝐹 = 0, 𝐺 = 0, 𝐻 = −𝛼 √

In this case, series solutions are sought. Apparently, 𝐹 & 𝐺 become small 𝑎𝑛𝑑 𝐻 → −𝛼, 𝑎𝑠 𝜉 →
∞. Therefore, the quadratic terms can be neglected. Use linearize equations for large values of 𝜉:

1) 𝐹 ′′ ≈ −𝐹 ′ 𝛼 ⇒ 𝐹 = 𝑒 −𝛼𝜉
2) 𝐺 ′′ ≈ −𝐺 ′ 𝛼 ⇒ 𝐺 = 𝑒 −𝛼𝜉

All it implies is that the asymptotic expansion of 𝐹, 𝐺, 𝑎𝑛𝑑 𝐻 must be exponentials of the form 𝑒 −𝛼𝜉 :

𝐴2 +𝐵2 −2𝛼𝜉 𝐴(𝐴2 +𝐵2 )

𝐹 = 𝐴𝑒 −𝛼𝜉 − 2𝛼 2
𝑒 + 4𝛼4 𝑒 −3𝛼𝜉 +⋯
𝐴, 𝐵 −𝛼𝜉 𝐵(𝐴2 +𝐵2 ) −3𝛼𝜉
𝑎𝑟𝑒 𝑢𝑛𝑘𝑛𝑜𝑤𝑛 𝐺 = 𝐵𝑒 − 12𝛼4 𝑒 +⋯
2𝐴 −𝛼𝜉
{ 𝐻 = −𝛼 + 𝑒 +⋯
𝛼

Similarly, for small values of 𝜉 (at the surface)

𝜉2 1
𝐹 = 𝑎𝜉 − 2
+ 3 𝑏𝜉 3 + ⋯
1
𝐺 = 1 + 𝑏𝜉 + 3 𝑎𝜉 3 + ⋯ See Frobenius method. F and H are 0 at the surface and G is 1.
2 3 1
{ 𝐻 = −𝑎𝜉 + 3 𝜉 + ⋯

The constants 𝐴, 𝐵, 𝑎, 𝑏 and 𝛼 should be chosen so that 𝐹, 𝐺, 𝐻 𝑎𝑛𝑑 𝐹 ′ , 𝐺 ′ , 𝐻′ remain continuous,

when one matches the two series. However, numerical integration yields 𝑎 = 0.51023, 𝑏 =
−0.616, 𝐴 = 0.934, 𝐵 = 1.208, 𝛼 = 0.88447.
𝑣𝑜
Thus, 𝑽𝒚 ≈ −0.89√𝜈𝜔 𝑎𝑠 𝑦 → ∞ (𝛼 = ) = −𝑣𝑜 (suction velocity or replenishment velocity
√𝜈𝜔
of fluid towards disk)
𝜔3 𝜈
𝑽𝒚 ≈ −0.51√ 𝜈 𝑦 2 𝑓𝑜𝑟 𝑦 ≪ √𝜔 (close to the surface) (check the dimensionless variable, H)

Similar calculations for F and G. F, G, H are shown below (check the solutions at the boundary):

𝜈 1.0 −𝐻
𝛿𝑜 = 3.6√ 0.8
𝜔
𝐺
0.2
Levich assumes 𝐹
0.1
2.0
𝑽𝒚 ~ −0.8𝑣𝑜 𝑎𝑡 𝛿𝑜 0
0.5 3.6 4.5
𝜔
𝜉=√ 𝑦
𝜈
130
 From (non-dimensionalized) velocity distribution trends shown in the plot,
𝜈
1. @𝜉 = 3.6, 𝑜𝑟 𝛿𝑜 = 3.6√ , 𝑉𝑦 ~0.8 of the limiting value, 𝑣𝑜 .
𝜔
@ the disk surface, 𝑉𝑦 = 0
2. @𝜉 = 3.6, 𝐺 ≈ 0.05 ⇒ 𝑉𝜃 → 0.05 𝑉𝜃 𝑚𝑎𝑥
@ the disk surface, 𝐺(𝜉) = 1 𝑎𝑛𝑑 𝑉𝜃 = 𝑅𝜔 (𝑓𝑟𝑜𝑚 𝑉𝜃 = 𝑟𝜔 𝐺(𝜉)).

Thus, 𝜹𝒐 defines boundary layer thickness on the disk surface, within which the radial and tangential
velocity components are not zero; only axial component exists beyond that layer.

Streamlines:

𝛿𝑜
𝛿𝑐

𝜈
𝛿𝑜 = 3.6√
𝜔
𝛿𝑐 ≪ 𝛿𝑜
(𝑆𝑐 ≫ 1: 𝑙𝑖𝑞𝑢𝑖𝑑)

Let us revert to the concentration boundary layer, 𝛿𝑐

𝜕𝐶 𝑉𝜃 𝜕𝐶 𝜕𝐶 𝜕2 𝐶 𝜕2 𝐶 1 𝜕𝐶 1 𝜕2 𝐶
𝑉𝑟 𝜕𝑟 + 𝑟 𝜕𝜃
+ 𝑉𝑦 𝜕𝑦 = 𝐷 (𝜕𝑦2 + 𝜕𝑟2 + 𝑟 𝜕𝑟 + 𝑟2 𝜕𝜃2 ) : species balance in the BL

𝑑𝐶 𝑑2 𝐶 𝑉𝑦 (𝑦)
⇒ 𝑉𝑦 =𝐷 ; } It is important to note that Vr ≠ 0 in BL.
𝑑𝑦 𝑑𝑦 2 𝐶𝑦 (𝑦)

BCs 1. 𝐶 = 0 𝑎𝑡 𝑦 = 0 (𝒅𝒊𝒔𝒌 𝒔𝒖𝒓𝒇𝒂𝒄𝒆) (Levich writes that this condition holds good for maximum
diffusional flux to the surface. However, such condition should hold good for the fast kinetics)

2. 𝐶 = 𝐶𝑜 𝑦→∞

Integrating the equation twice,

𝑦 1 𝑡
𝐶 = 𝑎1 ∫0 𝑒𝑥𝑝 {𝐷 ∫0 𝑉𝑦 (𝑧)𝑑𝑧} 𝑑𝑡 + 𝑎2

𝑎2 = 0 𝑓𝑟𝑜𝑚 𝐵𝐶 1
∞ 1 𝑡
BC 2: 𝐶𝑜 = 𝑎1 ∫0 𝑒𝑥𝑝 {𝐷 ∫0 𝑉𝑦 (𝑧)𝑑𝑧} 𝑑𝑡

131
 Lecture #25
Prototype example 16: Mass transfer to a slowly falling (liquid) drop in a vapor or gas.
Alternatively, consider mass transfer to a slowly rising (gas) bubble in a liquid (Levich).

𝑅𝑒 ≪ 1
𝐶1 ( ) 𝛿 𝑏𝑢𝑡 𝑛𝑜 𝛿𝑚
𝐶𝑜
𝑎𝑛𝑑 𝑃𝑒 ≫ 1 𝑐
𝑟
𝑔
𝑎

This example is analogous to the previous example of mass transfer to a falling spherical solid
particle. The difference lies in convective/diffusion at the liquid-vapor/gas interface vs. that at
the fluid-solid interface. The velocity of fluid motion at the stationary solid surface is zero (no-
slip condition with the relative velocity being zero). On the other hand, the liquid within the drop
(or gas within a bubble) is, however, circulating. Therefore, at such vapour-liquid drop interface,
although the normal velocity component is zero, the common tangential velocity in two phases
is non-zero. All it implies is that hydrodynamics must be resolved afresh before solving the
species balance equation in the boundary layer:

𝜕𝐶 1 𝜕𝐶 𝜕2 𝐶
𝑣𝑟 𝜕𝑟 + 𝑣𝜃 𝑟 𝜕𝜃 = 𝐷 𝜕𝑟 2 (1) (𝑠𝑎𝑚𝑒 𝑎𝑠 𝑒𝑥𝑎𝑚𝑝𝑙𝑒 15; 𝑜𝑛𝑒 𝑟𝑎𝑑𝑖𝑎𝑙 𝑡𝑒𝑟𝑚 𝑖𝑠 𝑑𝑟𝑜𝑝𝑝𝑒𝑑)

where, 𝐶 is the concentration of species diffusing from the surface of the drop into the liquid
inside. The BCs are
𝑟 → ∞ 𝐶 = 𝐶𝑜
𝑟=𝑎 𝐶 = 𝐶1
𝜃 = 0 𝑎𝑛𝑑 𝑟 = 𝑎 (𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑜𝑖𝑛𝑡), 𝐶 = 𝐶𝑜

What about 𝑣𝑟 and 𝑣𝜃 ? For spherical coordinate system,

1 𝜕𝜓 1 𝜕𝜓
𝑣𝜃 = − 𝑟 𝑠𝑖𝑛𝜃 𝜕𝑟 ; 𝑣𝑟 = ⃗⃗ = 0)
(∇. 𝑉
𝑟 2 𝑠𝑖𝑛𝜃 𝜕𝜃

We have to substitute ′𝑣𝜃 ′ from hydrodynamic solution, in which case NS equation must be
solved for both phases (outside and inside the drop) and the interfacial velocity and shear stress
should be matched (recall the example of the flow of two immiscible liquids having different
viscosities in a horizontal tube (BSL)). For the present case, see Levich for detailed hydrodynamic
or velocity calculations. Here, we skip the calculations and directly write down the solutions:

126
 𝜓 ≈ −𝑣𝑜 𝑎𝑦 𝑠𝑖𝑛2 𝜃
2(𝜌′ − 𝜌)𝑔𝑎2 𝜇 + 𝜇 ′ } (2) (𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑙 𝑜𝑟 𝑠𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑑𝑟𝑜𝑝, 𝑜𝑟 𝑖𝑡𝑠 𝐶𝑂𝑀)
𝑣=
3𝜇 2𝜇 + 3𝜇′
In this present case, the drop may also be considered motionless, with the coordinate system moving
along with the center of gravity of the falling drop, while the outer liquid moves in the opposite direction
at velocity, −𝑈; 𝑈 being the velocity of the drop’s motion in the inertial frame. The above- expression
passes into the Stoke’s equation for flow past a solid, with 𝜇 ≪< 𝜇′ :

2 (𝜌′ − 𝜌)𝑔𝑎2
𝑣𝑡 = ; (𝑛𝑜𝑡𝑒 𝑣 > 𝑣𝑡 )
9 𝜇

⇒ The liquid at the surface of the drop moves at the tangential velocity (𝒗𝜽 )𝒓=𝒂 = 𝒗𝒐 𝒔𝒊𝒏𝜽,
where, 𝑣𝑜 is the magnitude of the velocity at the drop’s equator:
𝜇 𝑣 (𝜌′ −𝜌)𝑔𝑎2
𝑣𝑜 = = (𝐻𝑒𝑟𝑒, 𝑣 𝑖𝑠 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑎𝑠 𝑖𝑛 𝑒𝑞. 2)
2 𝜇+𝜇′ 3(2𝜇+3𝜇 ′ )

Let us revert and work on 𝜃 & 𝜓 as the variables instead of r and 𝜃 (same as before):

𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶
= 𝜕𝜓 𝜕𝑟 = −𝑟𝑠𝑖𝑛𝜃 𝑣𝜃 (𝜕𝜓)
𝜕𝑟 𝜃

𝜕𝐶 𝜕𝐶 𝜕𝐶 𝜕𝜓 𝜕𝐶 𝜕𝐶
(𝜕𝜃) = (𝜕𝜃) + 𝜕𝜓 𝜕𝜃 = (𝜕𝜃) + 𝑟 2 𝑠𝑖𝑛𝜃𝑣𝑟 (𝜕𝜓)
𝑟 𝜓 𝜓 𝜃

Considering that we are solving 𝐶(𝜃, 𝑟) 𝑜𝑟 𝐶(𝜓, 𝜃) near the surface of the drop:
𝑦 = 𝑟 − 𝑎 , 𝑤ℎ𝑒𝑟𝑒 𝑦 ≪ 𝑎 𝑟
𝑦
𝑎

The species balance equation is transformed to

𝜕𝐶 𝜕2 𝐶
𝜕𝜃
= 𝐷𝑎3 (𝑣𝜃 )𝑦~0 𝑠𝑖𝑛2 𝜃 (𝜕𝜓2 ) − (3) (𝑠𝑎𝑚𝑒 𝑎𝑠 𝑏𝑒𝑓𝑜𝑟𝑒)

𝜕𝐶 𝜕2 𝐶
Substituting 𝑣𝜃 in eq. (3), = 𝐷𝑎3 𝑣𝑜 𝑠𝑖𝑛3 𝜃 (𝜕𝜓2 )
𝜕𝜃

Again, similar to the previous case/prototype example, introduce a new variable

𝑐𝑜𝑠3 𝜃
𝑡 = 𝐷𝑎3 𝑣𝑜 ∫ 𝑠𝑖𝑛3 𝜃𝑑𝜃 = 𝐷𝑎3 𝑣𝑜 ( − 𝑐𝑜𝑠𝜃) + 𝑎1 − (4) (what is 𝑎1 ? )
3

𝜕𝐶 𝜕2 𝐶
Or, = 𝜕𝜓2
𝜕𝑡

𝐵𝐶 1. 𝐶 = 𝐶𝑜 𝜓 → −∞;

127
 2. 𝐶 = 𝐶1 𝜓 → 0;
3. 𝐶 = 𝐶𝑜 𝑎𝑡 𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑠𝑡𝑎𝑔𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑜𝑖𝑛𝑡, 𝑟=𝑎 & 𝜃=0
2
(The last condition implies 𝐶 = 𝐶𝑜 @ 𝑡 = 𝑡𝑜 = 𝑎1 − 3 𝐷𝑎3 𝑣𝑜 )

The solution is
𝜓
−
2(𝐶𝑜 −𝐶1 ) 2√𝑡−𝑡𝑜
𝐶(𝜓, 𝑡) = ∫0 exp(−𝑧 2 ) 𝑑𝑧 + 𝐶1 (𝑎1 disappears mysteriously!)
√𝜋

Substitute the value of 𝜓 𝑎𝑠 (−𝑣𝑜 𝑎𝑦 𝑠𝑖𝑛2 𝜃) and 𝑡 𝑓𝑟𝑜𝑚 (4)

2(𝐶𝑜 −𝐶1 ) 𝑁 𝑎𝑣𝑜 𝑠𝑖𝑛2 𝜃𝑦

𝐶= ∫0 exp(−𝑧 2 ) 𝑑𝑧 + 𝐶1 𝑤ℎ𝑒𝑟𝑒, 𝑁 = 𝑐𝑜𝑠3 𝜃
√𝜋 2
2√𝐷𝑣𝑜 𝑎3 ( −𝑐𝑜𝑠𝜃+ )
3 3
Calculate flux,
1
𝜕𝐶 𝐷𝑣𝑜 𝑎𝑠𝑖𝑛2 𝜃(𝐶𝑜 −𝐶1 ) 𝐷𝑣 2 3 (1+𝑐𝑜𝑠𝜃)2 𝐷(𝐶𝑜 −𝐶1 )
𝐽𝑚 = 𝐷 (𝜕𝑦 ) = = ( 𝑎 𝑜 ) √𝜋 √ 2+𝑐𝑜𝑠𝜃 (𝐶𝑜 − 𝐶1 ) = 𝛿𝑐
𝑦=0 2 𝑐𝑜𝑠3 𝜃
2√𝜋𝐷𝑣𝑜 𝑎3 (3−𝑐𝑜𝑠𝜃+ 3 )
1
𝜋 𝑎𝐷 2 2+𝑐𝑜𝑠𝜃
where, 𝛿𝑐 𝑖𝑠 the mass BL thickness = √3 ( 𝑣 ) √(1+𝑐𝑜𝑠𝜃)2
𝑜

1 𝐻𝑖𝑔ℎ𝑒𝑠𝑡 𝑓𝑙𝑢𝑥 𝑖𝑠 𝑎𝑡 𝜃 = 0 (𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 𝑖𝑛𝑐𝑖𝑑𝑒𝑛𝑡)

Thus, 𝐽𝑚 ∝ 𝑣𝑜2 𝑤ℎ𝑒𝑟𝑒 𝑡ℎ𝑒 𝑑𝑟𝑜𝑝 𝑒𝑛𝑐𝑜𝑢𝑛𝑡𝑒𝑟𝑠 𝑓𝑟𝑒𝑠ℎ 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
1
−
∝𝑎 2 @ 𝜃 = 𝜋, 𝐽 𝑖𝑠 𝑠𝑚𝑎𝑙𝑙𝑒𝑠𝑡 (0)(𝑎𝑙𝑠𝑜, 𝑚𝑎𝑡ℎ𝑒𝑚𝑎𝑡𝑖𝑐𝑎𝑙𝑙𝑦
∝ 𝑓(𝜃)} 𝑡ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑠𝑖𝑛𝑔𝑢𝑙𝑎𝑟𝑖𝑡𝑦 𝑎𝑡 𝛿𝑐 → ∞)

𝐷(𝐶𝑜 −𝐶1 )
Note: The above-solution is valid when 𝑦 ≪ 𝑎. As an approximation, 𝐽|𝑟=𝑎 = 𝑎

𝜋
Total 𝑊(𝑚𝑎𝑠𝑠 𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒) = 2𝜋𝑎2 ∫0 𝐽 𝑠𝑖𝑛𝜃𝑑𝜃 (integrated over the surface)
1
𝐷𝑣 2 𝜋 (1+𝑐𝑜𝑠𝜃)
= 2√3𝜋𝑎2 ( 𝑜 ) ∫0 𝑠𝑖𝑛𝜃 𝑑𝜃 (𝐶𝑜 − 𝐶1 )
𝑎 √2+𝑐𝑜𝑠𝜃
1
𝜋 𝐷𝑣𝑜 2
= 8(𝐶𝑜 − 𝐶1 )√ 3 ( 𝑎
) 𝑎2
𝑣𝑜 𝑎 𝜈 𝑊𝑎
Introducing 𝑅𝑒 = , 𝑆𝑐 = 𝐷 , 𝑆ℎ (𝑆ℎ𝑒𝑟𝑤𝑜𝑜𝑑 #) = 4𝜋𝑎2 𝐷(𝐶 (how?),
𝜈 𝑜 −𝐶1 )
1 1
2 𝜇 2
𝑆ℎ = (𝑃𝑒 2 ) (𝜇+𝜇′ ) ; 𝑃𝑒 = 𝑅𝑒𝑆𝑐
√6𝜋
Also, compare 𝑊 with that for the diffusion to a solid spherical particle (previous example) under similar
1 1
3 3
condition (𝑠𝑎𝑚𝑒 𝑅𝑒): 𝑊 ∝ 𝑃𝑒 𝑜𝑟 𝑅𝑒 . The diffusional flow to the drop is greater than the solid
particle. Why? Because there is a sink in drop?? No.

⇒ At high 𝑹𝒆, complexities in determining the dependence of 𝑊 𝑜𝑟 𝑆ℎ on 𝑃𝑒 arise because of wake

formation downstream of the drop, or the deformation of drop to a non-spherical shape. By dimensionless
𝐷(𝐶𝑜 −𝐶1 )4𝜋𝑎 2
analysis, it can, however, be shown that 𝑊 ≈ at high Re.
√𝑃𝑒

128
 Lecture #26
Prototype example 17: Convection/diffusion to the surface of a circular rotating disk (Levich)

Note: If there is no rotation, then we would have been discussing potential flow of an
axisymmetric cylindrical jet striking a horizontal surface. Thus, boundary layer (momentum
and/or mass) over the disk surface arises in this case because of rotation of the disk. Also, fluid
acquires radial velocity under the influence of the centrifugal force, which in turn induces negative Y-
direction flow towards the disk, as a replenishment.

𝑉𝑟 = 𝑉𝜃 = 0
𝑉𝑦 = −𝑣0 } 𝑦 → ∞
𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑟𝑒𝑝𝑙𝑒𝑛𝑠ℎ𝑚𝑖𝑛𝑡 𝑜𝑓
𝜔 ↑ 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑙𝑎𝑦𝑒𝑟 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 ↓ 𝑦 }
𝑓𝑙𝑢𝑖𝑑 𝑡𝑜𝑤𝑎𝑟𝑑𝑠 𝑠𝑢𝑟𝑓𝑎𝑐𝑒
𝛿𝑜
𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 ↑
{ ′𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 ′ ↑ 𝑟 𝐵𝐶𝑠 @𝑦 = 0
𝑁𝑜𝑡𝑒:
𝑉𝑟 = 0 𝑉𝑟 = 0 𝑜𝑛 𝑑𝑖𝑠𝑘
⇒ No edge effect (infinitely wide disk) 𝜔 𝑉 = 𝜔𝑟
{ 𝜃 ≠ 0 𝑎𝑤𝑎𝑦 𝑓𝑟𝑜𝑚 𝑑𝑖𝑠𝑘
𝑉𝑦 = 0 {
⇒ Exact solution of the hydrodynamic = 0 𝑣𝑒𝑟𝑦 𝑓𝑎𝑟
𝑎𝑤𝑎𝑦
equations has been obtained by Von Korman and Cochran.

𝜕
Solution: NS equation for BL over disk surface (assume axial symmetry, 𝜕𝜃 = 0 (𝑛𝑜 𝜃 − 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒)

𝜕𝑉𝑟 𝑉2 𝜕𝑉 1 𝜕𝑝 𝜕2 𝑉 𝜕2 𝑉 1 𝜕𝑉𝑟 𝑉
𝑉𝑟 − 𝜃 + 𝑉𝑦 𝑟 = − + 𝜈 ( 2𝑟 + 2𝑟 + − 𝑟2 ) : 𝑟 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚
𝜕𝑟 𝑟 𝜕𝑦 𝜌 𝜕𝑟 𝜕𝑦 𝜕𝑟 𝑟 𝜕𝑟 𝑟
𝜕𝑉 𝑉𝑉 𝜕𝑉 𝜕2 𝑉 𝜕2 𝑉 1 𝜕𝑉𝜃 𝑉
𝑉𝑟 𝜃 − 𝑟 𝜃 + 𝑉𝑦 𝜃 = 𝜈 ( 2𝜃 + 2𝜃 + − 2𝜃 ) ∶ 𝜃 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚
𝜕𝑟 𝑟 𝜕𝑦 𝜕𝑦 𝜕𝑟 𝑟 𝜕𝑟 𝑟
𝜕𝑉𝑦 𝜕𝑉𝑦 1 𝜕𝑝 𝜕2 𝑉𝑦 𝜕2 𝑉𝑦 1 𝜕𝑉𝑦
{ 𝑉𝑟 𝜕𝑟 + 𝑉𝑦 𝜕𝑦 = − 𝜌 𝜕𝑦 + 𝜈 ( 𝜕𝑦2 + 𝜕𝑟2 + 𝑟 𝜕𝑟 ) : 𝑦 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚

𝜕𝑉𝑟 𝑉𝑟 𝜕𝑉𝑦
𝜕𝑟
+ 𝑟
+ 𝜕𝑦
= 0: Continuity

𝜕𝑝
(𝜕𝑟 ≈ 0: no ∇p along r) (As noted above, fluid acquires 𝑉𝑟 because of the centrifugal force)

Assume a similarity solution:

𝝎
𝑉𝑟 = 𝑟𝜔𝐹(𝜉) 𝝃 = √ 𝝂 𝒚 (𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑙𝑒𝑠𝑠)
𝑉𝜃 = 𝑟𝜔𝐺(𝜉)
𝑉𝑦 = √𝜈𝜔 𝐻(𝜉)
{ 𝑝 = −𝜌𝜈𝜔𝑃(𝜉) (𝑝 ~𝜌𝑉𝑦 2 )
Plug into NS equations, and Continuity:

129
 𝐹 2 − 𝐺 2 + 𝐹 ′ 𝐻 = 𝐹 ′′ ∶ 𝑟
2𝐹𝐺 + 𝐺 ′ 𝐻 = 𝐺 ′′ ∶ 𝜃
{ ′ ′ ′′
𝐻𝐻 = 𝑃 + 𝐻 ∶ 𝑦
2𝐹 + 𝐻 ′ = 0 ∶ 𝐶𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦

𝜉 = 0 𝐹 = 0, 𝐺 = 1, 𝐻 = 0 𝑣𝑜
BCs: Here, 𝛼 = 𝜈𝜔 (to be determined; 𝒗𝒐 𝐢𝐬 𝐮𝐧𝐤𝐧𝐨𝐰𝐧)
𝜉 → ∞ 𝐹 = 0, 𝐺 = 0, 𝐻 = −𝛼 √

In this case, series solutions are sought. Apparently, 𝐹 & 𝐺 become small 𝑎𝑛𝑑 𝐻 → −𝛼, 𝑎𝑠 𝜉 →
∞. Therefore, the quadratic terms can be neglected. Use linearize equations for large values of 𝜉:

1) 𝐹 ′′ ≈ −𝐹 ′ 𝛼 ⇒ 𝐹 = 𝑒 −𝛼𝜉
2) 𝐺 ′′ ≈ −𝐺 ′ 𝛼 ⇒ 𝐺 = 𝑒 −𝛼𝜉

All it implies is that the asymptotic expansion of 𝐹, 𝐺, 𝑎𝑛𝑑 𝐻 must be exponentials of the form 𝑒 −𝛼𝜉 :

𝐴2 +𝐵2 −2𝛼𝜉 𝐴(𝐴2 +𝐵2 )

𝐹 = 𝐴𝑒 −𝛼𝜉 − 2𝛼 2
𝑒 + 4𝛼4 𝑒 −3𝛼𝜉 +⋯
𝐴, 𝐵 −𝛼𝜉 𝐵(𝐴2 +𝐵2 ) −3𝛼𝜉
𝑎𝑟𝑒 𝑢𝑛𝑘𝑛𝑜𝑤𝑛 𝐺 = 𝐵𝑒 − 12𝛼4 𝑒 +⋯
2𝐴 −𝛼𝜉
{ 𝐻 = −𝛼 + 𝑒 +⋯
𝛼

Similarly, for small values of 𝜉 (at the surface)

𝜉2 1
𝐹 = 𝑎𝜉 − 2
+ 3 𝑏𝜉 3 + ⋯
1
𝐺 = 1 + 𝑏𝜉 + 3 𝑎𝜉 3 + ⋯ See Frobenius method. F and H are 0 at the surface and G is 1.
2 3 1
{ 𝐻 = −𝑎𝜉 + 3 𝜉 + ⋯

The constants 𝐴, 𝐵, 𝑎, 𝑏 and 𝛼 should be chosen so that 𝐹, 𝐺, 𝐻 𝑎𝑛𝑑 𝐹 ′ , 𝐺 ′ , 𝐻′ remain continuous,

when one matches the two series. However, numerical integration yields 𝑎 = 0.51023, 𝑏 =
−0.616, 𝐴 = 0.934, 𝐵 = 1.208, 𝛼 = 0.88447.
𝑣𝑜
Thus, 𝑽𝒚 ≈ −0.89√𝜈𝜔 𝑎𝑠 𝑦 → ∞ (𝛼 = ) = −𝑣𝑜 (suction velocity or replenishment velocity
√𝜈𝜔
of fluid towards disk)
𝜔3 𝜈
𝑽𝒚 ≈ −0.51√ 𝜈 𝑦 2 𝑓𝑜𝑟 𝑦 ≪ √𝜔 (close to the surface) (check the dimensionless variable, H)

Similar calculations for F and G. F, G, H are shown below (check the solutions at the boundary):

𝜈 1.0 −𝐻
𝛿𝑜 = 3.6√ 0.8
𝜔
𝐺
0.2
Levich assumes 𝐹
0.1
2.0
𝑽𝒚 ~ −0.8𝑣𝑜 𝑎𝑡 𝛿𝑜 0
0.5 3.6 4.5
𝜔
𝜉=√ 𝑦
𝜈
130
 From (non-dimensionalized) velocity distribution trends shown in the plot,
𝜈
1. @𝜉 = 3.6, 𝑜𝑟 𝛿𝑜 = 3.6√ , 𝑉𝑦 ~0.8 of the limiting value, 𝑣𝑜 .
𝜔
@ the disk surface, 𝑉𝑦 = 0
2. @𝜉 = 3.6, 𝐺 ≈ 0.05 ⇒ 𝑉𝜃 → 0.05 𝑉𝜃 𝑚𝑎𝑥
@ the disk surface, 𝐺(𝜉) = 1 𝑎𝑛𝑑 𝑉𝜃 = 𝑅𝜔 (𝑓𝑟𝑜𝑚 𝑉𝜃 = 𝑟𝜔 𝐺(𝜉)).

Thus, 𝜹𝒐 defines boundary layer thickness on the disk surface, within which the radial and tangential
velocity components are not zero; only axial component exists beyond that layer.

Streamlines:

𝛿𝑜
𝛿𝑐

𝜈
𝛿𝑜 = 3.6√
𝜔
𝛿𝑐 ≪ 𝛿𝑜
(𝑆𝑐 ≫ 1: 𝑙𝑖𝑞𝑢𝑖𝑑)

Let us revert to the concentration boundary layer, 𝛿𝑐

𝜕𝐶 𝑉𝜃 𝜕𝐶 𝜕𝐶 𝜕2 𝐶 𝜕2 𝐶 1 𝜕𝐶 1 𝜕2 𝐶
𝑉𝑟 𝜕𝑟 + 𝑟 𝜕𝜃
+ 𝑉𝑦 𝜕𝑦 = 𝐷 (𝜕𝑦2 + 𝜕𝑟2 + 𝑟 𝜕𝑟 + 𝑟2 𝜕𝜃2 ) : species balance in the BL

𝑑𝐶 𝑑2 𝐶 𝑉𝑦 (𝑦)
⇒ 𝑉𝑦 =𝐷 ; } It is important to note that Vr ≠ 0 in BL.
𝑑𝑦 𝑑𝑦 2 𝐶𝑦 (𝑦)

BCs 1. 𝐶 = 0 𝑎𝑡 𝑦 = 0 (𝒅𝒊𝒔𝒌 𝒔𝒖𝒓𝒇𝒂𝒄𝒆) (Levich writes that this condition holds good for maximum
diffusional flux to the surface. However, such condition should hold good for the fast kinetics)

2. 𝐶 = 𝐶𝑜 𝑦→∞

Integrating the equation twice,

𝑦 1 𝑡
𝐶 = 𝑎1 ∫0 𝑒𝑥𝑝 {𝐷 ∫0 𝑉𝑦 (𝑧)𝑑𝑧} 𝑑𝑡 + 𝑎2

𝑎2 = 0 𝑓𝑟𝑜𝑚 𝐵𝐶 1
∞ 1 𝑡
BC 2: 𝐶𝑜 = 𝑎1 ∫0 𝑒𝑥𝑝 {𝐷 ∫0 𝑉𝑦 (𝑧)𝑑𝑧} 𝑑𝑡

131
 𝜔3
In principle the integral should be evaluated from 0 to 𝛿𝑜 𝑤𝑖𝑡ℎ 𝑉𝑦 ≈ 0.51√ 𝜈 𝑦 2 𝑎𝑛𝑑 𝛿𝑜 𝑡𝑜 ∞ 𝑤𝑖𝑡ℎ 𝑉𝑦 ≈
−0.89√𝜈𝜔. Neglecting (turns out to be negligible) the diffusion or concentration within the 2nd integral,

𝒀 𝟑 1
∫𝟎 𝒆−𝒗 𝒅𝒗 𝑦 𝜈 3
𝑪 = 𝑪𝒐 ∞ 𝟑 (See Levich for details; heavy maths), where 𝑌 = 1 1 ≈ 2𝑦/𝛿𝑜 (𝐷)
∫𝟎 𝒆−𝒗 𝒅𝒗 3√6𝐷3 𝜈6
1.00 1
𝜔2
0.75
𝐶
𝐶𝑜 0.5

0.25
0.5 1.0 1.5 𝑦
𝑌=
𝛿𝑐
Diffusion flux to disk:
2 1
𝜕𝐶
𝐽 = 𝐷 𝜕𝑦| = 0.62 𝐷 3 𝜈 −6 𝜔1/2 𝐶𝑜 → 𝐶𝑜 𝑖𝑠 𝑡ℎ𝑒 free stream concentration
𝑦=0

⇒ 𝑪 = 𝟎 @𝒚 = 0 represents fast kinetics.

1 1
𝐷𝐶𝑜 𝐷 3 𝜈 𝐷 3
𝛿𝑐 = 𝐽
= 1.61 ( 𝜈 ) √𝜔 ≈ 0.5 ( 𝜈 ) 𝛿𝑜

𝜈
Thus 𝛿𝑐 ≪ 𝛿𝑜 𝑓𝑜𝑟 𝑆𝑐 ≫ 1 (𝑙𝑖𝑞𝑢𝑖𝑑) (𝛿𝑜 = 3.6√ )
𝜔

Note that 𝜹𝒄 is not a function of 𝒓 ⇒ the thickness of the boundary layer is constant over the entire
disk surface. Compare this situation with the boundary layer flow with suction. Also, the expression
𝟏 𝟏
𝒙
matches well with 𝛿𝑐 for flow over a flat plate 𝜹𝒄 ~𝑫𝟑 𝝂𝟔 √𝒗 , 𝑒𝑥𝑐𝑒𝑝𝑡 for the numerical coefficient.
𝒐

Total mass flow rate of the species

2 1
𝜋𝐷𝐶𝑜 𝑅2
𝑊= 1 ≈ 1.9𝐷 3 𝜈 −6 √𝜔 𝑅2 𝐶𝑜
𝐷 𝜈
1.61( )3 √
𝜈 𝜔
𝐷𝐶𝑜 𝐷
With finite kinetics, say 1st order and surface concentration, Cs, 𝐽 = 1 = 𝛿 (𝐶𝑜 − 𝐶𝑠 )
𝐷 𝜈 𝐷 𝑐
[1.61( 𝜈 )3 √𝜔 +
𝑘
]

(1) There are two rates in series:

𝑘 → ∞: diffusion controlled (fast kinetics) (same as the previous case)
𝑘 → 0 𝐽 ≈ 𝑘𝐶𝑜 (reaction controlled) 𝐶𝑠 → 𝐶𝑜
𝟏
(2) 𝑱 ∝ 𝝎𝟐
𝑐𝑎𝑛 𝑏𝑒 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙𝑙𝑦 𝑣𝑒𝑟𝑖𝑓𝑖𝑒𝑑
𝐽𝑜 𝑎𝑙𝑠𝑜, 𝑎𝑠𝑠𝑢𝑚𝑒 𝑎 𝑢𝑛𝑖𝑓𝑜𝑟𝑚𝑙𝑦
⇒ 𝑎𝑐𝑐𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑠𝑢𝑟𝑓𝑎𝑐𝑒
𝜔𝑅2
1 10 {𝑅𝑒 = 𝜈
~104 − 105 𝑙𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤
𝜔 2

132
 Lecture #27
Example 19: CFD-based transport modelling of a CVD reactor (gas and surface phase reaction (or
adsorption-desorption) in a vertical thermal reactor)

Preamble: At the outset you should have noticed that this example is titled “example”, and not
“prototype example” as titled for the previous 17 examples discussed in this lecture series.

CFD has significantly grown over the past twenty five years to the extent that traditional course
on transport phenomena is gradually giving way to engineering software like Fluent where the
focus is more on application using model simulation results, rather on understanding the
fundamental steps of transport phenomena. You can see a similar situation in the present use of
Aspin vis a vis Treybal’s book on mass transfer at the undergraduate level in chemical
engineering! Or, in the use of NAG or MATLAB libraries and a graduate level course on numerical
methods for solving ODEs or PDEs. While the emphasis on understanding the basics of transport
phenomena can never be undermined, it is good to discuss at least one example on the
applications of CFD, as a software, in chemical engineering. In this last lecture series, we solve
concentration profiles in a vertical thermal reactor stacked with silicon wafers, commonly used
in semiconductor industries, during a common purge cycle. To make the example simple, we
discuss here the purge schedule of a CVD reactor before the reaction is started. The problem
statement goes like this.

Problem statement: Consider a vertical CVD reactor stacked with silicon wafers. Before chemical
processing on the wafers is started, the reactor with the wafers must be purged using a high
purity inert gas (nitrogen or argon) to drive off moisture or any gaseous species left behind in the
reactor from the previous step, or atmospheric gaseous impurities, for example moisture,
intruded into the reactor. Purging is done from the top of the reactor, with the bottom end
closed. The purge gas flows through the annular space between the stack and reactor walls,
sweeping the impurities downwards along with it. From the perspective of chemical engineering,
or mass transfer, the purging step or process may be diffusion-controlled, as the purge gas will
not penetrate the gas space between the wafers because of symmetricity in flow conditions and
a relatively smaller wafer spacing or pitch. Thus, the impurities must diffuse radially outward to
be carried out of the reactor by the downward flowing purge gas in the annular space. We are
interested in developing a practical purge schedule or a mathematical model to predict gas
concentration levels everywhere in the reactor, including annular space and between the
wafers, and both in the gas phase and on the surface of the wafer. While doing so (i.e.,
developing model), we should remain focussed on Transport Phenomena as far as possible. To
this end, typical dimensions and conditions in such a CVD reactor are: wafer size: 2 – 20”
diameter, reactor diameter: 6 – 30”, wafer spacing: ¼ - 1”, number of wafers: 25 – 200, gas

133
Finite Difference-based
Numerical Methods in
Chemical Engineering

1
 Lecture #01
Finite Difference-based Numerical Methods in Chemical Engineering

Recommended Books:

1) Linear Algebra by Gilbert Strang , Cengage, 4th ed.

2) Numerical Methods for Engineers by S.K. Gupta, Wiley Eastern Ltd, 2nd ed.
3) Numerical Methods for Engineering Application by Joel H. Feriziger, Wiley, 2nd ed.

Objective: Learn a few basic numerical techniques (Finite Difference) to solve simple
ODEs/PDEs/algebraic equations derived from a variety of heat and mass transfer related
conservation equations.

 Analytical methods yield exact solutions, often in infinite series; use mathematical functions.
 Numerical methods yield approximate solutions; use numbers, computations; results are dependent
on the method accuracies and limited by computers machine errors.
Course focuses on
 Algebra (Matrix Operation)
 Methods (Finite Difference)
Requires knowledge of
 Computer Programming (because large # of iterations)
(there are engineering software, e.g., Matlab, NAG, IMSL, Polymath)
 No numerical analysis in this course!

Examples: (Applications in chemical engineering)

1. Distillation:
𝑉1 𝐿𝑜
1
2
𝑖−1
𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚 𝑖
𝑖
𝑖
𝑖+1

𝑁
Trays 𝑉𝑁+1 𝐿𝑁

Assume: Constant molar flowrates of vapor & liquid.

Binary components: A, B

(more volatile)

Species balance on 𝒊𝒕𝒉 plate:

2
 𝐿𝑖 = 𝑥𝑖 𝐿
𝑉𝑖 = 𝑦𝑖 𝑉
𝑦𝑖 = 𝑘𝑖 𝑥𝑖 (𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚)

𝑦 𝑉
Therefore, 𝑉𝑖 = ( 𝑖) ( ) 𝐿𝑖
𝑥𝑖 𝐿
or 𝑉𝑖 = 𝐴′𝑖 𝐿𝑖

And, 𝑉𝑖 + 𝐿𝑖 = 𝑉𝑖+1 + 𝐿𝑖−1 (𝑖𝑛 = 𝑜𝑢𝑡)

𝑉𝑖 ↓ 𝑙𝑖−1
⋯ 𝑖 𝑡ℎ 𝑝𝑙𝑎𝑡𝑒
𝑉𝑖+1 ↓ 𝑙𝑖

𝑉𝑖−1
𝑉𝑖 + 𝑉𝑖 ⁄𝐴′𝑖 − 𝐴′𝑖−1
− 𝑉𝑖+1 = 0

Re-arrange 𝐴𝑖−1 𝑉𝑖−1 + (1 − 𝐴𝑖 )𝑉𝑖 − 𝑉𝑖+1 = 0

𝑖 = 1, 𝑁

Thus, there are ‘N’ equations and (N+2) variables (𝑉𝑜 ⋯ ⋯ 𝑉𝑁+1 )

However, 𝑉𝑁+1 𝑎𝑛𝑑 𝐿𝑜 are known (inlet or boundary conditions)

Re-arrange for all trays 𝑖 = 1, 𝑁

(1 − 𝐴1 )𝑉1 − 𝑉2 = −𝐴𝑜 𝑉𝑜 (= 𝐿𝑜 ) (𝑘𝑛𝑜𝑤𝑛)

𝐴1 𝑉1 + ( 1 − 𝐴2 )𝑉2 − 𝑉3 = 0 (1)

𝐴𝑁−1 𝑉𝑁−1 + ( 1 − 𝐴𝑁+1 )𝑉𝑁 = 𝑉𝑁+1 (𝑘𝑛𝑜𝑤𝑛) (𝑁)

We have a set of algebraic equations (linear/non-linear) represented as

𝐴𝑋̅ = 𝑏̅

coefficient matrix column vectors or (𝑁 × 1) 𝑚𝑎𝑡𝑟𝑖𝑥

rows column

( 1 − 𝐴1 ) −1 0 0 0 0
𝐴1 ( 1 − 𝐴2 ) −1 0 0 0
𝐴=
0 𝐴2 (1 − 𝐴3 ) −1 0 0
[ 0 0 0 0 𝐴𝑁−1 (1 − 𝐴𝑁+1 )]𝑁×𝑁

rows columns

3
 𝑉1 𝐿0
𝑉 ⋮
𝑋̅ = { 2 } , 𝑏̅ = { } (Known)
⋮ ⋮
𝑉𝑁 𝑉𝑁+1

column vectors

Alternatively*,

( 1 − 𝐴1 ) −1 0 0 0 0
𝑉1 𝐿𝑜
( 1 − 𝐴1 ) −1 0 0 0 0
𝑉
𝐴1 ( 1 − 𝐴2 ) −1 0 0 0 [ 2] =[ ]
⋮ ⋮
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ 𝑉 𝑉𝑁+1 𝑁×1
[ 0 0 0 0 𝐴𝑁−1 (1 − 𝐴𝑁+1 )]𝑁×𝑁 𝑁 𝑁×1

∗ represented as matrix multiplication

Note:

Most of the finite difference methods lead to the discretization of ODEs/PDEs and a set of
algebraic equations.
You have to solve a set of algebraic equations using Gauss Elimination, G-Jordan, etc, methods.

Example 2: Particle settling

(𝑑𝑝 , 𝜌𝑝 , 𝑣𝑝 , 𝐴𝑝 , 𝑚𝑝 )

𝐻 𝑙𝑖𝑞𝑢𝑖𝑑/𝑔𝑎𝑠

𝑡 =?

(𝜌𝑓 , 𝑣𝑓 )

Force balance on the particle:

𝑑𝑣 1
𝑚𝑝 𝑑𝑡 = 𝑚𝑝 𝑔 − 𝑣𝑝 𝜌𝑓 𝑔 − 𝐶𝐷 (2 𝜌𝑓 𝑣 2 ) 𝐴𝑝

𝑤𝑒𝑖𝑔ℎ𝑡 𝑏𝑢𝑜𝑦𝑎𝑛𝑐𝑦 𝑑𝑟𝑎𝑔

𝑑𝑣 𝜌 1 𝑑𝑝2
= 𝑔 − 𝜌𝑓 𝑔 − 𝐶𝐷 (2 𝜌𝑓 𝑣 2 ) (𝜋 )
𝑑𝑡 𝑝 4

4
 𝑑𝑣
= 𝐴 − 𝐵𝐶𝐷 (𝑣)𝑣 2 𝐶𝐷 24⁄𝑅𝑒 , 𝑝
𝑑𝑡

= 𝐴 − 𝐵Φ(𝑣) 0.44

𝑡 = 0, 𝑣=0 𝑅𝑒 , 𝑝

Therefore, you have to solve an initial value problem (ODE)

Solution: linear or non-linear

homogeneous or non-homogeneous
𝑣𝑡 or a set of ODEs.
v

Methods: RK-4, Euler Forward, etc.

0 t

Example 3: Heating or cooling of a moving particle (𝛁𝑻(𝒓) = 𝟎)

𝑡=0 𝑣𝑝 = 0

𝑇 𝑇𝑓 𝑇 = 𝑇𝑜 > 𝑇𝑓 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡)

𝑟𝑝 ℎ𝑓 (𝑓𝑙𝑢𝑖𝑑 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒)

𝑇(𝑡) =? (𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡)

Assume/ Apply lump-body approach (𝑩𝒊 → 𝟎)

ℎ𝑑𝑝
1. Energy balance over entire particle →0 , 𝑘𝑝 >> ℎ𝑑𝑝
𝑘𝑝

𝑑𝑇
𝑚𝐶𝑝 𝑑𝑡 = −ℎ𝐴𝑠 (𝑇 − 𝑇𝑓 )

(𝐴𝑠 = 𝜋𝑑𝑝2 )
(𝑁𝑜 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑛𝑠𝑖𝑑𝑒 𝑡ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒)
ℎ𝑑𝑝
ℎ = ℎ(𝑁𝑢) ⇒ = 𝑓(𝑅𝑒,𝑝, 𝑃𝑟 )
𝑘𝑓

𝑣1 𝑑𝑝 𝜌𝑓 𝐶𝑝 𝜇
𝑜𝑟 ℎ = ℎ (𝑣𝑝 ); 𝑅𝑒 = , 𝑃𝑟 = ( )
𝜇𝑓 𝑘 𝑓

2 Time-dependence of velocity (see previous example) and energy balance equation can be
written as

5
 𝑑𝑣𝑝
= 𝐴 − 𝐵Φ(𝑣𝑝 )
𝑑𝑡
𝑑𝑇 } 𝑡=0 𝑣𝑝 = 0, 𝑇 = 𝑇𝑜 Type equation here.
= 𝐶 𝐷(𝑣𝑝 ) − 𝐸
𝑑𝑡

In this example, two ODEs are to be solved; first y1 and then y2.

𝑑𝑦1
= 𝑓(𝑦1 )
𝑑𝑡 } 𝑇𝑤𝑜 𝑂𝐷𝐸𝑠
𝑑𝑦2
= Φ(𝑦1 , 𝑦2 )
𝑑𝑡

Example 4: Heating or cooling of a stationary solid (𝛁𝑻 ≠ 𝟎) in a moving fluid

𝑡=0 𝑇 = 𝑇𝑜 > 𝑇𝑓 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡)

𝑓𝑙𝑢𝑖𝑑 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒
𝑟
𝑇(𝑡, 𝑟) =? 𝑟𝑝 𝑇𝑓

ℎ𝑓 (𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡)

Assume ∇𝑇(𝑟) = 0, 𝑖. 𝑒. , 𝑡ℎ𝑒𝑟𝑒 𝑖𝑠 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑛𝑠𝑖𝑑𝑒 𝑡ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒

Energy balance within the particle:

𝜕𝑇 𝑘 𝜕 𝜕𝑇
𝜌𝐶𝑝 𝜕𝑡 = 𝑟 2 𝜕𝑟 (𝑟 2 𝜕𝑟 )

𝑡=0 𝑇 = 𝑇𝑜 𝑟𝑝 ≥ 𝑟 ≥ 0 (𝐼𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

𝜕𝑇
= 0+ 𝑟 = 0 =0 (𝐵𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠)
𝜕𝑟
𝜕𝑇
𝑟 = 𝑟𝑝, − 𝑘 𝜕𝑟 = ℎ(𝑇 − 𝑇𝑓 )

Now, we have to solve a PDE equation (unsteady – state 1D on space)

Under the steady-state condition with/without a source term, the equation will be
reduced to a 2nd order ODE or a boundary value problem.

Example 5: Mass transfer in a tubular liquid flow (steady- state)

6
 𝐶𝐴 = 𝐶𝑖𝑛 𝑅
𝑟 (𝑤𝑎𝑡𝑒𝑟 − 𝑓𝑙𝑜𝑤) 𝑣𝑧 (𝑟)

𝑡=0 z
tube length = L

For 𝑅𝑒 < 2100

𝑟2
𝑣𝑧 (𝑟) = 2𝑣𝑜 (1 − 𝑅2 )

𝑐(𝑡, 𝑧, 𝑟) =?

Species (A) balance in the CV (2𝜋𝑟 ∆𝑧 ∆𝑟)

CV
𝜕𝐶𝐴
+ 𝑣𝑧 . ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 + 𝑟(= 0)
𝜕𝑡

(no reaction)

𝜕𝐶𝐴 𝑟 2 𝜕𝐶𝐴 𝜕𝐶𝐴 𝜕𝐶𝐴 1 𝜕 𝜕𝐶𝐴

+ 2𝑣𝑜 (1 − 2 ) + 𝑣𝑟 = 𝐷( 2 + (𝑟 ))
𝜕𝑡 𝑅 𝜕𝑧 𝜕𝑟 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟

𝑡 =0, 𝐶𝐴 = 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒

0+ 𝑧=0 𝐶𝐴 = 𝐶𝑖𝑛
𝜕𝐶𝐴
𝑧=𝐿 𝜕𝑧
=0 (𝑙𝑜𝑛𝑔 𝑡𝑢𝑏𝑒 𝑎𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑖𝑜𝑛)

𝜕𝐶𝐴
𝑟 =0, =0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)
𝜕𝑟
𝜕𝐶𝐴
=𝑅 − 𝜕𝑟
= 0 (𝑛𝑜𝑛 − 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒 𝑤𝑎𝑙𝑙)

In this example, we have a transient/unsteady-state 2D (space) problem or a PDE to solve.

Under the steady-state condition with/without a source term, the equation will be
reduced to an elliptic PDE or 2nd order (in both z and r) PDE.

7
Finite Difference-based
Numerical Methods in
Chemical Engineering

1
 Lecture #01
Finite Difference-based Numerical Methods in Chemical Engineering

Recommended Books:

1) Linear Algebra by Gilbert Strang , Cengage, 4th ed.

2) Numerical Methods for Engineers by S.K. Gupta, Wiley Eastern Ltd, 2nd ed.
3) Numerical Methods for Engineering Application by Joel H. Feriziger, Wiley, 2nd ed.

Objective: Learn a few basic numerical techniques (Finite Difference) to solve simple
ODEs/PDEs/algebraic equations derived from a variety of heat and mass transfer related
conservation equations.

 Analytical methods yield exact solutions, often in infinite series; use mathematical functions.
 Numerical methods yield approximate solutions; use numbers, computations; results are dependent
on the method accuracies and limited by computers machine errors.
Course focuses on
 Algebra (Matrix Operation)
 Methods (Finite Difference)
Requires knowledge of
 Computer Programming (because large # of iterations)
(there are engineering software, e.g., Matlab, NAG, IMSL, Polymath)
 No numerical analysis in this course!

Examples: (Applications in chemical engineering)

1. Distillation:
𝑉1 𝐿𝑜
1
2
𝑖−1
𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚 𝑖
𝑖
𝑖
𝑖+1

𝑁
Trays 𝑉𝑁+1 𝐿𝑁

Assume: Constant molar flowrates of vapor & liquid.

Binary components: A, B

(more volatile)

Species balance on 𝒊𝒕𝒉 plate:

2
 𝐿𝑖 = 𝑥𝑖 𝐿
𝑉𝑖 = 𝑦𝑖 𝑉
𝑦𝑖 = 𝑘𝑖 𝑥𝑖 (𝑒𝑞𝑢𝑖𝑙𝑖𝑏𝑟𝑖𝑢𝑚)

𝑦 𝑉
Therefore, 𝑉𝑖 = ( 𝑖) ( ) 𝐿𝑖
𝑥𝑖 𝐿
or 𝑉𝑖 = 𝐴′𝑖 𝐿𝑖

And, 𝑉𝑖 + 𝐿𝑖 = 𝑉𝑖+1 + 𝐿𝑖−1 (𝑖𝑛 = 𝑜𝑢𝑡)

𝑉𝑖 ↓ 𝑙𝑖−1
⋯ 𝑖 𝑡ℎ 𝑝𝑙𝑎𝑡𝑒
𝑉𝑖+1 ↓ 𝑙𝑖

𝑉𝑖−1
𝑉𝑖 + 𝑉𝑖 ⁄𝐴′𝑖 − 𝐴′𝑖−1
− 𝑉𝑖+1 = 0

Re-arrange 𝐴𝑖−1 𝑉𝑖−1 + (1 − 𝐴𝑖 )𝑉𝑖 − 𝑉𝑖+1 = 0

𝑖 = 1, 𝑁

Thus, there are ‘N’ equations and (N+2) variables (𝑉𝑜 ⋯ ⋯ 𝑉𝑁+1 )

However, 𝑉𝑁+1 𝑎𝑛𝑑 𝐿𝑜 are known (inlet or boundary conditions)

Re-arrange for all trays 𝑖 = 1, 𝑁

(1 − 𝐴1 )𝑉1 − 𝑉2 = −𝐴𝑜 𝑉𝑜 (= 𝐿𝑜 ) (𝑘𝑛𝑜𝑤𝑛)

𝐴1 𝑉1 + ( 1 − 𝐴2 )𝑉2 − 𝑉3 = 0 (1)

𝐴𝑁−1 𝑉𝑁−1 + ( 1 − 𝐴𝑁+1 )𝑉𝑁 = 𝑉𝑁+1 (𝑘𝑛𝑜𝑤𝑛) (𝑁)

We have a set of algebraic equations (linear/non-linear) represented as

𝐴𝑋̅ = 𝑏̅

coefficient matrix column vectors or (𝑁 × 1) 𝑚𝑎𝑡𝑟𝑖𝑥

rows column

( 1 − 𝐴1 ) −1 0 0 0 0
𝐴1 ( 1 − 𝐴2 ) −1 0 0 0
𝐴=
0 𝐴2 (1 − 𝐴3 ) −1 0 0
[ 0 0 0 0 𝐴𝑁−1 (1 − 𝐴𝑁+1 )]𝑁×𝑁

rows columns

3
 𝑉1 𝐿0
𝑉 ⋮
𝑋̅ = { 2 } , 𝑏̅ = { } (Known)
⋮ ⋮
𝑉𝑁 𝑉𝑁+1

column vectors

Alternatively*,

( 1 − 𝐴1 ) −1 0 0 0 0
𝑉1 𝐿𝑜
( 1 − 𝐴1 ) −1 0 0 0 0
𝑉
𝐴1 ( 1 − 𝐴2 ) −1 0 0 0 [ 2] =[ ]
⋮ ⋮
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ 𝑉 𝑉𝑁+1 𝑁×1
[ 0 0 0 0 𝐴𝑁−1 (1 − 𝐴𝑁+1 )]𝑁×𝑁 𝑁 𝑁×1

∗ represented as matrix multiplication

Note:

Most of the finite difference methods lead to the discretization of ODEs/PDEs and a set of
algebraic equations.
You have to solve a set of algebraic equations using Gauss Elimination, G-Jordan, etc, methods.

Example 2: Particle settling

(𝑑𝑝 , 𝜌𝑝 , 𝑣𝑝 , 𝐴𝑝 , 𝑚𝑝 )

𝐻 𝑙𝑖𝑞𝑢𝑖𝑑/𝑔𝑎𝑠

𝑡 =?

(𝜌𝑓 , 𝑣𝑓 )

Force balance on the particle:

𝑑𝑣 1
𝑚𝑝 𝑑𝑡 = 𝑚𝑝 𝑔 − 𝑣𝑝 𝜌𝑓 𝑔 − 𝐶𝐷 (2 𝜌𝑓 𝑣 2 ) 𝐴𝑝

𝑤𝑒𝑖𝑔ℎ𝑡 𝑏𝑢𝑜𝑦𝑎𝑛𝑐𝑦 𝑑𝑟𝑎𝑔

𝑑𝑣 𝜌 1 𝑑𝑝2
= 𝑔 − 𝜌𝑓 𝑔 − 𝐶𝐷 (2 𝜌𝑓 𝑣 2 ) (𝜋 )
𝑑𝑡 𝑝 4

4
 𝑑𝑣
= 𝐴 − 𝐵𝐶𝐷 (𝑣)𝑣 2 𝐶𝐷 24⁄𝑅𝑒 , 𝑝
𝑑𝑡

= 𝐴 − 𝐵Φ(𝑣) 0.44

𝑡 = 0, 𝑣=0 𝑅𝑒 , 𝑝

Therefore, you have to solve an initial value problem (ODE)

Solution: linear or non-linear

homogeneous or non-homogeneous
𝑣𝑡 or a set of ODEs.
v

Methods: RK-4, Euler Forward, etc.

0 t

Example 3: Heating or cooling of a moving particle (𝛁𝑻(𝒓) = 𝟎)

𝑡=0 𝑣𝑝 = 0

𝑇 𝑇𝑓 𝑇 = 𝑇𝑜 > 𝑇𝑓 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡)

𝑟𝑝 ℎ𝑓 (𝑓𝑙𝑢𝑖𝑑 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒)

𝑇(𝑡) =? (𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡)

Assume/ Apply lump-body approach (𝑩𝒊 → 𝟎)

ℎ𝑑𝑝
1. Energy balance over entire particle →0 , 𝑘𝑝 >> ℎ𝑑𝑝
𝑘𝑝

𝑑𝑇
𝑚𝐶𝑝 𝑑𝑡 = −ℎ𝐴𝑠 (𝑇 − 𝑇𝑓 )

(𝐴𝑠 = 𝜋𝑑𝑝2 )
(𝑁𝑜 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑛𝑠𝑖𝑑𝑒 𝑡ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒)
ℎ𝑑𝑝
ℎ = ℎ(𝑁𝑢) ⇒ = 𝑓(𝑅𝑒,𝑝, 𝑃𝑟 )
𝑘𝑓

𝑣1 𝑑𝑝 𝜌𝑓 𝐶𝑝 𝜇
𝑜𝑟 ℎ = ℎ (𝑣𝑝 ); 𝑅𝑒 = , 𝑃𝑟 = ( )
𝜇𝑓 𝑘 𝑓

2 Time-dependence of velocity (see previous example) and energy balance equation can be
written as

5
 𝑑𝑣𝑝
= 𝐴 − 𝐵Φ(𝑣𝑝 )
𝑑𝑡
𝑑𝑇 } 𝑡=0 𝑣𝑝 = 0, 𝑇 = 𝑇𝑜 Type equation here.
= 𝐶 𝐷(𝑣𝑝 ) − 𝐸
𝑑𝑡

In this example, two ODEs are to be solved; first y1 and then y2.

𝑑𝑦1
= 𝑓(𝑦1 )
𝑑𝑡 } 𝑇𝑤𝑜 𝑂𝐷𝐸𝑠
𝑑𝑦2
= Φ(𝑦1 , 𝑦2 )
𝑑𝑡

Example 4: Heating or cooling of a stationary solid (𝛁𝑻 ≠ 𝟎) in a moving fluid

𝑡=0 𝑇 = 𝑇𝑜 > 𝑇𝑓 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡)

𝑓𝑙𝑢𝑖𝑑 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒
𝑟
𝑇(𝑡, 𝑟) =? 𝑟𝑝 𝑇𝑓

ℎ𝑓 (𝑐𝑜𝑛𝑣𝑒𝑐𝑡𝑖𝑣𝑒 ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡)

Assume ∇𝑇(𝑟) = 0, 𝑖. 𝑒. , 𝑡ℎ𝑒𝑟𝑒 𝑖𝑠 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝑔𝑟𝑎𝑑𝑖𝑒𝑛𝑡 𝑖𝑛𝑠𝑖𝑑𝑒 𝑡ℎ𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒

Energy balance within the particle:

𝜕𝑇 𝑘 𝜕 𝜕𝑇
𝜌𝐶𝑝 𝜕𝑡 = 𝑟 2 𝜕𝑟 (𝑟 2 𝜕𝑟 )

𝑡=0 𝑇 = 𝑇𝑜 𝑟𝑝 ≥ 𝑟 ≥ 0 (𝐼𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

𝜕𝑇
= 0+ 𝑟 = 0 =0 (𝐵𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠)
𝜕𝑟
𝜕𝑇
𝑟 = 𝑟𝑝, − 𝑘 𝜕𝑟 = ℎ(𝑇 − 𝑇𝑓 )

Now, we have to solve a PDE equation (unsteady – state 1D on space)

Under the steady-state condition with/without a source term, the equation will be
reduced to a 2nd order ODE or a boundary value problem.

Example 5: Mass transfer in a tubular liquid flow (steady- state)

6
 𝐶𝐴 = 𝐶𝑖𝑛 𝑅
𝑟 (𝑤𝑎𝑡𝑒𝑟 − 𝑓𝑙𝑜𝑤) 𝑣𝑧 (𝑟)

𝑡=0 z
tube length = L

For 𝑅𝑒 < 2100

𝑟2
𝑣𝑧 (𝑟) = 2𝑣𝑜 (1 − 𝑅2 )

𝑐(𝑡, 𝑧, 𝑟) =?

Species (A) balance in the CV (2𝜋𝑟 ∆𝑧 ∆𝑟)

CV
𝜕𝐶𝐴
+ 𝑣𝑧 . ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 + 𝑟(= 0)
𝜕𝑡

(no reaction)

𝜕𝐶𝐴 𝑟 2 𝜕𝐶𝐴 𝜕𝐶𝐴 𝜕𝐶𝐴 1 𝜕 𝜕𝐶𝐴

+ 2𝑣𝑜 (1 − 2 ) + 𝑣𝑟 = 𝐷( 2 + (𝑟 ))
𝜕𝑡 𝑅 𝜕𝑧 𝜕𝑟 𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟

𝑡 =0, 𝐶𝐴 = 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒

0+ 𝑧=0 𝐶𝐴 = 𝐶𝑖𝑛
𝜕𝐶𝐴
𝑧=𝐿 𝜕𝑧
=0 (𝑙𝑜𝑛𝑔 𝑡𝑢𝑏𝑒 𝑎𝑝𝑝𝑟𝑜𝑥𝑖𝑚𝑎𝑡𝑖𝑜𝑛)

𝜕𝐶𝐴
𝑟 =0, =0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)
𝜕𝑟
𝜕𝐶𝐴
=𝑅 − 𝜕𝑟
= 0 (𝑛𝑜𝑛 − 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒 𝑤𝑎𝑙𝑙)

In this example, we have a transient/unsteady-state 2D (space) problem or a PDE to solve.

Under the steady-state condition with/without a source term, the equation will be
reduced to an elliptic PDE or 2nd order (in both z and r) PDE.

7
 Lecture #02

Linear Algebraic Equations

𝑓1 (𝑋1 , 𝑋2 , ⋯ ⋯ 𝑋𝑛 ) = 0
𝑓2 (𝑋1 , 𝑋2 , ⋯ ⋯ 𝑋𝑛 ) = 0
} system of 𝑛 linear algebraic equations
⋮
𝑓𝑛 (𝑋1 , 𝑋2 , ⋯ ⋯ 𝑋𝑛 ) = 0

or

𝑎11 𝑋1 + 𝑎12 𝑋2 + ⋯ ⋯ + 𝑎1𝑛 𝑋𝑛 = 𝑏1

𝑟𝑜𝑤 𝑐𝑜𝑙𝑢𝑚𝑛
𝑛 # 𝑜𝑓 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠 𝑡𝑜 𝑠𝑜𝑙𝑣𝑒 𝑛 # 𝑜𝑓𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠
𝑎21 𝑋1 + 𝑎22 𝑋2 + ⋯ ⋯ + 𝑎2𝑛 𝑋𝑛 = 𝑏2
⋮
𝑎𝑛1 𝑋1 + 𝑎𝑛2 𝑋2 + ⋯ ⋯ + 𝑎𝑛𝑛 𝑋𝑛 = 𝑏𝑛 }

or 𝐴𝑋 = 𝑏 𝐴𝑋̅ = 𝑏̅

𝑤ℎ𝑒𝑟𝑒 𝐴 = 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑚𝑎𝑡𝑟𝑖𝑥

𝑎11 𝑎12 ⋯ 𝑎1𝑛
𝑎21 𝑎22 ⋯ 𝑎2𝑛
=[ ⋮ ⋮ ⋮ ⋮ ] (𝑎𝑖𝑗 𝑖𝑠 𝑎 𝑔𝑒𝑛𝑒𝑟𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡; 𝑖, 𝑗 = 1 𝑡𝑜 𝑛)
𝑎𝑛1 𝑎𝑛2 ⋯ 𝑎𝑛𝑛 𝒏×𝒏

row column
𝑋1 𝑏1
𝑋 𝑏
𝑋 = 𝑚𝑎𝑡𝑟𝑖𝑥 = [ 2 ] 𝑏 = 𝑚𝑎𝑡𝑟𝑖𝑥 = [ 2 ]
⋮ ⋮
𝑋𝑛 𝑛×1 𝑏𝑛 𝑛×1
or,
𝑋1 𝑏1
𝑋 𝑏
𝑋̅ = 𝑐𝑜𝑙𝑢𝑚𝑛 𝑣𝑒𝑐𝑡𝑜𝑟 = { 2 } , 𝑏̅ = 𝑐𝑜𝑙𝑢𝑚𝑛 𝑣𝑒𝑐𝑡𝑜𝑟 = { 2 }
⋮ ⋮
𝑋𝑛 𝑏𝑛

8
 Note: In general, a matrix has 𝑛 × 𝑚 elements

Therefore, aij :
i = 1 to n
j = 1 to m
If n = m, there may be unique solution and the rank of the matrix = m
In general, rank >, =, < m (3 cases)
(For more, refer a book on matrix operation)
In this course, we will be solving a number of equations with as many variables:
3x + 2y + z = 5
3x + 2y = 5
} or x + y − z = 2 } etc.
x−y=6
{ 2x + 2y − 2z = 4

Some properties of a matrix and operation

Symmetric matrix, 𝑎𝑖𝑗 = 𝑎𝑗𝑖

Square matrix, n=m
⋱ 0
Diagonal matrix, [ ] (𝑜𝑛𝑙𝑦 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠)
0 ⋱

Banded matrix (tridiagonal)

𝑎11 𝑎12 0 0 ⋯ 0
𝑎21 𝑎22 𝑎23 0 ⋯ 0
0 𝑎32 𝑎33 𝑎34 ⋯ 0
⋮ ⋮ ⋮ ⋮ ⋮ ⋮
[ 0 0 0 0 𝑎𝑛𝑛−1 𝑎𝑛𝑛 ] 𝑛×𝑛
𝑚𝑎𝑡𝑟𝑖𝑥

(The first and last rows have only two non-zero elements and all other rows have 3 elements,
one on the diagonal and the other two on each side of the diagonal element)

Upper/Lower triangular matrix:

(All elements below and above the diagonal are zero, respectively)

𝑎11 𝑎12 𝑎13 ⋯ 𝑎1𝑛 𝑎11

𝑎22 𝑎23 ⋯ 𝑎2𝑛 𝑎21 𝑎22
UTM: ⋱ ⋱ ⋮ LTM: 𝑎31 𝑎32 ⋱ 𝟎
𝟎 ⋱ 𝑎𝑛−1𝑛 ⋮ ⋮ ⋱ ⋱
[ 𝑎𝑛𝑛 ]𝒏×𝒏 [𝑎𝑛1 𝑎𝑛2 ⋯ 𝑎𝑛𝑛−1 𝑎𝑛𝑛 ]𝒏×𝒏

Operation:
Summation: [𝐴]𝑚×𝑛 = [𝐵]𝑚×𝑛 + [𝐶]𝑚×𝑛

9
Multiplication (rules):
[𝐴]𝑚×𝑛 × [𝐵]𝑛×𝑙 = [𝐶]𝑚×𝑙
same

Therefore, multiplication is not permissible for [𝐵] × [𝐴]

𝑛×𝑙 𝑚×𝑛
{
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡
𝑙≠𝑚

Also, these multiplications are permissible:

5 1 5 1 0
−3 0 −4 −3 2
[3 2 ] ×[ ] ; [3 2 ] ×[ ] ; [1 3 2]1×3 × [ 5 ]
1 0 2×2 2 7 −3 2×3
0 3 3×2 0 3 3×2 −8 3×1

Rule: 𝑪𝒊𝒋 = ∑𝒌𝒍=𝟏 𝒂𝒊𝒌 × 𝒃𝒌𝒋

3 1 (3 × 5 + 1 × 7) 3 × 6 + 1 × 2
5 6
examples: [8 6 ] ×[ ] = [(8 × 5 + 6 × 7) (8 × 6 + 6 × 2)]
7 2 2×2
0 4 3×2 0 × 5 + 4 × 7) (0 × 6 + 4 × 2
3×2

Inverse of a matrix:
1 0 0
[𝐴]−1 ⇒ [𝐴]𝑚×𝑛 × [𝐴 ]−1
𝑛×𝑚 = [𝐼] ≡ [ 0 1 0]
0 0 1

Identity matrix

For 𝟐 × 𝟐 matrix
1 𝑎 −𝑎12 𝑎11 𝑎12
[𝐴]−1 = [−𝑎22 𝑎11 ] 𝑤ℎ𝑒𝑟𝑒 𝐴 = [𝑎 𝑎22 ]
|𝐴| 21 21

Note: |𝐴| = (𝑎11 × 𝑎22 − 𝑎12 × 𝑎21 ) ≠ 0

else, matrix A is considered to be singular.
For bigger size matrices, there are special methods to determine[𝐴]−1 , discussed later.

Transpose of a matrix [𝑨]𝑻 :

𝑎 𝑎12 ⋯ 𝑎1𝑛 𝑎11 𝑎21 ⋯ 𝑎𝑚1

[𝐴] = [𝑎 11 𝑎 ⋯ 𝑎 ], [𝐴]𝑇 = [𝑎 ⋯ ⋯ 𝑎𝑚𝑛 ]𝑛×𝑚
𝑚1 𝑚2 𝑚𝑛 1𝑛
(rows have been interchanged with columns)

10
 𝑎1
𝑎2
example: [𝐴] = [ ⋮ ] ⇒ [𝐴]𝑇 = [𝑎1 𝑎2 ⋯ 𝑎𝑛 ]1×𝑛
𝑎𝑛 𝑛×1

Let us get back to solving a set of algebraic equations:

[𝐴]{𝑋} = {𝑏}
𝑜𝑟 [𝐴][𝑋] = [𝑏] } 𝑏 𝑇 = [𝑏1 𝑏2 ⋯ ⋯ 𝑏𝑛 ], 𝑋 𝑇 = [𝑋1 𝑋2 ⋯ ⋯ 𝑋𝑛 ]
𝑜𝑟 𝐴𝑋⃗ = 𝑏⃗⃗

𝑿 = 𝑨−𝟏 𝒃

𝐹𝑖𝑛𝑑𝑖𝑛𝑔 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑡𝑜 𝑡ℎ𝑒 𝑠𝑒𝑡 𝑜𝑓 𝑙𝑖𝑛𝑒𝑎𝑟 𝑎𝑙𝑔𝑒𝑏𝑟𝑎𝑖𝑐 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑖𝑠

⇒ { 𝑛𝑜𝑡ℎ𝑖𝑛𝑔 𝑏𝑢𝑡 𝑓𝑖𝑛𝑑𝑖𝑛𝑔 𝑖𝑛𝑣𝑒𝑟𝑠𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑚𝑎𝑡𝑟𝑖𝑥, 𝑓𝑜𝑙𝑙𝑜𝑤𝑒𝑑 𝑏𝑦
𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛 𝑤𝑖𝑡ℎ 𝒃.

Generally, Cremer’s rule works nicely for small matrix:

|𝑨𝒋 |
𝑿𝒋 =
|𝑨|

𝐴𝑗 is determined by replacing 𝑗 𝑡ℎ column of A by b.

3 5 7 7
Therefore, if 𝐴 = [ 2 0 8] ; 𝑏 = {3}
−1 2 5 2
3 7 7
𝑋2 = [ 2 3 8]⁄|𝐴| ; |𝐴| ≠ 0
−1 2 5
3 5 7
𝑋3 = [ 2 0 3]⁄|𝐴| , 𝑒𝑡𝑐.
−1 2 2
Let us briefly discuss unique solution, singularity, and rank (r) of a (𝑚 × 𝑛) matrix. (Read the
textbook for details)

Consider the following set of equations:

2𝑥1 + 3𝑥2 = 11
4𝑥1 + 6𝑥2 = 22

Note: Two equations are not independent.

2 3
and |𝐴| = [ ] =0
4 6 2×2

11
Therefore, there is no unique solution, coefficient matrix is singular and 𝑟 of such matrix = 1 <
2 𝑜𝑟 𝑚(# 𝑜𝑓 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛𝑠) < 𝑛(# 𝑜𝑓 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠)

2𝑥1 + 3𝑥2 = 11
𝑥1 + 𝑥2 = 4

Note: Two equations are independent.

|𝐴| = [2 3] ≠ 0; 𝑎𝑛𝑑 𝑚 = 𝑛, 𝑟 = 2(𝑛)
1 1 2×2

There is a unique solution.

Geometrically, two equations represent two straight lines intersecting at a unique point.

Similarly, 3 independent equations for 𝑋2

3 variables will represent 3 planes on a ⇒

geometrical space; two planes intersecting

will yield a straight line, and two straight lines 𝑋1

intersecting will yield a point (𝑋1 , 𝑋2 , 𝑋3 ), or the unique solution to the equations.

2𝑥1 + 3𝑥2 − 𝑥3 = 11
𝑥1 + 𝑥2 + 𝑥3 = 4
𝑥1 + 2𝑥2 + 2𝑥3 = 7

Note: There are three linearly independent equations representing three non-parallel
equations. In this case, m = 𝑛, 𝑟 = 3(𝑛).

12
 Lecture #03
Gauss Elimination

It is the most widely used method to solve a set of linear algebraic equations. The method
reduces the original matrix to an upper triangular matrix which can also be used to determine
the determinant of the matrix. Interestingly, one can also inspect the intermediate
steps/solutions to determine singularity, rank, and number of independent equations.

General steps: 𝐴𝑥̅ = 𝑏̅ 𝑤ℎ𝑒𝑟𝑒 𝐴 𝑖𝑠 𝑎 3 × 3 𝑚𝑎𝑡𝑟𝑖𝑥

𝑎11 𝑥1 + 𝑎12 𝑥2 + 𝑎13 𝑥3 = 𝑏1 ⋯ ⋯ ⋯ 𝑅1

𝑎21 𝑥1 + 𝑎22 𝑥2 + 𝑎23 𝑥3 = 𝑏2 ⋯ ⋯ ⋯ 𝑅2

𝑎31 𝑥1 + 𝑎32 𝑥2 + 𝑎33 𝑥3 = 𝑏3 ⋯ ⋯ ⋯ 𝑅3

𝑎11 𝑎12 𝑎13 𝑥1 𝑏1

or [𝑎21 𝑎22 𝑎23 ] {𝑥2 } = {𝑏2 }
𝑎31 𝑎32 𝑎33 𝑥3 𝑏3

𝑅1 , first row is called pivot row and first non-zero element (𝑎11 ) is called pivotal element.
Subsequent operations are performed around pivot row and pivotal element.

𝑎11 𝑎12 𝑎13 𝑥1 𝑏1

′ ′
Step 1: [ 0 𝑎22 𝑎23 ] {𝑥2 } = {𝑏2 ′ }
0 𝑎32 ′ 𝑎33 ′ 𝑥3 𝑏3 ′
𝑎
𝑎22 ′ = 𝑎22 − (𝑎21 ) × 𝑎12 Briefly, the first equation is divided by
11
𝑎 a11 , multiplied by a21 , and substracted
where 𝑎23 ′ = 𝑎23 − (𝑎21 ) × 𝑎13 (2)
11 from the second equation to yield the
′ 𝑎21
𝑏2 = 𝑏2 − (𝑎 ) × 𝑏1 } modified equation (2)
11

𝑎11 ≠ 0

Similarly,
𝑎
𝑎32 ′ = 𝑎32 − (𝑎31 ) × 𝑎12 Briefly, the first equation is divided by
11
𝑎 a11 , multiplied by a31 , and substracted
𝑎33 ′ = 𝑎33 − (𝑎31 ) × 𝑎13 (3)
11 from the 3rd equation to yield the
′ 𝑎31
𝑏3 = 𝑏3 − (𝑎 ) × 𝑏1 } modified equation (3)
11

𝑎
or, in the simplest form 𝑅2 ′ = 𝑅2 − 𝑅1 (𝑎21 )
11

13
 𝑎
𝑅3 ′ = 𝑅3 − 𝑅1 (𝑎31 )
11

Note: Multiplication or division of an equation, or subtraction or addition of one independent

equation from/to another independent equation does not yield a new independent equation.
Therefore, at the end of step 1 we still have three linearly independent equations, although
modified, to solve.

Geometrically, two planes (non-parallel) will intersect to yield a straight line. Therefore, step1
has yielded two straight lines by the intersections of plane/(eq 1) with plane 2 (eq 2) and with
plane 3 (eq 3).

Step2: Follow the similar procedure. Now 𝑅2 ′ (2nd row) becomes pivotal equation and
𝑎22 ′ becomes pivotal element.

𝑎11 𝑎12 𝑎13 𝑥1 𝑏1

[ 0 𝑎22 ′ 𝑎23 ′ ] {𝑥2 } = { 𝑏2 ′ }
0 0 𝑎33 ′′ 𝑥3 𝑏3 ′′

where,

𝑎 ′
𝑎33 ′′ = 𝑎33 ′ − (𝑎32 ′ ) × 𝑎23 ′
22

𝑎 ′
𝑏3 ′′ = 𝑏3 ′ − (𝑎32 ′ ) × 𝑏2 ′
22

𝑎 ′
or in the simplest form, 𝑅3 ′′ = 𝑅3 ′ − 𝑅2 ′ (𝑎32 ′ )
22

Geometrically, two straight lines (𝑅2 ′ 𝑎𝑛𝑑 𝑅3 ′ ) intersect at a point.

Note: At the end of step (2) an upper triangular matrix is obtained. The three modified
algebraic equations are as follows:

𝑎11 𝑥1 + 𝑎12 𝑥2 + 𝑎13 𝑥3 = 𝑏1

𝑎22 ′ 𝑥2 + 𝑎23 ′ 𝑥3 = 𝑏2 ′ }
𝑎33 ′′ 𝑥3 = 𝑏33 ′′

This is a new set of three linearly independent algebraic equations to solve! Here ‘new’ means
‘modified’.

14
Reverse (back) substitution (to determine 𝒙
̅):

𝑥3 = 𝑏3 ′′ ⁄𝑎33 ′′ (from the last row)

𝑏2 ′ −𝑎23 ′ 𝑥3
𝑥2 = (from the 2nd last row)
𝑎22 ′

𝑏1 −𝑎12 𝑥2 −𝑎13 𝑥3
𝑥1 = (from the 1st row)
𝑎11

In general, for the ith variable of ‘n’ equations

(𝑖−1) (𝑖−1)
𝑏𝑖 − ∑𝑛
𝑗=𝑖+1 𝑎𝑖𝑗 𝑥𝑗
𝑥𝑖 = (𝑖−1)
𝑎𝑖𝑖

A large number of equations (𝑛 > 3) requires a programming code to solve.

Evaluation of determinant:

Consider a simple 2 × 2 matrix

𝑎11 𝑎12
𝐴 = [𝑎 𝑎22 ]2×2 ⇒ 𝑑𝑒𝑡 𝐴 = (𝑎11 × 𝑎22 − 𝑎21 × 𝑎12 )
21

Apply GE method
𝑎11 𝑎12 𝑎21
𝐴′ = [ 0 ′
𝑎22 ′ ]2×2 ; 𝑎22 = (𝑎22 − (𝑎11 ) × 𝑎12 )

The determinant of the modified matrix 𝐴′ = (𝑎11 × 𝑎22 − 𝑎21 × 𝑎12 ) = 𝑠𝑎𝑚𝑒 𝑎𝑠 det(𝐴)

“Value of determinant is not changed by the forward elimination step of GE”. This must be
true because forward step only modifies the equations.

Take 3 × 3 matrix:
𝑎11 𝑎12 𝑎13 𝑎11 𝑎12 𝑎13
𝐺𝐸
𝑎
[ 21 𝑎22 𝑎23 ] ⇒ [ 0 𝑎22 ′ 𝑎23 ′ ]
𝑎31 𝑎32 𝑎33 0 0 𝑎33 ′′

[𝐴] [𝐴′ ]

𝑑𝑒𝑡[𝐴] = 𝑑𝑒𝑡[𝐴′ ] = 𝑎11 𝑎22 ′ 𝑎33 ′′ (can be checked)

Therefore,

𝑑𝑒𝑡[𝐴] ⇒ 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑜𝑓 𝑖𝑡𝑠 𝑢𝑝𝑝𝑒𝑟 𝑡𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑥

15
 Note that the swapping of rows of a matrix does not alter values of determinant. However,
sign changes: (−1)𝑘 , where k is the number of times rows are swapped.

Re-visit

𝑑𝑒𝑡[𝐴]3×3 = 𝑎11 𝑎22 ′ 𝑎33 ′′ (𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑜𝑓 𝑈𝑇𝑀 𝑜𝑓 𝐴)

Note: 1. For a matrix to be non-singular 𝑑𝑒𝑡[𝐴] ≠ 0

𝑜𝑟 𝑎11 𝑎22 ′ 𝑎33 ′′ ≠ 0

Alternatively, the corresponding three equations are linearly independent; rank of

[𝐴] = #𝑜𝑓 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 = #𝑜𝑓 𝑒𝑞 𝑛𝑠 = 3.

Geometrically, three planes intersect at a unique point in space, or none of them is ∥ to the
other.

Alternatively,

If a33 ′′ = 0, det[A] = 0 ⇒ the matrix is singular; rank of the matrix = 2(< 3); only first two
equations are linearly independent; the problem is under-determined and one more equation is
required to solve(𝑥1 , 𝑥2 , 𝑥3 ). Algebraically, 3rd plane is ∥ to one of the other two planes.

Pivoting and ill-conditioning:

𝐴𝑑𝑗|𝐴|
𝐴𝑋̅ = 𝑏̅ ⇒ 𝑋̅ = 𝐴−1 𝑏̅ ; 𝐴−1 = |𝐴|

Therefore, for 𝑋̅ to be solved or A to be invertible, |𝐴| ≠ 0. Yet if |𝐴| →

0 (𝑜𝑟 𝑖𝑡 𝑖𝑠 𝑎 𝑠𝑚𝑎𝑙𝑙 𝑛𝑢𝑚𝑏𝑒𝑟), a small change in |𝐴| will lead to a large change in 𝐴−1.
𝑑𝑌 1
[Note: 𝑌 = 1⁄𝑋 , = − 𝑋2
𝑑𝑋

In other words, a small change in X will lead to large change in Y]

In such case, the problem is said to be ill-conditioned. To invert such a matrix, large machine
precision may be required, and one has to be careful with round-off errors.

Recall, determinant of a matrix is the product of the diagonal elements of UTM. Therefore,
pivotal elements should not only be zero, but also large, or the matrix should be “diagonally
strong”.

16
Example: 0.0001𝑥1 + 𝑥2 = 1 (𝐴)

𝑥1 + 𝑥2 = 2

Working with 4th decimal, apply GE with 0.0001 as the pivotal element to obtain
𝑋 𝑇 = [0.0000 1.0000] which is wrong. Now, swap the equations as

𝑥1 + 𝑥2 = 2

0.0001𝑥1 + 𝑥2 = 1

Apply GE to obtain 𝑥 𝑇 = [1.0000 1.0000] which is 4th decimal accurate.

This is called pivoting. Swap the row, so that the pivotal element is relatively larger. Also,
scaling (or multiplying equation (A) by 10000) can be done before applying GE.

Example: Solve by GE:

1 −1 2 𝑥1 −8
[2 −2 3] {𝑥2 } = {−20}
1 1 1 𝑥3 −2
1 −1 2 𝑥1 −8
Step 1: [0 0 −1] {𝑥2 } = {−4}
0 2 −1 𝑥3 +6
Row pivoting is required (or swap 2nd equation with 3rd so that pivotal elemental is non – zero).

1 −1 2 𝑥1 −8
[0 2 −1] {𝑥2 } = {+6}
0 0 −1 𝑥3 −4
We have now UTM. Reverse substitution will yield

𝑥 𝑇 = [−11 5 4]

𝑑𝑒𝑡[𝐴] = 1 × 2 × −1 = −2 ≠ 0 (matrix is singular, rank = 3, all 3 equations are linearly

independent ). Also, note that we swapped one time. So, there is the multiplication with -1.

Column pivoting (generally it should be avoided)

1 2 −1 𝑥1 −8
[0 −1 0 ] {𝑥2 } = {−4}
0 −1 2 𝑥3 +6
Column 2 is replaced with column 3 to make the pivotal element ≠ 0. However, note that the
variables 𝑥2 𝑎𝑛𝑑 𝑥3 are also swapped in the column vector 𝑥̅ , else you would be solving a
different set of equations!

17
𝑥1 + 2𝑥2 − 𝑥3 = −8 𝑥1 + 2𝑥3 − 𝑥2 = −8
−𝑥2 = −4 }(left) original equation/(right) swapped equation: −𝑥3 = −4 }
−𝑥2 + 2𝑥3 = +6 2𝑥2 − 𝑥3 = +6

Proceed as follows:

1 −2 −1 𝑥1 −8
[0 𝑥
−1 0 ] { 3 } = {−4}
0 0 2 𝑥2 10

Reverse substitution will yield 𝑥 𝑇 = [−11 , 5 , 4]

18
 Lecture #04
Gauss-Jordan
1. Procedure is similar to that of GE

2. Normalizer the pivotal element

3. Eliminate the unknown from all rows, below and above the pivot row

4. No back substitution is required as required in GE.

𝐴𝑋̅ = 𝑏̅
𝑎11 𝑎12 𝑎13 𝑥1 𝑏1
[𝑎21 𝑎22 𝑎23 ] {𝑥2 } = {𝑏2 }
𝑎31 𝑎32 𝑎33 3×3 𝑥3 𝑏3
𝑎
Step 1: 𝑎1𝑗 ′ = 𝑎1𝑗 𝑗 = 1,3 (𝑛𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒 𝑎11 (𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡) to 1.0)
11

𝑏1 ′ = 𝑏1 ⁄𝑎11

𝑎2𝑗 ′ = 𝑎2𝑗 − 𝑎1𝑗 ′ × 𝑎21

𝑎3𝑗 ′ = 𝑎3𝑗 − 𝑎1𝑗 ′ × 𝑎31 𝑠𝑎𝑚𝑒 𝑎𝑠 𝑡ℎ𝑒 𝐺𝐸 𝑚𝑒𝑡ℎ𝑜𝑑; 𝑚𝑎𝑘𝑖𝑛𝑔
′ ′ 𝑡ℎ𝑒 𝑓𝑖𝑟𝑠𝑡 𝑐𝑜𝑙𝑢𝑚𝑛 𝑏𝑒𝑙𝑜𝑤 𝑡ℎ𝑒 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤 𝑡𝑜 𝑏𝑒 0.
𝑏2 = 𝑏2 − 𝑏1 × 𝑎21
′ ′
𝑏3 = 𝑏3 − 𝑏1 × 𝑎31 }

1 𝑎12 ′ 𝑎13 ′ 𝑏1 ′
[0 𝑎22 ′ 𝑎23 ′ ] { 𝑋̅ } = {𝑏2 ′ }
0 𝑎32 ′ 𝑎33 ′ 3×3 𝑏3 ′

𝑎 ′
Step2: 𝑎2𝑗 ′′ = 𝑎 2𝑗 ′ 𝑗 = 2,3 (𝑛𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡 𝑜𝑓 2𝑛𝑑 𝑟𝑜𝑤 to 1.0)
22

𝑏2 ′′ = 𝑏2 ′ ⁄𝑎22 ′

′′ ′
𝑎1𝑗 = 𝑎1𝑗 − 𝑎2𝑗 ′ × 𝑎12 ′ 𝑠𝑢𝑏𝑡𝑟𝑎𝑐𝑡 𝑡ℎ𝑒 𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑟𝑜𝑤 𝑎𝑏𝑜𝑣𝑒.
}
𝑏1 ′′ = 𝑏1 ′ − 𝑏2 ′ × 𝑎12 ′ 𝑵𝒐𝒕𝒆 𝒕𝒉𝒊𝒔 𝒊𝒔 𝒕𝒉𝒆 𝒆𝒙𝒕𝒓𝒂 𝒔𝒕𝒆𝒑 𝒊𝒏 𝑮𝑬.

′ ′
𝑎3𝑗 ′′ = 𝑎3𝑗 − 𝑎2𝑗 × 𝑎32 ′ 𝑠𝑢𝑏𝑡𝑟𝑎𝑐𝑡 𝑡ℎ𝑒 𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑟𝑜𝑤 𝑏𝑒𝑙𝑜𝑤.
′′ ′ ′ }
𝑏3 = 𝑏3 − 𝑏2 × 𝑎32 ′ 𝑇ℎ𝑖𝑠 𝑖𝑠 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑠𝑡𝑒𝑝 𝑎𝑠 𝑖𝑛 𝐺𝐸.

19
 1 0 𝑎13 ′′ 𝑏1 ′′
[0 1 𝑎23 ′′ ] 𝑋̅ = {𝑏2 ′′ }
0 0 𝑎33 ′′ 𝑏3 ′′

Step 3: 𝑏3 ′′′ = 𝑏3 ′′ ⁄𝑎33 ′′ (normalize pivotal element to 1)

𝑏2 ′′′ = 𝑏2 ′′ − 𝑏3 ′′′ 𝑎23 ′′ 𝑠𝑢𝑏𝑡𝑟𝑎𝑐𝑡 𝑏𝑜𝑡ℎ 𝑟𝑜𝑤𝑠

′′′ ′′ ′′′ ′′ } 1 𝑎𝑛𝑑 2 𝑎𝑏𝑜𝑣𝑒 𝑓𝑟𝑜𝑚 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤. 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑟𝑜𝑤 𝑏𝑒𝑙𝑜𝑤 𝑡ℎ𝑒 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤.
𝑏1 = 𝑏1 − 𝑏3 𝑎13

1 0 0 𝑥1 𝑏1 ′′′
[0 1 0] {𝑥2 } = {𝑏2 ′′′ }
0 0 1 𝑥3 𝑏3 ′′′

Therefore, 𝑋 𝑇 = 𝑏 (′′′)𝑇 : 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛.

No back substitution is required for a unit matrix!

Make a note that G-J method can also be used to determine inverse of the matrix, if working on
the augmented matrix. Check text books for more details.

1 −1 2 −8
example: [1 1 1] { 𝑋̅ } = { −2 }
2 −2 3 −20
1 −1 2 𝑥1 −8
⇒ [0 2 −1] [𝑥2 ] = [ 6 ] (pivotal element already had ‘1’ ⇒ no need to normalize!)
0 0 −1 𝑥3 −4
1 −1 2 𝑥1 −8
⇒ [0 𝑥
1 −0.5] [ 2 ] = [ 3 ] (pivotal row was divided by 2 to make pivotal element to be 1)
0 0 −1 𝑥3 −4
1 0 1.5 𝑥1 −5
⇒ [0 1 −0.5] [𝑥2 ] = [ 3 ] (rows were subtracted above and below from the pivotal row)
0 0 −1 𝑥3 −4
1 0 1.5 𝑥1 −5
⇒ [0 𝑥
1 −0.5] [ 2 ] = [ 3 ] (pivotal element was made to be 1)
0 0 1 𝑥3 4
1 0 0 𝑥1 −11
⇒ [0 ]
1 0 2[ 𝑥 ] = [ 5 ] (rows 1 & 2 subtracted from the pivotal row)
0 0 1 3 𝑥 4
𝑋 𝑇 = [−11 5 4]𝑇

You should ensure that the programming code for G-J is written by modifying the code for GE,
instead of writing afresh, by changing the indices for i and j.

20
LU Decomposition: Two similar/identical methods

Dolittle Crout’s

𝐴𝑋̅ = 𝑏̅

(This method is considered to be the best if only right hand side 𝑏̅ changes from problem/case
to problem/case)

1 0 0 𝑈11 𝑈12 𝑈13

[𝑙21 1 0] [ 0 𝑈22 𝑈23 ]
𝑙31 𝑙32 1 0 0 𝑈33
𝐴 = 𝐿𝑈 ⇒
𝐷𝑜𝑙𝑖𝑡𝑡𝑙𝑒
𝑑𝑒𝑐𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛𝑠
𝑙11 0 0 1 𝑈12 𝑈13
𝑖𝑛𝑡𝑜 𝐿 𝑎𝑛𝑑 𝑈
[𝑙21 𝑙22 0 ] [0 1 𝑈23 ]
𝑙31 𝑙32 𝑙33 0 0 1
{ 𝐶𝑟𝑜𝑢𝑡
In either case, steps are as follows:

𝐴 = 𝐿𝑈 (𝑑𝑒𝑐𝑜𝑚𝑝𝑜𝑠𝑒)
𝐿𝑈 𝑋̅ = 𝑏̅
} ⇒ 𝐿𝑑̅ = 𝑏̅, 𝑠𝑜𝑙𝑣𝑒 𝑓𝑜𝑟 𝑑̅ 𝑓𝑖𝑟𝑠𝑡
𝑑̅
𝑈𝑋 = 𝑑̅ ⇒ 𝑠𝑜𝑙𝑣𝑒 𝑓𝑜𝑟 𝑋̅
̅

1 −1 2 −8
example: [1 ̅
1 1] { 𝑋 } = { −2 }
2 −2 3 −20
1 0 0 𝑈11 𝑈12 𝑈13
𝐴 = [𝑙21 1 0] [ 0 𝑈22 𝑈23 ]
𝑙31 𝑙32 1 3×3 0 0 𝑈33 3×3

𝑈11 𝑈12 𝑈13

𝑙21 𝑈12 𝑙21 𝑈13
𝑙21 𝑈11 ( + ) ( + )
𝑈22 𝑈23
= 𝑙31 𝑈13
𝑙31 𝑈12 +
𝑙31 𝑈11 ( + ) 𝑙32 𝑈23
𝑙32 𝑈22 +
[ ( 33 )]3×3
𝑈

Match,

𝑈11 = 1, 𝑈12 = −1, 𝑈13 = 2

21
 𝑙21 𝑈11 = 1, 𝑙21 𝑈12 + 𝑈12 = 1, 𝑙21 𝑈13 + 𝑈23 = 1

𝑙31 𝑈11 = 2, 𝑙31 𝑈12 + 𝑙32 𝑈23 = −2, 𝑙31 𝑈13 + 𝑙32 𝑈23 + 𝑈33 = 3

Therefore, there are 9 unknown & 9 equations to solve and the problem is well defined.

How do we obtain LU?

Well, use GE to reduce A to an upper triangular matrix U.

However, in addition, store multiplication coefficients of GE as the coefficients 𝑙21 , 𝑙31 , 𝑎𝑛𝑑 𝑙32
of lower triangular matrix

𝑎 𝑘 = 1 𝑡𝑜 𝑛 − 1
as: 𝑙𝑖𝑘 = 𝑎 𝑖𝑘 ; }
𝑘𝑘 𝑖 = 𝑘 + 1 𝑡𝑜 𝑛
It is clear that the programming code used for GE to reduce the matrix A to UTM (U) is the
same as that for determining U of LU decomposition method. In addition, the coefficients of L
are also determined automatically by defining an extra coefficient 𝑙𝑖𝑘 (see above) in the same
programming code.

Example:

1 −1 2
𝐴 = [1 1 1]
2 −2 3

Step 1:

1 0 0 𝑈11 𝑈12 𝑈13

= [𝑙21 1 0] [ 0 𝑈22 𝑈23 ]
𝑙31 𝑙32 1 0 0 𝑈33

L U
𝑁𝑜𝑡𝑒 𝑙21 𝑖𝑠 𝑡ℎ𝑒
1 0 0 1 −1 2 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡
= [𝑙21 = 1 1 0] [0 2 −1] 𝑡𝑜 𝑚𝑎𝑘𝑒 𝑡ℎ𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡
𝑙31 = 2 𝑙32 = 0 1 0 0 −1 𝑈12 = 0. 𝑆𝑖𝑚𝑖𝑙𝑎𝑟𝑖𝑡𝑦,
{𝑙31 𝑎𝑛𝑑 𝑙32 𝑎𝑟𝑒 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒𝑑
L U

1 −1 2 1 0 0 1 −1 2
Therefore, [1 1 1] = [ 1 1 0] [0 2 −1]
2 −2 3 2 0 1 0 0 −1

A L U

22
Step 2: 𝐿𝑑⃗ = 𝑏⃗⃗

1 0 0 𝑑1 −8
[1 1 0] [𝑑2 ] = [ −2 ]
2 0 1 𝑑3 −20

𝑑1 = −8 ; 𝑑2 = −2 + 8 = 6; 𝑑3 = −20 + 16 = −4

𝑑𝑇 = [−8 6 − 4]

(Note: you have done forward substitution or solved a LTM, which is just opposite to solving
an UTM or reverse/backward substitution step of GE)

Step 3: 𝑈𝑋̅ = 𝑑̅

1 −1 2 𝑋1 −8
[0 2 −1] {𝑋2 } = { 6 }
0 0 −1 𝑋3 −4
(Note this is the reverse/backward substitution of GE)

𝑋3 = 4, 𝑋2 = (6 + 4)⁄2 = 5, 𝑋1 = −8 + 5 − 8 = −11

𝑋 𝑇 = [−11 5 4]𝑇 Ans.

You should ensure that you are not writing a fresh code for the LU decomposition method.
Rather, you simply modify the code you wrote earlier for the 1st part (forward substitution) of
GE, or the code to reduce A to UTM. Then, write a code for inverting LTM and use the
previously written code for the back-substitution or the reverse substitution of GE. In other
words, the code of LU decomposition has three sub-parts.

23
 Lecture #05
Matrix Inverse 𝑨 𝑨−𝟏 = 𝑰(𝒅𝒆𝒇𝒊𝒏𝒊𝒕𝒊𝒐𝒏 𝒇𝒐𝒓 𝑨−𝟏 )
1 0 0
[0 1 0]
0 0 1
𝑎11 𝑎12 𝑎13
𝑎
If 𝐴 = [ 21 𝑎22 𝑎23 ], then
𝑎31 𝑎32 𝑎33

𝑎11 𝑎12 𝑎13 𝑏11 𝑏12 𝑏13 1 0 0

𝑎
[ 21 𝑎22 𝑎23 ] [𝑏21 𝑏22 𝑏23 ] = [0 1 0]
𝑎31 𝑎32 𝑎33 𝑏31 𝑏32 𝑏33 0 0 1
𝐴 𝐴−1 𝐼
Making use of the rule of matrix multiplication (row×column), it is clear that
𝑎11 𝑎12 𝑎13 𝑏11 1 1
[𝑎21 𝑎22 𝑎23 ] [𝑏21 ] = [0] ⇒ 𝐴𝑏̅1 = [0]
𝑎31 𝑎32 𝑎33 𝑏31 0 0
𝑏12 0
and [ ⃗⃗⃗⃗⃗2 = [0 1 0]𝑇
] [𝑏22 ] = [1] ⇒ 𝐴𝑏
𝑏32 0
𝑏13 0
[ ̅̅̅3 = [0 0 1]𝑇
] [𝑏23 ] = [0] ⇒ 𝐴𝑏
𝑏33 1
Therefore, 𝑏𝑖𝑗 can be calculated from as many equations.

However, LU decomposition can be used to determine 𝐴−1 as well:

𝐴𝑏̅1 = 𝐿𝑈 𝑏̅1 = [1 0 0]𝑇

̅̅̅
𝑑1
or 𝐿 𝑑1 = [1 0 0]𝑇 ⇒ 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 ̅̅̅
𝑑1

So, 𝑈𝑏̅1 = ̅̅̅

𝑑1 ⇒ 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 𝑏̅1
Similarly,
̅̅̅2 = 𝐿𝑈 ̅̅̅
𝐴𝑏 𝑏2 = [0 1 0]𝑇
or 𝐿 ̅̅̅
𝑑2 = [0 1 0]𝑇 ⇒ 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 ̅̅̅
𝑑2

24
 So, ̅̅̅2 = ̅̅̅
𝑈𝑏 𝑑2 ⇒ 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 ̅̅̅
𝑏2 , 𝑠𝑜 𝑓𝑜𝑟𝑡ℎ,
25 5 1
Example: [𝐴] = [ 64 8 1] = 𝐿𝑈
144 12 1
25 5 1
[ 64 8 1]
144 12 1
Apply GE (Forward step)

𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 𝑢𝑠𝑒𝑑 𝑡𝑜 𝑚𝑎𝑘𝑒

𝑠𝑡
25 5 1 1 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑜𝑤𝑠, 𝑏𝑜𝑡𝑡𝑜𝑚
[0 −4.8 0.56 ] ⇒ 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑖𝑣𝑜𝑡𝑎𝑙 𝑟𝑜𝑤 𝑡𝑜 𝑏𝑒 𝑧𝑒𝑟𝑜
144
0 −16.8 −4.36 = 64⁄25 𝑎𝑛𝑑 25
𝑜𝑟 2.56 𝑎𝑛𝑑 5.76

𝑇ℎ𝑖𝑠 𝑖𝑠 𝑎𝑛 𝑈𝑇𝑀 .
25 5 1 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑢𝑠𝑒𝑑 𝑡𝑜 𝑚𝑎𝑘𝑒 𝑡ℎ𝑒
[0 −4.8 −1.56] ⇒ 1𝑠𝑡 𝑒𝑙𝑒𝑚𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑙𝑎𝑠𝑡 𝑟𝑜𝑤
0 0 0.7 𝑡𝑜 𝑏𝑒 𝑧𝑒𝑟𝑜
= 16.8⁄4.8 = 3.5

Therefore,
1 0 0
𝐿 = [2.56 1 0]
5.76 3.5 1
or,
25 5 1 1 0 0 25 5 1
[ 64 8 1] = [2.56 1 0] [ 0 −4.8 −1.56]
144 1 1 5.76 3.5 1 0 0 0.7
A L U
Now, determine 𝐴−1

Step 1: 𝐿𝑈 𝑏̅1 = [1 0 0]𝑇

̅̅̅
𝑑1

25
 1 0 0 𝑑1 1
[2.56 1 0] {𝑑2 } = {0}
5.76 3.5 1 𝑑3 0
L ̅̅
𝒅̅𝟏̅
𝑑1 = 1, 𝑑2 = 0 − 2.56 = −2.56, 𝑑3 = 0 − 5.76 + 3.50 × 2.56 = 3.2
25 5 1 𝑏11 𝑑1
2. [0 ] {
−4.8 −1.56 21𝑏 } = { 𝑑 2}
0 0 0.7 𝑏31 𝑑3
U ̅̅̅
𝒃𝟏

𝑏31 = 3.2⁄0.7 = 4.571, 𝑏21 = (−2.56 − 4.571 × (−1.56))/−4.8 = −0.9524

𝑏11 = (1 − 4.571 − 5 × (−0.9524))/25.0 = 0.04762

or, 𝑏1𝑇 = [0.04762 − 0.9524 4.571]𝑇

1 0 0 𝑑1 0
3. [2.56 1 0] {𝑑2 } = {1} ⇒ 𝑑1 = 0, 𝑑2 = 1, 𝑑3 = −3.5
5.76 3.5 1 𝑑3 0
L ̅𝒅̅̅𝟏̅

25 5 1 𝑏12 0
[0 −4.8 −1.56] {𝑏22 } = { 1 }
0 0 0.7 𝑏32 −3.5
U ̅̅̅
𝒃𝟐
𝑇
𝑏̅2 = [−0.0833 1.417 − 5.0]𝑇
Similarly,
𝑇
𝑏̅3 = [0.03571 − 0.4643 1.429]𝑇

0.04762 −0.0833 0.03571

𝐴−1 = [−0.9524 1.417 −0.4643]
4.571 −0.5 1.429

̅ and programming codes:

̅=𝒃
⇒ Methods so far to solve 𝑨𝑿

(a) GE: To solve a large number of algebraic equations, a programming code is

required. As earlier shown, GE contains two distinct steps: (1) forward
elimination to convert 𝑨 ̅ to an upper triangular matrix (UTM), (2)

26
 Backward/reverse substitution to solve UTM. Therefore, while writing the code;
it is recommended that one clearly distinguishes the two steps as “subroutines”
(# 1 and 2). You will later see that the other methods also often require one or
both of the steps/codes.

(b) GJ: The programming code for GJ is similar to that (code 1) of the forward
elimination of GE, with some modification. The second part (code) of GE, ie.
reverse substitution is not required. The modification or extra step is simple.
First, pivotal elements should be normalized to 1; in addition to subtracting the
modified pivotal row from the rows below the ‘𝑖 𝑡ℎ ’ row of GE, the pivotal row is
also subtracted from the rows above. Therefore, one extra line is included in the
code for GJ.

(c) The code to determine the determinant of the matrix A is the same as that (code
1) for converting A to UTM, ie. the first or forward elimination step of GE.
Further, add a line to determine the ∏𝑛𝑖=1 𝑎𝑖𝑖 product of the diagonal elements of
the UTM ie. one can put the flag to check if any of the 𝑎𝑖𝑖 is zero, the matrix is
singular and cannot be inverted. One can also write a simple code to check the
number # of zeros on the diagonal elements and therefore, determine the rank
of the matrix as (𝑛 − #).

(d) LU method: The programming code (# 3) is similar to that (code 1) of the

forward step of GE, to determine UTM. An extra line in the programming loop is
𝑎
required to store the coefficients 𝑙𝑖𝑘 = (𝑎 𝑖𝑘 ), used to make the elements of the
𝑘𝑘
first column of the UTM to be zero, as the elements of the empty columns of the
LTM.

𝐴 = [𝑙11 ]×[ ]
𝑙21 𝑙22

LTM UTM
(e) Inverse of the matrix: First, the programming code (# 3) of LU is to be used. The
forward substitution on LTM will yield/give 𝑑̅ (𝑐𝑜𝑙𝑢𝑚𝑛 𝑚𝑎𝑡𝑟𝑖𝑥) and the
backward substitution on UTM will yield 𝑏̅ (𝑐𝑜𝑙𝑢𝑚𝑛 𝑚𝑎𝑡𝑟𝑖𝑥)(already available;
# 2). Therefore, one new code/subroutine (# 4) is required for the forward
substitution on LTM, which is not different from that (# 2) for the reverse
substitution on UTM:

 𝐴 = 𝐿𝑈 (𝑐𝑜𝑑𝑒 # 3)
 𝐹𝑜𝑟𝑤𝑎𝑟𝑑 𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛 → 𝑑̅ (𝑐𝑜𝑑𝑒 #4)
 𝐵𝑎𝑐𝑘𝑤𝑎𝑟𝑑 𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛 → 𝑏̅ (𝑐𝑜𝑑𝑒 # 2)

27
 Lecture #06
Thomas Algorithm (Tridiagonal matrix)

𝐴𝑋̅ = 𝑏̅

𝑎11 𝑎12 0 ⋯ 0 0 0 0
𝑋1 𝑏1
𝑎21 𝑎22 𝑎23 ⋯ 0 0 0 0
𝑋2 𝑏2
0 𝑎32 𝑎33 𝑎34 0 ⋯ 0 0
𝑋3 = 𝑏3
0 0 𝑎43 𝑎44 𝑎45 ⋯ 0 0
⋮ ⋮
{𝑋𝑛 } {𝑏𝑛 }
[ 0 0 0 0 0 0 𝑎𝑛𝑛−1 𝑎𝑛𝑛 ]

Tridiagonal matrix has 3 non-zero elements in all of its rows, except in the 1st and last row,
with one element each on the left and right of the diagonal element. The 1st and last rows have
one element on the right and left of the diagonal elements, respectively. It is a banded matrix
around its diagonal:

𝑋1 𝑏1
𝑋2 𝑏2
𝑋3 = 𝑏3
⋮ ⋮
[ ] {𝑋𝑛 } {𝑏𝑛 }

Therefore, a tridiagonal matrix can be represented using single subscripted indices for its
elements:

𝑏1 𝑐1 0 0 ⋯ 0
𝑎2 𝑏2 𝑐2 0 ⋯ 0
0 𝑎3 𝑏3 𝑐3 ⋯ 0
Zero
𝑎4 𝑏4 𝑐4
𝑧𝑒𝑟𝑜
[ 𝑎𝑛 𝑏𝑛 ]

Note that the element in the 𝑖 𝑡ℎ row is represented as (𝑎𝑖 , 𝑏𝑖 , 𝑐𝑖 ). All bs are on the diagonal.

Therefore, 𝐴𝑋̅ = 𝑑̅ , where 𝐴 is a tridiagonal matrix and it represents the following set of linear
algebraic equations:

28
 𝑏1 𝑥1 + 𝑐1 𝑥2 = 𝑑1
𝑎2 𝑥1 + 𝑏2 𝑥2 + 𝑐2 𝑥3 = 𝑑2
𝑎3 𝑥2 + 𝑏3 𝑥3 + 𝑐3 𝑥4 = 𝑑3
𝑎4 𝑥3 + 𝑏4 𝑥4 + 𝑐4 𝑥5 = 𝑑4

𝑎𝑛 𝑥𝑛−1 + 𝑏𝑛 𝑥𝑛 = 𝑑𝑛 }

⇒ You must have noted that for the 𝑖 𝑡ℎ row, 𝑏𝑖 is the diagonal element multiplied with the
variable 𝑥𝑖 in the some row, whereas 𝑎𝑖 is multiplied with 𝑥𝑖−1 (the variable above 𝑖 𝑡ℎ row) and
𝑐𝑖 is multiplied with 𝑥𝑖+1 (the variable below 𝑖 𝑡ℎ row). Naturally, the first and last rows have
only two variables.

⇒(Tridiagonal system is quite common when using Finite Difference 2nd order method to solve
boundary value problems or partial differential 𝑒𝑞 𝑛𝑠 .)

Thomas algorithm to solve such system:

Step 1:
𝑐 𝑑
𝒙𝟏 is eliminated : (𝑏2 − 𝑏1 𝑎2 ) 𝑥2 + 𝑐2 𝑥3 = (𝑑2 − 𝑏1 𝑎2 )
1 1

(by dividing 1st row with 𝑏1 , multiplying with 𝑐1 and subtracting from row 2)
𝑐 𝑑
𝒙𝟐 is eliminated : (𝑏3 − 𝑏2 𝑎3 ) 𝑥3 + 𝑐3 𝑥4 = (𝑑3 − 𝑏2 𝑎3 )
2 2

⋮
𝑐 𝑑
𝒙𝒌−𝟏 is eliminated : (𝑏𝑘 − 𝑏𝑘−1 𝑎𝑘 ) 𝑥𝑘 + 𝑐𝑘 𝑥𝑘+1 = (𝑑𝑘 − 𝑏𝑘−1 𝑎𝑘 )
𝑘−1 𝑘−1

It is clear that;

𝐴𝑙𝑙 𝑎𝑠 𝑎𝑟𝑒 𝑒𝑙𝑖𝑚𝑖𝑛𝑎𝑡𝑒𝑑

𝐴𝑙𝑙 𝑏𝑠 𝑎𝑟𝑒 𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑
}
𝐴𝑙𝑙 𝑐𝑠 𝑎𝑟𝑒 𝑢𝑛𝑎𝑙𝑡𝑒𝑟𝑒𝑑
𝐴𝑙𝑙 𝑑𝑠 𝑎𝑟𝑒 𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑
𝑐 𝑑
Last row: 𝑏𝑛 = 𝑏𝑛 − 𝑏𝑛−1 𝑎𝑛 ; 𝑑𝑛 = 𝑑𝑛 − 𝑏𝑛−1 𝑎𝑛
𝑛−1 𝑛−1

(Note that 𝑎𝑛 has moved to RHS, leaving behind 𝑏𝑛 𝑥𝑛 only on the LHS)

29
(It is important to note that 𝑘 𝑡ℎ row uses the latest modified values of 𝑏𝑘 and 𝑑𝑘 from the
previous (k-1) step. Therefore, there is no need to store the previous values of (a, b, d) while
writing the code)
𝑑𝑛
Back-substitution: 𝑋𝑛 = (𝑛𝑡ℎ 𝑟𝑜𝑤 ℎ𝑎𝑠 𝑜𝑛𝑙𝑦 𝑏𝑛 𝑎𝑛𝑑 𝑑𝑛 )
𝑏𝑛

𝑑𝑛−1 −𝑐𝑛−1 𝑥𝑛
𝑋𝑛−1 = (𝑛 − 1 𝑟𝑜𝑤 ℎ𝑎𝑠 𝑏𝑛−1 , 𝑐𝑛−1 , 𝑎𝑛𝑑 𝑑𝑛−1 )
𝑏𝑛−1

⇒ A pseudo programming code can be written as follows:

Tridiagonal (N, a, b, c, d, X)

𝑑𝑜 𝑖 = 1, 𝑁

𝑎(𝑖) =

𝑏(𝑖) =

𝑐(𝑖) =

𝑑(𝑖) =

end do

𝑑𝑜 𝑖 = 2, 𝑁

𝑐(𝑖 − 1)
𝑏(𝑖) = 𝑏(𝑖) − 𝑎(𝑖)
𝑏(𝑖 − 1)
𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛
𝑑(𝑖 − 1)
𝑑(𝑖) = 𝑑(𝑖) − 𝑎(𝑖)
𝑏(𝑖 − 1)}

end do
𝑑(𝑛)
𝑋𝑛 = 𝑏(𝑛)

𝑑𝑜 (𝑖) = 𝑁 − 1, 1, −1

𝑑(𝑖) − 𝑐(𝑖)𝑋𝑖+1
𝑋𝑖 = } 𝑟𝑒𝑣𝑒𝑟𝑠𝑒 𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛
𝑏(𝑖)

end do

30
 Indirect Methods

GE, GJ, LU decomposition are the direct methods to solve 𝐴𝑋̅ = 𝑏̅. Simple iterations can also
be done to solve a set of algebraic equations, Jacobi and Gauss-Seidel being the two commonly
used indirect methods.

2 1 0 𝑋1 1
Ex. 𝑋
[1 2 1] { 2 } = {2}
0 1 1 𝑋3 4
(1) (1) (1)
Make a guess 𝑋1 , 𝑋2 , 𝑋3

Jacobi Gauss-Seidel
(1) (1)
(2) 1−𝑋2 1−𝑋2
𝑋1 = =
2 2

(1) (1) (2) (1)

(2) 2−𝑋1 −𝑋3 2−𝑋1 −𝑋3
𝑋2 = =
2 2

(1) (2)
(2) 4−𝑋2 4−𝑋2
𝑋3 = =
1 1

(𝑘−1)
Takes all 𝑋𝑠 of the previous iteration Uses the most latest iterated values of 𝑋𝑠 .

In general: 𝐴𝑋̅ = 𝑏̅
(𝑘)
(𝑘+1) 𝑏𝑖 −∑𝑛
𝑗=1 𝑎𝑖𝑗 𝑋𝑗
Jacobi: 𝑋𝑖 = (𝑗 ≠ 𝑖)
𝑎𝑖𝑖 (≠0)

diagonal elements

(𝑘+1) (𝑘)
𝑏𝑖 − ∑𝑖−1
𝑗=1 𝑎𝑖𝑗 𝑋𝑗 − ∑𝑛𝑗=𝑖+1 𝑎𝑖𝑗 𝑋𝑗
𝑢𝑝𝑑𝑎𝑡𝑒𝑑 𝑜𝑙𝑑
(𝑘+1) 𝑙𝑎𝑡𝑒𝑠𝑡 𝑣𝑎𝑙𝑢𝑒𝑠
G-S: 𝑋𝑖 =
𝑣𝑎𝑙𝑢𝑒𝑠

{ 𝑎𝑖𝑖 (≠ 0)

31
At any 𝑖 𝑡ℎ row:

1⋯⋯𝑖 − 1 𝑖 𝑖 + 1⋯⋯ 𝑁

𝑢𝑝𝑑𝑎𝑡𝑒𝑑 𝑜𝑙𝑑
𝑣𝑎𝑙𝑢𝑒𝑠 𝑣𝑎𝑙𝑢𝑒𝑠
(𝑘 + 1) (𝑘)

Schematically,

elements: multiply with the variables 1 to 𝑥𝑖−1

𝑋1
𝑋2
⋮
1 ⋯ 𝑖−1 𝑖 𝑖+1 ⋯ 𝑛 𝑋 𝑖
⋮
⋮
[ ] {𝑋𝑛 }

elements: multiply with the

𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 𝑥𝑖+1 𝑡𝑜 𝑥𝑛 .

G-S (modified)

𝑋𝑖 = 𝜆𝑋𝑖 (𝑛𝑒𝑤) + (1 − 𝜆)𝑋𝑖 (𝑜𝑙𝑑)

𝜆=1 (𝑢𝑛𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑)

0 < 𝜆 < 1 ⇒ 𝑢𝑛𝑑𝑒𝑟 − 𝑟𝑒𝑙𝑎𝑥𝑎𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟

1 < 𝜆 < 2 ⇒ 𝑜𝑣𝑒𝑟 − 𝑟𝑒𝑙𝑎𝑥𝑎𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟

(′𝜆′ 𝑖𝑠 𝑜𝑓𝑡𝑒𝑛 𝑎𝑠𝑠𝑢𝑚𝑒𝑑 𝑑𝑒𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑢𝑝𝑜𝑛 𝑔𝑢𝑒𝑠𝑒𝑠, 𝑒𝑡𝑐)

Look differently! : LDU method

32
 𝑎11 𝑎12 ⋯ 𝑎1𝑛 0 𝑎11
⋮ 𝑎21 0 ⋱
⋮ = 0 + ⋱ +
⋮ 0
[𝑎𝑛1 ⋯ ⋯ 𝑎𝑛𝑛 ]𝑛×𝑛 [𝑎𝑛1 ⋯ 𝑎𝑛−1 ] [ 𝑎𝑛𝑛 ]
0 ⋯ ⋯ 𝑎1𝑛
0
0
𝑎𝑛−1𝑛
[ 0 ]
strictly lower 𝐷 𝑠𝑡𝑟𝑖𝑐𝑡𝑙𝑦 𝑢𝑝𝑝𝑒𝑟
𝑡𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑥 + (𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 + 𝑡𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑥
(𝐿) 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑜𝑛𝑙𝑦) (𝑈)

𝐴=𝐿+𝐷+𝑈

𝐴𝑋̅ = 𝑏̅

(𝐿 + 𝐷 + 𝑈)𝑋̅ = 𝑏⃗⃗ ⇒ 𝐷𝑋̅ = 𝑏̅ − 𝐿𝑋̅ − 𝑈𝑋̅

𝑏̅ − 𝐿𝑋̅ − 𝑈𝑋̅
𝑜𝑟 𝑋̅ =
𝐷
Jacobi:

𝑏̅ − (𝐿 + 𝑈)𝑋̅ (𝑘)
𝑋̅ (𝑘+1) =
𝐷
(𝑘)
(𝑘+1)
𝑏𝑖 − ∑𝑛𝑗=1 𝑋𝑗 𝑎𝑖𝑗
𝑜𝑟 𝑋𝑖 = ; 𝑗 ≠ 𝑖 (𝑗 = 𝑖 𝑤𝑖𝑙𝑙 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑒𝑙𝑒𝑚𝑒𝑛𝑡)
𝑎𝑖𝑖

𝐆 − 𝐒 ∶ (𝐿 + 𝐷)𝑋̅ = 𝑏̅ − 𝑈𝑋̅

(𝐿 + 𝐷)𝑋̅ (𝑘+1) = 𝑏̅ − 𝑈𝑋̅ (𝑘)

𝐷𝑋̅ (𝑘+1) = 𝑏̅ − 𝑈𝑋̅ (𝑘) − 𝐿𝑋̅ (𝑘+1)

(𝑘+1)
(𝑘+1) 𝑏𝑖 −∑𝑖−1
𝑗=1 𝑋𝑖 𝑎𝑖𝑗 −∑𝑛 𝑘
𝑗=𝑖+1 𝑋𝑖 𝑎𝑖𝑗
𝑋𝑖 =
𝑎𝑖𝑖
(𝑠𝑎𝑚𝑒 𝑎𝑠 𝑏𝑒𝑓𝑜𝑟𝑒 ) ( 𝑎𝑖𝑖 ≠ 0 )

Both methods will lead to the same solutions, with different # of iterations.

33
 Lecture #07
Homogeneous linear algebraic equations

-Represents a special class of problems, also known as the Eigenvalue or Characteristic type of
problems

- Mathematically represented as 𝐴𝑥̅ = 0 or 𝑏̅ = 0 (𝑛𝑢𝑙𝑙 𝑣𝑒𝑐𝑡𝑜𝑟)

𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ ⋯ 𝑎1𝑛 𝑥𝑛 = 0

⋮
𝑜𝑟 ( )
⋮
𝑎𝑛1 𝑥1 + 𝑎𝑛2 𝑥2 + ⋯ ⋯ 𝑎𝑛𝑛 𝑥𝑛 = 0

Naturally, a trivial solution is 𝑥̅ = 0 . In the simplest geometrical term, two straight lines or
three planes intersect at the origin:

𝑥1 − 𝑥2 = 0
}⇒ 𝑥2
𝑥1 − 0.3𝑥2 = 0

𝑥1

⇒ A more interesting problem to solve is in seeking a non-trivial solution when 𝑑𝑒𝑡(𝐴) =

0 𝑜𝑟 𝑟 < 𝑚. Such simultaneous set of homogeneous linear equations when 𝑑𝑒𝑡(𝐴) = 0 is
common in several engineering and mechanics applications, and also in the initial and boundary
value problems. Such situation is better known as the eigenvalue problem and is represented by
eigenvalues and the corresponding eigenvectors.

Compare this situation to the earlier discussed set of non-homogeneous linear algebraic
equation 𝐴𝑋̅ = 𝑏̅ , where we sought a unique solution, when 𝑑𝑒𝑡|𝐴| ≠ 0. Comparatively,

Non-homogeneous Homogeneous equations

𝐴𝑋̅ = 𝑏̅ 𝐴𝑋̅ = 0

det(𝐴) ≠ 0 det(𝐴) = 0

𝑟=𝑚 𝑟<𝑚

Unique solution Non-trivial solution

𝐸𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚

34
Let us look at

𝐴𝑋̅ = 𝜆𝑋̅ 1 where A is the coefficient matrix and ′𝜆′ is a non-zero number.

matrix scalar quantity

or (𝐴 − 𝜆𝐼)𝑋̅ = 0 2 where 𝐼 is an identity matrix

1 0 0
𝐼 = [0 1 0 ]
0 0 1
Equation 1 or 2 represents eigenvalue problem and such equations commonly occur in several
𝑑𝑦̅
dynamic studies of distillation, adsorption and CSTR, viz. in solving = 0. Equation (2) can
𝑑𝑡
also be written as

(𝑎11 − 𝜆)𝑋1 + 𝑎12 𝑋2 + ⋯ ⋯ 𝑎1𝑛 𝑋𝑛 = 0

𝑎21 𝑋1 + (𝑎22 − 𝜆)𝑋2 + ⋯ ⋯ 𝑎2𝑛 𝑋𝑛 = 0
⋮
⋮
{𝑎𝑛1 𝑋1 + 𝑎𝑛2 𝑋2 + ⋯ ⋯ (𝑎𝑛𝑛 − 𝜆)𝑋𝑛 = 0}

or

𝑎11 𝑎12 ⋯ 𝑎1𝑛 𝑋1 𝑋1

⋮ 𝑋2 𝑋2
[ ⋮ ]{ } = 𝜆{ }
⋮ ⋮
𝑎𝑛1 𝑎𝑛2 ⋯ 𝑎𝑛𝑛 𝑋𝑛 𝑋𝑛

𝑐𝑜𝑙𝑢𝑚𝑛 𝑎 𝑠𝑐𝑎𝑙𝑎𝑟 𝑐𝑜𝑙𝑢𝑚𝑛

𝐴(𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑚𝑎𝑡𝑟𝑖𝑥)
𝑣𝑒𝑐𝑡𝑜𝑟 𝑞𝑢𝑎𝑛𝑡𝑖𝑡𝑦 𝑣𝑒𝑐𝑡𝑜𝑟

Re-visit 𝐴𝑋̅ = 𝜆𝑋̅ - eigenvalue problem

Most vectors 𝑋̅ will not satisfy such an equation. A common vector (𝑖𝑋1 + 𝑗𝑋2 ) will change
direction and magnitude to (𝑖𝑏1 + 𝑗𝑏2 ) on the transformation by A. Only certain special
vectors 𝑋̅, called eigenvectors, corresponding to only special numbers, 𝜆 (𝑒𝑖𝑡ℎ𝑒𝑟 + 𝑣𝑒 𝑜𝑟 −
𝑣𝑒), called eigenvalues, will satisfy the above equation. In such case, the eigenvector 𝑋̅ does not
change its direction or does not rotate when transformed by the coefficient matrix A, but is
only scaled by the eigenvalue 𝜆. See the geometrical representation below:

35
 𝑋2

𝑋2
𝑋̅
𝑏̅ 𝜆𝑋⃗
𝑋̅

𝑋1 𝑋1

𝐴𝑋̅ = 𝑏̅ 𝐴𝑋̅ = 𝜆𝑋̅

(𝑛𝑜𝑡 𝑎𝑛 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚) (𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚)

(𝑋̅ 𝑖𝑠 𝑡ℎ𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑒𝑡𝑜𝑟 𝑜𝑓 𝐴; 𝜆
𝑖𝑠 𝑡ℎ𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝐴)

It turns out that (𝑛 × 𝑛) matrix 𝐴 will give ‘𝑛’ 𝜆𝑠 (𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠), and each eigenvalue will give
‘𝑛’ eigenvectors that will also be linearly independent. As an example,

𝑎11 𝑎12 𝜆1 → {𝑋}1

[𝑎
21 𝑎22 ]2×2 will have 2 𝜆𝑠 (𝜆1 & 𝜆2 ) ⇒ 𝜆2 → {𝑋}2

Both vectors {𝑋}1 , & {𝑋}2 will satisfy 𝐴𝑋̅ = 𝜆𝑋̅. They will also be linearly independent so that
they can be used as a basis for the space-description. They may be orthogonal or non-
orthogonal.

𝑋2

𝜆1
𝜆2

𝑋1

In the figure above, dotted and solid lines represent two vectors scaled by the respective
eigenvalues1 and 2, respectively.

Similarly, (3 × 3) A matrix will give 3𝜆𝑠 (𝜆1 , 𝜆2 , 𝜆3 ). Each 𝜆 will give three independent
vectors {𝑋}1 , {𝑋}2 , {𝑋}3 satisfying 𝐴𝑋̅ = 𝜆𝑋̅.

36
 𝑑2 𝑋1
𝑚1 = −𝐾𝑋1 + 𝐾(𝑋2 − 𝑋1 ) 𝑋
𝑑𝑡 2 𝑋̅ = { 1 }
𝑑 2 𝑋2 𝑋2
𝑚2 = −𝐾(𝑋2 − 𝑋1 ) − 𝐾𝑋2 }
𝑑𝑡 2
𝑋𝑖 = 𝐴𝑖 𝑠𝑖𝑛(𝜔𝑡) (𝑎𝑠𝑠𝑢𝑚𝑒 𝑎 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛)

( 𝑜𝑟 𝑋̅ = 𝐴̅ 𝑠𝑖𝑛(𝜔𝑡))

Therefore,

𝑋𝑖 ′′ = −𝐴𝑖 𝜔2 𝑠𝑖𝑛𝜔𝑡

Substitute,
2𝐾 𝐾
(𝑚 − 𝜔2 ) 𝐴1 − 𝑚 𝐴2 = 0
1 1
𝐾 2𝐾
}
2
− 𝑚 𝐴1 + (𝑚 − 𝜔 ) 𝐴2 = 0
2 2

𝑜𝑟

2𝐾 𝐾
( − 𝜔2 ) −
𝑚1 𝑚1 𝐴1
|| 𝐾 2𝐾 || {𝐴 } = 0
2
− ( − 𝜔2 )
𝑚2 𝑚2

Therefore, the characteristic equation describes an eigenvalue problem (or spring dynamics is
an eigenvalue problem). It has assumed the form of

(𝐴 − 𝜔2 𝐼)𝑋̅ = 0
2𝐾 𝐾
−𝑚 𝑋
𝑚1
where 𝜔2 = 𝜆, 𝐴=| 1
| and 𝑋̅ = { 1 }
−𝑚
𝐾 2𝐾 𝑋2
2 𝑚2

𝑒𝑖𝑔𝑒𝑛 𝑣𝑎𝑙𝑢𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑒𝑐𝑡𝑜𝑟

Example (From Kreysig's Math book)

𝑚1 = 𝑚2 = 40 𝑘𝑔 ; K = 200 𝑁⁄𝑚

For a non-trivial solution 𝑑𝑒𝑡(𝐴 − 𝜔2 𝐼)𝑋̅ = 0

𝑒𝑞 𝑛𝑠 : (10 − 𝜔2 )𝐴1 − 5𝐴2 = 0

−5𝐴1 + (10 − 𝜔2 )𝐴2 = 0

44
 𝑑𝑒𝑡( ) : (10 − 𝜔2 )2 − 25 = 0

𝜆1 = 15, 𝜆2 = 5 𝑜𝑟 𝜔1 = √𝜆1 , 𝜔2 = √𝜆2

1
𝜆1 = 15 𝐴1 = −𝐴2 𝑜𝑟 𝐴(1) = { }
−1
1
𝜆2 = 5 𝐴1 = 𝐴2 𝑜𝑟 𝐴(2) = { }
1
General solution:

𝑋̅ = 𝐶1 𝐴̅(1) 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝐴̅(2) 𝑠𝑖𝑛𝜔2 𝑡

𝑋 1 1
𝑜𝑟 { 1 } = 𝐶1 { } 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 { } 𝑠𝑖𝑛𝜔2 𝑡
𝑋2 −1 1
𝑜𝑟 𝑋1 = 𝐶1 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝑠𝑖𝑛𝜔2 𝑡
}
𝑎𝑛𝑑 𝑋2 = −𝐶1 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝑠𝑖𝑛𝜔2 𝑡

Graphical representation of block oscillations:

𝜆1 =15 𝜆2 =5
𝐴1 𝐴2 𝐴1 𝐴2
𝜔1 =√15 𝜔2 =√5

45
 Lecture #09
One more example of eigenvalue or characteristic type of problem:

Let us revisit Jacobi iteration to invert a matrix.

𝐴𝑋̅ = 𝑏̅

⇒ (𝐿 + 𝐷 + 𝑈)𝑋̅ = 𝑏̅

⇒ 𝐷𝑋̅ = 𝑏̅ − (𝐿 + 𝑈)𝑋̅

𝑜𝑟 𝑋̅ = 𝐷−1 𝑏̅ − 𝐷−1 (𝐿 + 𝑈)𝑋̅

or 𝑋̅ = 𝑆𝑋̅ + 𝐶̅

𝑜𝑟 𝑋̅ (𝑘+1) = 𝑆𝑋̅ (𝑘) + 𝐶̅ ∶ 𝐽𝑎𝑐𝑜𝑏𝑖 𝐼𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑓𝑜𝑟𝑚𝑢𝑙𝑎

S is called stationary matrix because very often, matrix 𝐴 𝑜𝑟 (𝐿, 𝐷, 𝑈) do not change and are
fixed. It is 𝑏̅ (right hand side) that varies from one problem to the other, as a forcing function.

Define, error vector 𝑒̅ (𝑘) = 𝑋̅ (𝑘) − 𝑋̅

Approximate
𝑇𝑟𝑢𝑒 𝑣𝑎𝑙𝑢𝑒
𝑣𝑎𝑙𝑢𝑒 𝑎𝑡 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛
#𝑘
Similarly,

𝑒̅ (𝑘+1) = 𝑋̅ (𝑘+1) − 𝑋̅

= 𝑆(𝑋̅ (𝑘) − 𝑋̅) = 𝑆𝑒̅ (𝑘)

𝑜𝑟 𝑒̅ (𝑘+1) = 𝑆𝑒̅ (𝑘) (𝑦𝑜𝑢 𝑠ℎ𝑜𝑢𝑙𝑑 𝑟𝑒𝑎𝑙𝑖𝑧𝑒 𝑡ℎ𝑎𝑡 𝑖𝑡

𝑜𝑟 𝑒̅ (𝑘) = 𝑆 𝑘 𝑒̅ (0) 𝑖𝑠 𝑎𝑛 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚⁄𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
𝐴𝑋̅ = 𝜆𝑋̅ 𝑤𝑖𝑡ℎ 𝜆 = 1)

For 𝑘 → ∞ 𝑆 𝑘 → 0 𝑓𝑜𝑟 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑐𝑒 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑒̅ (𝑘) → 0 𝑎𝑠 𝑒 (0) ≠ 0

Therefore, criteria for the convergence: 𝑆 𝑘 → 0 𝑎𝑠 𝑘 → ∞

Let {𝜈𝑗 }𝑗=1,𝑛 be the eigenvectors corresponding to the 𝜆𝑗 eigenvalue of the stationary matrix

46
𝑆(𝑛 × 𝑛). Recall eigenvectors are linearly independent. So, they can form the basis of n
dimensional space. Alternatively, any vector can be represented as the linear combination of the
eigenvectors.

Therefore,

where Sνj = λj νj
𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗
or Sν1 = λ1 ν1 , etc.

𝑒 (1) = 𝑆𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗

𝑒 (2) = 𝑆 2 𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗 2
(𝑢𝑠𝑒 𝑆𝜈𝑗 = 𝜆𝑗 𝜈𝑗 )
⋮
𝑒 (𝑘) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗 𝑘

For convergence, 𝑘 → ∞ 𝑒 (𝑘) → 0 𝑜𝑟 |𝜆𝑗 | < 1 (𝑗 = 1 ⋯ 𝑛)

For a 𝑛 × 𝑛 S matrix, there will be 𝜆1 ⋯ 𝜆𝑛 eigenvalues.

If |𝜆𝑚𝑎𝑥 | < 1 𝑡ℎ𝑒𝑛 𝑎𝑙𝑙 𝜆𝑗 < 1

Therefore, for Jacobi Iteration to converge: |𝜆𝑚𝑎𝑥 | < 1

(𝑀𝑎𝑥 (𝜆𝑗 ) ≡ 𝑠𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝑟𝑎𝑑𝑖𝑢𝑠 𝜌(𝑆) < 1)

There is the Gershgorin Theorem (Check the book by Strang) which states that in such case

∑|𝑎𝑖𝑗 | < |𝑎𝑖𝑖 |

𝑛𝑜𝑛 − 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙

𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠
Such matrix is said to be diagonally strong. It is a desirable characteristic of a matrix to be
inverted. Compare two extreme situations:

and

non-zero

Note that zero or no iterations are required to solve an identity matrix! Also, a diagonally
strong matrix will require relatively fewer number of iterations. On the other hand, the above
right hand matrix is a singular matrix and cannot be inverted. Similarly, a matrix with smaller

47
value-numbers on its diagonal relative to the other elements in the same row will require a
relatively larger number of iterations to converge.

Power Method

The previous examples have shown that the ‘dynamics’ of a 1st order system can be
inspected or characterized by determining 𝜆𝑚𝑎𝑥 𝑎𝑛𝑑 𝜆𝑚𝑖𝑛 , instead of all 𝜆𝑠 . Therefore, a 𝑛 × 𝑛
matrix will have 𝑛𝜆𝑠 , and the overall response of the system will be the least sensitive to
𝜆𝑚𝑖𝑛 and most sensitive to 𝜆𝑚𝑎𝑥 . The response corresponding to the other 𝜆𝑠 will be
intermediate. Power method is commonly used to determine 𝜆𝑚𝑎𝑥 and also, 𝜆𝑚𝑖𝑛 .

If [𝐴]𝑛×𝑛 matrix has 𝑛 𝜆𝑠(𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠), then one can write |𝜆1 | > |𝜆2 | > ⋯ ⋯ |𝜆𝑛 |

𝜆𝑚𝑎𝑥 𝜆𝑚𝑖𝑛

̅ = λmax ̅
And AX ̅ = λmin ̅
X and AX X

Method: Make a guess of 𝑋 𝑇 = [1 0 1] 𝑜𝑟 [1 1 0] (say, for a 3 × 3 matrix)

Then 𝐴𝑋̅ (0) ⇒ 𝑋̅ (1)

𝑋̅ (1) 𝐴𝑋 (0)
Use 𝑋̅ (1) = ‖𝑋̅ (1) ‖
= ‖𝐴𝑋 (0)‖

magnitude (you are scaling or

( normalizing the vector, )
2 2 2
i. e. a⃗⃗⁄|a| or a⃗⃗⁄√a1 + a2 + a3

𝐴𝑋̅ (𝑘−1)
𝑋̅ (𝑘) = ‖𝐴𝑋̅ (𝑘−1)‖ , 𝑘 = 1, 2,3 (# 𝑜𝑓 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠)

Can be shown that when 𝑘 → ∞ ‖𝐴𝑋̅ (𝑘−1) ‖ → |𝜆𝑚𝑎𝑥 |

In other words, after sufficient # of iterations, one obtains maximum eigenvalue and the vector
𝑋̅ (𝑘−1) when transformed by ‘A’ matrix does not rotate and is just scaled by 𝜆𝑚𝑎𝑥 , i.e.

𝐴𝑋̅ (𝑘−1) = 𝜆𝑚𝑎𝑥 𝑋̅ (𝑘)

In other wors, ̅
X (k−1) and ̅
X (k) have the same directions.

𝜆𝑚𝑎𝑥 𝑋̅ (𝑘+1)

𝑋̅ (𝑘)

48
 0 2 3
Example: Determine 𝜆𝑚𝑎𝑥 of 𝐴 = [−10 −1 2]
−2 4 7
1
𝜈 (1) = {0} ≡ [1 0 0]𝑇 (1𝑠𝑡 𝑔𝑢𝑒𝑠𝑠)
0
0 2 3 1 0
[−10 −1 2]{0} {−10} 0.0000
(2) −2 4 7 0 −2
𝜈 = ‖𝐴𝜈 (1) ‖
= = {−0.9806}
√100+4
−0.1961

|𝜆| = √104 = 10.198

0.0000 −2.5495
[A]{−0.9806} { 0.5884 }
−0.1961 −5.2951
𝜈 (3) = ‖Aν(2) ‖
= √2.54952 +.58842 +5.29512
={ }

𝜆2 = 5.9063

After 20 iterations,

−0.3714
(20)
𝑋 = { 0.55698} with λ(20) = 3.0014
0.7428
You will see a gradual convergence from 10.198 to 3.0014.

The basics remains the same:

𝐴𝑋 (19) = 𝜆𝑚𝑎𝑥 𝑋 (20) and 𝑋 (19) 𝑎𝑛𝑑 𝑋 (20) will have approximately the same direction.

⇒ It can be shown that 𝝀𝒎𝒊𝒏 𝐨𝐟 𝐀 = 𝟏⁄𝝀 𝒘𝒉𝒆𝒓𝒆 𝝀𝒎𝒂𝒙 is the highest eigenvalue of the
𝒎𝒂𝒙
inverse of matrix 𝑨 𝒐𝒓 𝑨−𝟏 :

̅ = λmax X
AX ̅

̅ = A−1 (λmax X
or X ̅) = λmax (A−1 X
̅)

̅ = (1⁄
or A−1 X ̅
)X
λ max

̅ ≡ λ′max X
or BX ̅

where B is A−1 and λmin of A is (1⁄ ) of B where λ′max is the highest λ of B.

λ′max

49
The λmax to λmin ratio is known as stiffness ratio and indicates the stiffness (maximum rate of
change in the functional value corresponding to λmax relative to minimum rate corresponding to
λmin) of the system. A ratio > 10 indicates the system to be stiff requiring a fine step size for
calculations (to be discussed later)

Notes (From Strang's book):

If X is an eigenvector of A corresponding to 𝜆𝑚𝑎𝑥 and A is invertible, then X is also an

eigenvector of 𝐴−1 corresponding to the inverse of its 𝜆𝑚𝑎𝑥 .

Eigenvectors corresponding to different 𝜆𝑠 are linearly independent.

A matrix and its transpose have the same eigenvalues.

A matrix is singular if and only if it has zero 𝜆 . A non-singular matrix has all non zero 𝜆𝑠 .

∑ 𝜆𝑖 = 𝑡𝑟𝑎𝑐𝑒 𝑜𝑓 𝐴 = ∑𝑛𝑖=1 𝑎𝑖𝑖

∏ 𝜆𝑖 = det(𝐴)

The 𝜆𝑠 of an upper or lower triangular matrix are the elements on its main diagonal.

[It also follows that det A = det (UTM of A) = ∏(𝑎𝑖𝑖 )𝑈𝑇𝑀 ]

Quiz 1

50
 𝑑2 𝑋1
𝑚1 = −𝐾𝑋1 + 𝐾(𝑋2 − 𝑋1 ) 𝑋
𝑑𝑡 2 𝑋̅ = { 1 }
𝑑 2 𝑋2 𝑋2
𝑚2 = −𝐾(𝑋2 − 𝑋1 ) − 𝐾𝑋2 }
𝑑𝑡 2
𝑋𝑖 = 𝐴𝑖 𝑠𝑖𝑛(𝜔𝑡) (𝑎𝑠𝑠𝑢𝑚𝑒 𝑎 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛)

( 𝑜𝑟 𝑋̅ = 𝐴̅ 𝑠𝑖𝑛(𝜔𝑡))

Therefore,

𝑋𝑖 ′′ = −𝐴𝑖 𝜔2 𝑠𝑖𝑛𝜔𝑡

Substitute,
2𝐾 𝐾
(𝑚 − 𝜔2 ) 𝐴1 − 𝑚 𝐴2 = 0
1 1
𝐾 2𝐾
}
2
− 𝑚 𝐴1 + (𝑚 − 𝜔 ) 𝐴2 = 0
2 2

𝑜𝑟

2𝐾 𝐾
( − 𝜔2 ) −
𝑚1 𝑚1 𝐴1
|| 𝐾 2𝐾 || {𝐴 } = 0
2
− ( − 𝜔2 )
𝑚2 𝑚2

Therefore, the characteristic equation describes an eigenvalue problem (or spring dynamics is
an eigenvalue problem). It has assumed the form of

(𝐴 − 𝜔2 𝐼)𝑋̅ = 0
2𝐾 𝐾
−𝑚 𝑋
𝑚1
where 𝜔2 = 𝜆, 𝐴=| 1
| and 𝑋̅ = { 1 }
−𝑚
𝐾 2𝐾 𝑋2
2 𝑚2

𝑒𝑖𝑔𝑒𝑛 𝑣𝑎𝑙𝑢𝑒 𝑒𝑖𝑔𝑒𝑛𝑣𝑒𝑐𝑡𝑜𝑟

Example (From Kreysig's Math book)

𝑚1 = 𝑚2 = 40 𝑘𝑔 ; K = 200 𝑁⁄𝑚

For a non-trivial solution 𝑑𝑒𝑡(𝐴 − 𝜔2 𝐼)𝑋̅ = 0

𝑒𝑞 𝑛𝑠 : (10 − 𝜔2 )𝐴1 − 5𝐴2 = 0

−5𝐴1 + (10 − 𝜔2 )𝐴2 = 0

44
 𝑑𝑒𝑡( ) : (10 − 𝜔2 )2 − 25 = 0

𝜆1 = 15, 𝜆2 = 5 𝑜𝑟 𝜔1 = √𝜆1 , 𝜔2 = √𝜆2

1
𝜆1 = 15 𝐴1 = −𝐴2 𝑜𝑟 𝐴(1) = { }
−1
1
𝜆2 = 5 𝐴1 = 𝐴2 𝑜𝑟 𝐴(2) = { }
1
General solution:

𝑋̅ = 𝐶1 𝐴̅(1) 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝐴̅(2) 𝑠𝑖𝑛𝜔2 𝑡

𝑋 1 1
𝑜𝑟 { 1 } = 𝐶1 { } 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 { } 𝑠𝑖𝑛𝜔2 𝑡
𝑋2 −1 1
𝑜𝑟 𝑋1 = 𝐶1 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝑠𝑖𝑛𝜔2 𝑡
}
𝑎𝑛𝑑 𝑋2 = −𝐶1 𝑠𝑖𝑛𝜔1 𝑡 + 𝐶2 𝑠𝑖𝑛𝜔2 𝑡

Graphical representation of block oscillations:

𝜆1 =15 𝜆2 =5
𝐴1 𝐴2 𝐴1 𝐴2
𝜔1 =√15 𝜔2 =√5

45
 Lecture #09
One more example of eigenvalue or characteristic type of problem:

Let us revisit Jacobi iteration to invert a matrix.

𝐴𝑋̅ = 𝑏̅

⇒ (𝐿 + 𝐷 + 𝑈)𝑋̅ = 𝑏̅

⇒ 𝐷𝑋̅ = 𝑏̅ − (𝐿 + 𝑈)𝑋̅

𝑜𝑟 𝑋̅ = 𝐷−1 𝑏̅ − 𝐷−1 (𝐿 + 𝑈)𝑋̅

or 𝑋̅ = 𝑆𝑋̅ + 𝐶̅

𝑜𝑟 𝑋̅ (𝑘+1) = 𝑆𝑋̅ (𝑘) + 𝐶̅ ∶ 𝐽𝑎𝑐𝑜𝑏𝑖 𝐼𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑓𝑜𝑟𝑚𝑢𝑙𝑎

S is called stationary matrix because very often, matrix 𝐴 𝑜𝑟 (𝐿, 𝐷, 𝑈) do not change and are
fixed. It is 𝑏̅ (right hand side) that varies from one problem to the other, as a forcing function.

Define, error vector 𝑒̅ (𝑘) = 𝑋̅ (𝑘) − 𝑋̅

Approximate
𝑇𝑟𝑢𝑒 𝑣𝑎𝑙𝑢𝑒
𝑣𝑎𝑙𝑢𝑒 𝑎𝑡 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛
#𝑘
Similarly,

𝑒̅ (𝑘+1) = 𝑋̅ (𝑘+1) − 𝑋̅

= 𝑆(𝑋̅ (𝑘) − 𝑋̅) = 𝑆𝑒̅ (𝑘)

𝑜𝑟 𝑒̅ (𝑘+1) = 𝑆𝑒̅ (𝑘) (𝑦𝑜𝑢 𝑠ℎ𝑜𝑢𝑙𝑑 𝑟𝑒𝑎𝑙𝑖𝑧𝑒 𝑡ℎ𝑎𝑡 𝑖𝑡

𝑜𝑟 𝑒̅ (𝑘) = 𝑆 𝑘 𝑒̅ (0) 𝑖𝑠 𝑎𝑛 𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚⁄𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
𝐴𝑋̅ = 𝜆𝑋̅ 𝑤𝑖𝑡ℎ 𝜆 = 1)

For 𝑘 → ∞ 𝑆 𝑘 → 0 𝑓𝑜𝑟 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑐𝑒 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑒̅ (𝑘) → 0 𝑎𝑠 𝑒 (0) ≠ 0

Therefore, criteria for the convergence: 𝑆 𝑘 → 0 𝑎𝑠 𝑘 → ∞

Let {𝜈𝑗 }𝑗=1,𝑛 be the eigenvectors corresponding to the 𝜆𝑗 eigenvalue of the stationary matrix

46
𝑆(𝑛 × 𝑛). Recall eigenvectors are linearly independent. So, they can form the basis of n
dimensional space. Alternatively, any vector can be represented as the linear combination of the
eigenvectors.

Therefore,

where Sνj = λj νj
𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗
or Sν1 = λ1 ν1 , etc.

𝑒 (1) = 𝑆𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗

𝑒 (2) = 𝑆 2 𝑒 (0) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗 2
(𝑢𝑠𝑒 𝑆𝜈𝑗 = 𝜆𝑗 𝜈𝑗 )
⋮
𝑒 (𝑘) = ∑𝑛𝑗=1 𝑐𝑗 𝜈𝑗 𝜆𝑗 𝑘

For convergence, 𝑘 → ∞ 𝑒 (𝑘) → 0 𝑜𝑟 |𝜆𝑗 | < 1 (𝑗 = 1 ⋯ 𝑛)

For a 𝑛 × 𝑛 S matrix, there will be 𝜆1 ⋯ 𝜆𝑛 eigenvalues.

If |𝜆𝑚𝑎𝑥 | < 1 𝑡ℎ𝑒𝑛 𝑎𝑙𝑙 𝜆𝑗 < 1

Therefore, for Jacobi Iteration to converge: |𝜆𝑚𝑎𝑥 | < 1

(𝑀𝑎𝑥 (𝜆𝑗 ) ≡ 𝑠𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝑟𝑎𝑑𝑖𝑢𝑠 𝜌(𝑆) < 1)

There is the Gershgorin Theorem (Check the book by Strang) which states that in such case

∑|𝑎𝑖𝑗 | < |𝑎𝑖𝑖 |

𝑛𝑜𝑛 − 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙 𝑑𝑖𝑎𝑔𝑜𝑛𝑎𝑙

𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠
Such matrix is said to be diagonally strong. It is a desirable characteristic of a matrix to be
inverted. Compare two extreme situations:

and

non-zero

Note that zero or no iterations are required to solve an identity matrix! Also, a diagonally
strong matrix will require relatively fewer number of iterations. On the other hand, the above
right hand matrix is a singular matrix and cannot be inverted. Similarly, a matrix with smaller

47
value-numbers on its diagonal relative to the other elements in the same row will require a
relatively larger number of iterations to converge.

Power Method

The previous examples have shown that the ‘dynamics’ of a 1st order system can be
inspected or characterized by determining 𝜆𝑚𝑎𝑥 𝑎𝑛𝑑 𝜆𝑚𝑖𝑛 , instead of all 𝜆𝑠 . Therefore, a 𝑛 × 𝑛
matrix will have 𝑛𝜆𝑠 , and the overall response of the system will be the least sensitive to
𝜆𝑚𝑖𝑛 and most sensitive to 𝜆𝑚𝑎𝑥 . The response corresponding to the other 𝜆𝑠 will be
intermediate. Power method is commonly used to determine 𝜆𝑚𝑎𝑥 and also, 𝜆𝑚𝑖𝑛 .

If [𝐴]𝑛×𝑛 matrix has 𝑛 𝜆𝑠(𝑒𝑖𝑔𝑒𝑛𝑣𝑎𝑙𝑢𝑒𝑠), then one can write |𝜆1 | > |𝜆2 | > ⋯ ⋯ |𝜆𝑛 |

𝜆𝑚𝑎𝑥 𝜆𝑚𝑖𝑛

̅ = λmax ̅
And AX ̅ = λmin ̅
X and AX X

Method: Make a guess of 𝑋 𝑇 = [1 0 1] 𝑜𝑟 [1 1 0] (say, for a 3 × 3 matrix)

Then 𝐴𝑋̅ (0) ⇒ 𝑋̅ (1)

𝑋̅ (1) 𝐴𝑋 (0)
Use 𝑋̅ (1) = ‖𝑋̅ (1) ‖
= ‖𝐴𝑋 (0)‖

magnitude (you are scaling or

( normalizing the vector, )
2 2 2
i. e. a⃗⃗⁄|a| or a⃗⃗⁄√a1 + a2 + a3

𝐴𝑋̅ (𝑘−1)
𝑋̅ (𝑘) = ‖𝐴𝑋̅ (𝑘−1)‖ , 𝑘 = 1, 2,3 (# 𝑜𝑓 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠)

Can be shown that when 𝑘 → ∞ ‖𝐴𝑋̅ (𝑘−1) ‖ → |𝜆𝑚𝑎𝑥 |

In other words, after sufficient # of iterations, one obtains maximum eigenvalue and the vector
𝑋̅ (𝑘−1) when transformed by ‘A’ matrix does not rotate and is just scaled by 𝜆𝑚𝑎𝑥 , i.e.

𝐴𝑋̅ (𝑘−1) = 𝜆𝑚𝑎𝑥 𝑋̅ (𝑘)

In other wors, ̅
X (k−1) and ̅
X (k) have the same directions.

𝜆𝑚𝑎𝑥 𝑋̅ (𝑘+1)

𝑋̅ (𝑘)

48
 0 2 3
Example: Determine 𝜆𝑚𝑎𝑥 of 𝐴 = [−10 −1 2]
−2 4 7
1
𝜈 (1) = {0} ≡ [1 0 0]𝑇 (1𝑠𝑡 𝑔𝑢𝑒𝑠𝑠)
0
0 2 3 1 0
[−10 −1 2]{0} {−10} 0.0000
(2) −2 4 7 0 −2
𝜈 = ‖𝐴𝜈 (1) ‖
= = {−0.9806}
√100+4
−0.1961

|𝜆| = √104 = 10.198

0.0000 −2.5495
[A]{−0.9806} { 0.5884 }
−0.1961 −5.2951
𝜈 (3) = ‖Aν(2) ‖
= √2.54952 +.58842 +5.29512
={ }

𝜆2 = 5.9063

After 20 iterations,

−0.3714
(20)
𝑋 = { 0.55698} with λ(20) = 3.0014
0.7428
You will see a gradual convergence from 10.198 to 3.0014.

The basics remains the same:

𝐴𝑋 (19) = 𝜆𝑚𝑎𝑥 𝑋 (20) and 𝑋 (19) 𝑎𝑛𝑑 𝑋 (20) will have approximately the same direction.

⇒ It can be shown that 𝝀𝒎𝒊𝒏 𝐨𝐟 𝐀 = 𝟏⁄𝝀 𝒘𝒉𝒆𝒓𝒆 𝝀𝒎𝒂𝒙 is the highest eigenvalue of the
𝒎𝒂𝒙
inverse of matrix 𝑨 𝒐𝒓 𝑨−𝟏 :

̅ = λmax X
AX ̅

̅ = A−1 (λmax X
or X ̅) = λmax (A−1 X
̅)

̅ = (1⁄
or A−1 X ̅
)X
λ max

̅ ≡ λ′max X
or BX ̅

where B is A−1 and λmin of A is (1⁄ ) of B where λ′max is the highest λ of B.

λ′max

49
The λmax to λmin ratio is known as stiffness ratio and indicates the stiffness (maximum rate of
change in the functional value corresponding to λmax relative to minimum rate corresponding to
λmin) of the system. A ratio > 10 indicates the system to be stiff requiring a fine step size for
calculations (to be discussed later)

Notes (From Strang's book):

If X is an eigenvector of A corresponding to 𝜆𝑚𝑎𝑥 and A is invertible, then X is also an

eigenvector of 𝐴−1 corresponding to the inverse of its 𝜆𝑚𝑎𝑥 .

Eigenvectors corresponding to different 𝜆𝑠 are linearly independent.

A matrix and its transpose have the same eigenvalues.

A matrix is singular if and only if it has zero 𝜆 . A non-singular matrix has all non zero 𝜆𝑠 .

∑ 𝜆𝑖 = 𝑡𝑟𝑎𝑐𝑒 𝑜𝑓 𝐴 = ∑𝑛𝑖=1 𝑎𝑖𝑖

∏ 𝜆𝑖 = det(𝐴)

The 𝜆𝑠 of an upper or lower triangular matrix are the elements on its main diagonal.

[It also follows that det A = det (UTM of A) = ∏(𝑎𝑖𝑖 )𝑈𝑇𝑀 ]

Quiz 1

50
 Lecture #10
Nonlinear Algebraic Equations

Solving 𝐹1 (𝑋1 ⋯ ⋯ 𝑋𝑛 ) = 0
𝐹2 (𝑋1 ⋯ ⋯ 𝑋𝑛 ) = 0
} 𝑜𝑟 𝐹̅ (𝑋̅) = 0
⋮
𝐹𝑛 (𝑋1 ⋯ ⋯ 𝑋𝑛 ) = 0

𝐹1 (𝑋̅)
𝑜𝑟 𝐹̅ (𝑋̅) = { ⋮ }=0
̅
𝐹𝑛 (𝑋)

where, 𝐹1 (𝑋̅) takes the form such as (𝑋1 − sin 𝑋1 ) = 0

𝑜𝑟 𝑋1 − sin 𝑋2 = 0
}
and 𝑋1 = 𝑒 𝑋2

Solving a set of nonlinear algebraic equations is naturally difficult.

Let us begin with solving one nonlinear equation:

𝑓(𝛼) = 0 (𝑅𝑜𝑜𝑡 𝑓𝑖𝑛𝑑𝑖𝑛𝑔) 𝛼 =?

Closed Methods

1. Bracketing method (for a monotonically increasing or decreasing function)

𝑓(𝑋)

𝑓(𝛼) = 0

𝑋𝑙1 𝑋𝑙2

𝑋𝑚2 𝑋𝑚1 𝑋𝑈 𝑋

Step 1: Corner the root. Find 𝑋𝑙 𝑎𝑛𝑑 𝑋𝑈 such that

𝑓(𝑋𝑙 )𝑓(𝑋𝑈 ) < 0 ∶

All it means is that one root is cornered or bracketed. Some functions may have multiple

51
roots (be careful and corner all three roots separately by plotting the function qualitatively but
accurately).

𝑁𝑜𝑡𝑒: 𝑇ℎ𝑖𝑠 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑡ℎ𝑟𝑒𝑒 𝑟𝑜𝑜𝑡𝑠

𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑋𝑙 𝑎𝑛𝑑 𝑋𝑈 .

𝑋𝑈

𝑋𝑙
𝑋𝑈 +𝑋𝑙
Step 2: 𝑋𝑚 (1) = = 𝑋𝑜𝑙𝑑
2

Check f(X m ) = 0. If true, you have the root or Xm = α

If not, check ⇒ if f(Xm )f(Xl ) < 0. If true,

XU = Xm (replace XU with Xm )

else Xl = Xm (replace Xl with Xm )

(This way, you are cornering the root or coming closer to the root)

(Note: The other check f(X)f(XU ) < 0 will also work. If – ve Xl = Xm else

XU = Xm )
XU +Xl
Step 3: Xm (2) = = Xnew (XU & Xl are new values)
2

Xm new −Xm old

you can begin error Check |ξa | = | | × 100 <∈ (specified by user)
Xm old
{ checks after 2nd else go to step 2 (keep iterating)
iterations only X(i+1) −X(i)
till | | × 100 <∈
X(i)

Example: Van der Waals gas law/equation:

a
(P + ) (V − b) = constant
V2

52
 It is clear that this equation is polynomial in V with 𝑛 = 3.

Let us assume that the other variables and constants are known and the simplified
equation takes the following form:

f(V) = 0

V 3 − 0.165V − 3.993 × 10−4 = 0

Calculate V or find the roots

Step 0: Bracket the root: V{0, 0.11} f(V) Xm (1)

Xl = 0, XU = 0.11 by plotting
0.11
1. Check f(Xl )f(XU ) < 0 V
So that the root is bracketed (cornered)

(1) 0+0.11
2. X m = = 0.055 (3 digits accurate after decimal)
2

f(0.055)f(0) = 3.993 × 10−4 × 6.655 × 10−5 > 0

(You would have checked whether f(0.055) x f(0.11) < 0 or not as well!)
Xl = 0.055, XU = 0.11 (remains the same)

Xl +Xu 0.055+0.011
3. Xm (2) = = = 0.0825 = Xnew
2 2

f(Xm (2) ) = −1.622 × 10−4

Check f(Xl )f(Xm ) < 0

XU = 0.0825, Xl = 0.055

0.0825−0.055
4: Check ∈a = | | × 100 = 50.00% > 0.2% (∈s )
0.055
0.055+0.0825
Therefore, X m = = 0.00875
2
Do 10 iterations to obtain Xm = 0.0625 with ∈a = 0.1721% < 0.2%

53
Ans. V (root) = 0.0625.
You should also prepare a table while solving:

>
Xl XU Xm ∈ f(Xm )f(Xl ) your
< 0
}
> 0 decision
or f(Xm )f(XU )
<
1. Yes
2. Xm No
⋮

10.

2. False – Position method (Regular-Falsi)

(This method is similar to the previous one, except draw a straight line connecting
XU and Xm to determine the intersection on X axis as the approximate value for the root)

Step1: Bracket root

f(X) f(XU )f(Xl ) < 0

(1) f(XU )
2. X r = XU − × (Xl − XU )
f(Xl ) − f(XU )

f(XU ) 0 − f(Xl )
X (X = )
U − Xr Xr − Xl

α(root) XU f(Xl ) − Xl f(XU )

(1) = = Xold
Xl Xr f(Xl ) − f(XU )

3. Check: f(Xr )f(Xl ) < 0 or f(Xr )f(XU ) < 0

XU = Xr or Xl = Xr

XU f(Xl ) − Xl f(XU )
Xr (2) = = Xnew
f(Xl ) − f(XU )

Check ∈a <∈s else go to 2; iterate till convergence. Note that it is difficult to choose
between two methods. In general, this method is faster. There are always exceptions. Try
f(X) = X10 − 1!

54
 Open Methods

1. Fixed Point Iteration.

Step1. f(X) = 0 ⇒ Modify⁄transform it to
f(X) = X − g(X) = 0
eg. f(X) = X − cos X
or X − eX
(Try for x 2 − 1 as X − g(X) = 0 or X = g(X)

Sometimes it is not straight forward & there may be more than one combination.
The working formula is
Y = X = g(X) (2 equations and find its interaction)

Y=X f(X) = 0 split into

Y Y = X and Y = g(X)
Y = g(X)
Therefore, a sequence is generated.
X 3 X 2 X1 X 0 Xn+1 = g(Xn )

Sometimes, one has to start from left to the root; sometimes there is a divergence or
spiral convergence, depending on type of functions.

Y or
𝑔(𝑋)

𝑋0 𝑋1 𝑋2
fast convergence slow convergence divergence
(spiral)
Criterion for convergence: |𝐠 ′ (𝐗)| < 𝐘 ′ (𝐗) = 𝟏 at 𝐗 = 𝐗 𝐧

55
Note: Closed methods often generate at least one root. Open methods may sometimes result in
divergence depending upon type of function and starting guess, Xo . However, if they converge,
they will do more quickly than the closed methods.

Step 2: Start with Xo

X1 = g(Xo )
X1 −X0
Step 3: Check ∈a = | | <∈s
X1

If not, Xold = Xr (iterated value)

Xr = g(Xold )
Xr −Xold
Check ∈a = | | <∈s
Xold

Else keep on iterating, keeping in mind the sequence Xn+1 = g(Xn ).

2. Newton-Raphson Method
(Most popular method to solve a non-linear algebraic equation)

f(X) = 0 tangent Step1: Choose

tangent Xr = Xo (guess)
f(Xr )
2. f ′ (Xr ) = tangent at X r =
X0 − X1
f(Xr )
X2 X0 X1 = X r −
f′ (Xr )

(Note: f ′ (Xr ) ≠ 0)
X1 −Xr
3. If ∈a = | | × 100 <∈s stop
Xr

α = X1 (root)
else Xr = X1 (repeat the steps)
f(Xi )
Sequence: Xi+1 = Xi − ; f ′ (Xi ) ≠ 0
f′ (Xi )

Xi+1 −Xi
and ∈i+1 = | | × 100
Xi

56
Notes: ⇒
 f ′ (X) ≠ 0 is the major drawback of the method. There is a price for computing f ′ (X)
 Simple to code and convergence is fast (see later)
 There are cases of divergence and ‘missing’ roots, depending on function and initial guess

divergence

Y missing root

X0
X1 X 3 X 2 X 0 X X1
root
In all such cases, the best strategy is to plot and locate all roots graphically (qualitatively but
accurately), and also, start close to the root.

The NR method can be derived from the Taylor series expansion showing the order of error in the
method.
f′′ (Xi )
f(X i+1 ) = f(X i ) + f ′ (Xi )(Xi+1 − X i ) + (X i+1 − X i )2 + ⋯ ⋯
2!

≃ f(X i ) + f ′ (Xi )(Xi+1 − X i ) + 0(h2 )

(error)
If f(X i+1 ) = 0, root is located
f(X )
Xi+1 = Xi − f′ (Xi )
i

Example (NR) : 𝑓(𝑋) = 𝑋 3 − 0.165𝑋 2 + 3.993 × 10−4 ; 𝑓 ′ (𝑋) = 3𝑋 2 − 0.33𝑋

Step 1. 𝑋0 = 0.05: 𝑃𝑙𝑜𝑡 𝑓(𝑋)𝑎𝑛𝑑 𝑐ℎ𝑜𝑜𝑠𝑒 𝑋0 𝑐𝑙𝑜𝑠𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑖𝑛𝑡𝑒𝑟𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓(𝑋) 𝑤𝑖𝑡ℎ 𝑋 – axis
𝑓(𝑋 )
2. 𝑓 ′ (𝑋0 ) = 3𝑋0 2 − 0.33𝑋0  𝑋1 = 𝑋0 − 𝑓′ (𝑋0 ) = 0.06242
0

0.06242−0.0500
3. ∈ = | | × 100 = 24.84%
0.0500

𝑓(𝑋 )
4. 𝑋2 = 𝑋1 − 𝑓′ (𝑋1 ) = 0.06238
1

∈ = 0.06411 ⇒ 𝛼 = 0.06238 (𝑟𝑜𝑜𝑡) (accurate to the 1st significant digit after decimal)

57
 Lecture #11
Newton-Raphson (Multi-variables)

F1 (X1 , X2 ⋯ Xn ) = 0 ⃗F⃗(X
̅) = 𝟎
F2 (X1 , X2 ⋯ Xn ) = 0
]⇒ ̅) = 0, i = 1 ⋯ n
or Fi (X
⋮
Fn (X1 , X2 ⋯ Xn ) = 0 or Fi (Xi ) = 0, i = 1 ⋯ n

Taylor-expansion of multi-variables:

𝜕𝐹1 (𝑋1 (𝑘+1) − 𝑋1 (𝑘) ) + 𝜕𝐹1 (𝑋2

(𝑘+1)
− 𝑋2 (𝑘) ) +
𝐹1 (𝑋̅ (𝑘+1) ) ≈ 𝐹1 (𝑋̅ (𝑘) ) + | |
𝜕𝑋1 𝑋̅ (𝑘) ≠ 𝑋1 𝜕𝑋2 𝑋̅ (𝑘) ≠ 𝑋2

𝜕𝐹1 (𝑋𝑛 (𝑘+1) − 𝑋𝑛 (𝑘) )

⋯⋯ (1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟 𝑎𝑐𝑐𝑢𝑟𝑎𝑡𝑒)
𝜕𝑋𝑛 𝑋̅ (𝑘) ≠ 𝑋𝑛

𝜕𝐹2 (𝑋1 (𝑘+1) − 𝑋1 (𝑘) ) + 𝜕𝐹2 (𝑋2

(𝑘+1)
− 𝑋2 (𝑘) ) +
𝐹2 (𝑋̅ (𝑘+1) ) ≈ 𝐹2 (𝑋̅ (𝑘) ) + | |
𝜕𝑋1 𝑋̅ (𝑘) ≠ 𝑋1 𝜕𝑋2 𝑋̅ (𝑘) ≠ 𝑋2

𝜕𝐹2 (𝑋𝑛 (𝑘+1) − 𝑋𝑛 (𝑘) )

⋯⋯
𝜕𝑋𝑛 𝑋̅ (𝑘) ≠ 𝑋𝑛

𝜕𝐹𝑛 (𝑋1 (𝑘+1) − 𝑋1 (𝑘) ) + 𝜕𝐹𝑛 (𝑋2

(𝑘+1)
− 𝑋2 (𝑘) ) +
𝐹𝑛 (𝑋̅ (𝑘+1) ) ≈ 𝐹𝑛 (𝑋̅ (𝑘) ) + | | (𝑘)
𝜕𝑋1 𝑋̅ (𝑘) ≠ 𝑋1 𝜕𝑋2 𝑋̅ ≠ 𝑋2

𝜕𝐹𝑛 (𝑋𝑛 (𝑘+1) − 𝑋𝑛 (𝑘) )

⋯⋯
𝜕𝑋𝑛 𝑋̅ (𝑘) ≠ 𝑋𝑛

If 𝑋̅ (𝑘+1) = 𝛼 (𝑟𝑜𝑜𝑡 𝑎𝑡 (𝑘 + 1)𝑡ℎ 𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠) ⇒ 𝐹𝑖 (𝑋̅ (𝑘+1) ) = 0,

(𝑖 = 1 ⋯ 𝑛)

58
 𝜕𝐹1 𝜕𝐹1 𝜕𝐹1 (𝑘)
⋯
0 𝐹1 (𝑘) 𝜕𝑋1 𝜕𝑋2 𝜕𝑋𝑛
∆𝑋1
𝜕𝐹2
0 𝐹2 ⋯ ⋯ ⋯ ∆𝑋2
𝜕𝑋1
⋮ = ⋮ +
⋮
⋮
⋮ ⋮ ⋮
⋮
{0} {𝐹𝑛 } 𝜕𝐹𝑛 𝜕𝐹𝑛 𝜕𝐹𝑛 {∆𝑋𝑛 }
[𝜕𝑋1 ⋯
𝜕𝑋2 𝜕𝑋𝑛 ]

(∆𝑋1 = 𝑋1 (𝑘+1) − 𝑋1 (𝑘) , 𝑒𝑡𝑐)

𝜕𝐹1 𝜕𝐹1 (𝑘)

⋯ ∆𝑋1 𝐹1 (𝑘)
𝜕𝑋1 𝜕𝑋𝑛
𝑜𝑟 ⋮ { ⋮ } = −{ ⋮ }
𝜕𝐹𝑛
⋯
𝜕𝐹𝑛 ∆𝑋𝑛 𝐹𝑛
[𝜕𝑋1 𝜕𝑋𝑛 ]

𝑜𝑟 ̅̅̅
𝐽𝐽∆𝑋̅ (𝑘) = −𝐹̅ (𝑘)

𝑜𝑟 ∆𝑋̅ (𝑘) = −̅̅̅

𝐽𝐽−1 𝐹̅ (𝑘)

𝑜𝑟 𝑋̅ (𝑘 + 1) = 𝑋̅ (𝑘) − ̅̅̅
𝐽𝐽−1 𝐹̅ (𝑘) ⇒ Multi-variable NR method (similar in form to single
variable NR method)

Example: 𝐹1 (𝑌̅) = 4 − 8𝑌1 + 4𝑌2 − 2𝑌1 3 = 0

𝐹2 (𝑌̅) = 1 − 4𝑌1 + 3𝑌2 + 𝑌2 2 = 0

Starting guess values

𝑌̅ (1) = [𝑌1 𝑌2 ]𝑇(1) = [0.5 0.5]𝑇

Apply NR multivariable method formula:

𝑋̅ (𝑘+1) = 𝑋̅ (𝑘) − 𝐽𝐽
̅ −1 𝐹̅ (𝑘)

𝜕𝐹1 𝜕𝐹1 −1
𝑌 (𝑘+1) 𝑌 (𝑘)
𝜕𝑌1 𝜕𝑌2 𝐹 (𝑘)
𝑜𝑟 { 1 } = { 1} − { 1}
𝑌2 𝑌2 𝜕𝐹2 𝜕𝐹2 𝐹2
[ 𝜕𝑌1 𝜕𝑌2 ]

59
 −1
𝑌 1 0.5 (−8 − 6𝑌1 2 ) 4 𝐹
{ 1} = { } − [ ] { 1}
𝑌2 0.5 −4 (3 + 2𝑌2 ) 𝐹2 0.5,0.5

0.5 −9.5 4 −1 1.75

={ }−[ ] { }
0.5 −4 4 0.75
0.5 1 4 −4 1.75
={ } + 22 [ ]{ }
0.5 4 −9.5 0.75
(Note: For 𝟐 × 𝟐 matrix simple Cramer’s rule was applied to determine the inverse of
the matrix, else apply 𝑮 − 𝑱 𝒐𝒓 𝑳𝑼 decomposition method can be used for a large size
matrix)
1
𝑌1 (1) = 0.5 + 22 (4 × 1.75 − 4 × 0.75) = 0.6818

1
𝑌2 (1) = 0.5 + (4 × 1.75 − 9.5 × 0.75) = 0.4943
22
Continue iteration till there is a convergence. One can prepare a table like this:
𝜕𝐹1 𝜕𝐹1 𝜕𝐹2 𝜕𝐹2
# 𝑌1 𝑌2 𝐹1 𝐹2 𝑌1 𝑌2 ∈𝑎1 ∈𝑎2
𝜕𝑌1 𝜕𝑌2 𝜕𝑌1 𝜕𝑌2
1

Secant Method
(Another common open method; it does not require derivative to compute. Requires two
starting adjacent guess values instead)

Step1:

Straight line: Choose 𝑋𝑖−1 𝑎𝑛𝑑 𝑋𝑖 adjacent guess values near the root ′𝛼′.

2. Draw straight line connecting 𝑓(𝑋𝑖−1 ) 𝑎𝑛𝑑 𝑓(𝑋𝑖 ). Find

𝑓(𝑥) intersection on X axis → 𝑋𝑖+1

𝑓(𝑋 ) 𝑓(𝑋𝑖−1 )
′𝛼′ 𝑓 ′ (𝑋𝑖+1 ) = 𝑋 −𝑋𝑖 =𝑋
𝑖 𝑖+1 𝑖−1 −𝑋𝑖+1

𝑖 𝑖−1 (𝑋 −𝑋 )
𝑋𝑖+1 𝑋 𝑋𝑖+1 = 𝑋𝑖 − 𝑓(𝑋𝑖 ) × 𝑓(𝑋 )−𝑓(𝑋
𝑖 𝑖−1 )

𝑓(𝑋𝑖 )
𝑜𝑟 𝑋𝑖+1 = 𝑋𝑖 − 𝑓(𝑋𝑖 )−𝑓(𝑋𝑖−1 ) : sequence
𝑋𝑖 −𝑋𝑖−1

60
Check if 𝑓(𝑋𝑖+1 ) → (𝑦𝑜𝑢 ℎ𝑎𝑣𝑒 ℎ𝑖𝑡 𝑡ℎ𝑒 𝑟𝑜𝑜𝑡), else repeat with 𝑋𝑖 𝑎𝑛𝑑 𝑋𝑖+1 following the above
sequence.

 The method converges faster than Bisection method, but slower that NR. It is preferred
over NR to avoid 𝑓 ′ = 0.
𝑓(𝑋 )
 If 𝑋𝑖 → 𝑋𝑖−1 , note 𝑋𝑖+1 = 𝑋𝑖 − 𝑓′ (𝑋𝑖 ) (same as NR!)
𝑖

At this stage, let us look at Taylor’s series. Any function f(X) can be expanded as:

𝑓 ′′ (𝑋0 ) 𝑓 𝑛 (𝑋0 )(𝑋−𝑋0 )𝑛

𝑓(𝑋) = 𝑓(𝑋0 ) + 𝑓 ′ (𝑋0 )(𝑋 − 𝑋0 ) + (𝑋 − 𝑋0 )2 + ⋯ + ⋯ 𝑡𝑜 ∞
2! 𝑛!

exact

𝑓 ′′ (𝑋𝑖 ) 𝑓 𝑛 (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑖 )𝑛

or, 𝑓(𝑋𝑖+1 ) = 𝑓(𝑋𝑖 ) + 𝑓 ′ (𝑋𝑖 )(𝑋 − 𝑋𝑖 ) + (𝑋𝑖+1 − 𝑋𝑖 )2 + ⋯ + ⋯ 𝑡𝑜 ∞
2! 𝑛!

exact

𝑓 ′ (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑖 ) 𝑓 ′′ (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑖 )2

≈ 𝑓(𝑋𝑖 ) + + + ⋯ 0(ℎ3 )
1! 2!

approximate

or truncation error is 0(ℎ3 ) (accurate to 2nd order)

𝑓 ′ (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑖 )

≈ 𝑓(𝑋𝑖 ) + + 0(ℎ2 )
1!

approximate

(accurate to 1st order or truncation error is 0(ℎ2 ))

(error is one order greater than the accuracy of the method)

𝑓 𝑛 (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑛 )𝑛
= 𝑓(𝑋𝑖 ) + 𝑓 ′ (𝑋𝑖 )(𝑋𝑖+1 − 𝑋𝑖 ) + ⋯ + 𝑅𝑛+1
𝑛!

exact!

𝑓 𝑛+1 (𝜉)(𝑋𝑖+1 −𝑋𝑛 )𝑛+1

where, 𝑅𝑛+1 = (𝑐𝑎𝑛 𝑏𝑒 𝑒𝑥𝑎𝑐𝑡𝑙𝑦 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒𝑑)
𝑛+1!

However, 𝜉 is ‘some’ number between 𝑋𝑖+1 𝑎𝑛𝑑 𝑋𝑖 .

Therefore,

𝑓 ′ (𝑋𝑖 )(𝑋𝑖+1 −𝑋𝑖 ) 𝑓 2 (𝜉)(𝑋𝑖+1 −𝑋𝑖 )2

𝑓(𝑋𝑖+1 ) = 𝑓(𝑋𝑖 ) + 1!
+ 2!

exact
61
 𝑓 ′ (𝜉)(𝑋𝑖+1 −𝑋𝑖 )
= 𝑓(𝑋𝑖 ) + (1st Mean Value Theorem)
1!

exact
𝑓(𝑋𝑖+1 )−𝑓(𝑋𝑖 )
In other words, 𝑓 ′ (𝜉) = (𝑋𝑖+1 −𝑋𝑖 )

All it means there is ′𝜉′ somewhere between 𝑋𝑖 𝑎𝑛𝑑 𝑋𝑖+1 , where the tangent equals the divided
difference between two end points 𝑋𝑖 − 𝑋𝑖+1 .

Note: Taylor series is used to determine the error or accuracy of a numerical method, or to
determine the convergence of the method.

Convergent Analysis

Fixed Point Iteration method

𝑓(𝑋) = 𝑋 − 𝑔(𝑋)

𝑜𝑟 𝑋 (𝑛+1) = 𝑔(𝑋 (𝑛) ) (𝑖𝑡𝑒𝑟𝑎𝑡𝑖𝑜𝑛)

If 𝛼 is the root then 𝛼 − 𝑔(𝛼) = 0

Error at (𝑛 + 1)𝑡ℎ iteration =|True Value − Approximate Value|

𝑒 𝑛+1 = |𝛼 − 𝑋 (𝑛+1) |

= |𝑔(𝛼) − 𝑔(𝑋 (𝑛) )|

= |𝑔′ (𝑋̃)(𝛼 − 𝑋 (𝑛) | ∶ Recall Mean Value Theorem. ̃

X is between

α and X (n)

= |𝑔′ (𝑋̃)𝑒 𝑛 |

𝑒 𝑛+1
= |𝑔′ (𝑋̃)| < 1 for error to decrease as 𝑛 → ∞ where 𝑋̃ is {𝛼, 𝑋 𝑛 }
𝑒𝑛

NR Convergent
𝑓
𝑋 𝑘+1 = 𝑋 𝑘 − 𝑓𝑘′ ; 𝑓𝑘′ ≠ 0
𝑘

𝑓
𝑒 (𝑘+1) = |𝛼 − 𝑋 𝑘+1 | = |𝛼 − 𝑋 𝑘 + 𝑓𝑘′ |
𝑘

62
 (𝑋 −𝑋 ) 2
𝑓(𝑋 𝑘+1 ) = 𝑓(𝑋 𝑘 ) + 𝑓𝑘 ′ (𝑋 𝑘+1 − 𝑋 𝑘 ) + 𝑓 ′′ (𝑋̃) 𝑘+12! 𝑘

exact (𝑋̃: {𝑋𝑘 , 𝑋𝑘+1 })

If 𝑋 𝑘+1 is the root (𝛼)

(𝛼 − 𝑋𝑘 )2 ′′
0 = 𝑓(𝑋 𝑘 ) + 𝑓𝑘 ′ (𝛼 − 𝑋 𝑘 ) + 𝑓 (𝑋̃)
2!

𝑓 ′′ (𝑋̃)
∴ 𝑒 (𝑘+1) = | (𝛼 − 𝑋𝑘 )2 |
2𝑓 ′ 𝑘

𝑓 ′′ (𝑋̃ )
= | 2𝑓′ (𝑒 𝑘 )2 |
𝑘

𝑒 (𝑘+1) 𝑓 ′′ (𝑋̃ )
= |2𝑓′ (𝛼)| ; 𝑓 ′ (𝛼) ≠ 0
(𝑒 𝑘 )2

If 𝑘 → ∞ 𝑒 (𝑘+1) decreases by the square of the error in the previous step. In other words, the
method has quadratic convergence (fast convergence), although the method is only 1st order
accurate. This is the special feature of NR method.

63
 Lecture #12
Function Approximation: Interpolation
Fit a line passing through the data points to
predict the functional value at an intermediate interpolated value
data. The line must pass through the data. The method is used
to interpolate, for example, certain properties such as enthalpy and
vapour pressure, at an intermediate point where the functional 𝑓𝑖
values or properties are not available.
Therefore, the data provided for the interpolation must
be reliable/accurate. 𝑋𝑖
Mathematically, interpolating function must be smooth
(differential); continuous and pass through all data points.
Compare it to ‘regression’ (best fit to the data) when the line need not pass through each data.
Regression is used for prediction when data are experimentally measured or approximately
calculated. Therefore, the regressed data may have errors.

Yi experimental data
𝑓𝑖

𝑋𝑖
(No regression topic to be covered in this course; no extrapolation as well!)

The main motivation for developing an interpolating function is not only to predict the
functional value at an interpolating/intermediate point, but also to approximate a continuous f(x)
or a discrete data set, y(x) as the nth degree polynomial, Pn(x); thus the name of the current topic:
function approximation. You will see in the later lessons that once you have your own
approximated simple Pn(x), you will forgo the complex f(x) and work on Pn(x) instead, not only
for calculating functional values at the intermediate points but also for computing f’(x) and f’’(x)
as Pn’(x) and Pn’’(x), respectively. Also, all formulae for integration are also derived from Pn(x).

64
Interpolating function for ‘n’ data points
𝑃𝑛−1 (𝑋) = 𝑎𝑛−1 𝑋 𝑛−1 + 𝑎𝑛−2 𝑋 𝑛−2 + ⋯ 𝑎0 : a polynomial of degree (𝑛 − 1).
If there are ‘n’ data points, a unique polynomial of degree (n-1) passes through the points:
Ex. 1. 𝑦 = 𝑚𝑥 + 𝑏 (1st order polynomial) passing through two data points. Two unknown &
two equations to solve m and b:

2. Three data points

P2
y = ax 2 + bx + c (2nd order polynomial, a quadratic equation)
(3 equations to solve three unknowns 𝑎, 𝑏 & 𝑐)
3. Thus, ‘n’ data points will have a 𝑢𝑛𝑖𝑞𝑢𝑒 polynomial of degree (n − 1) passing through the
points.
𝑦 = 𝑎𝑛−1 𝑥 𝑛−1 + 𝑎𝑛−2 𝑥 𝑛−2 + ⋯ ⋯ 𝑎0
Therefore,
𝑦1 = 𝑎𝑛−1 𝑥1 𝑛−1 + 𝑎𝑛−2 𝑥1 𝑛−2 + ⋯ ⋯ 𝑎0
𝑦2 = 𝑎𝑛−1 𝑥2 𝑛−1 + 𝑎𝑛−2 𝑥2 𝑛−2 + ⋯ ⋯ 𝑎0 n
⋮ equations
𝑦𝑛 = 𝑎𝑛−1 𝑥𝑛 𝑛−1 + 𝑎𝑛−2 𝑥𝑛 𝑛−2 + ⋯ ⋯ 𝑎0 }
or

𝑦1 1 𝑥1 𝑥1 2 ⋯ 𝑥1 𝑛−1 𝑎0
𝑦2 1 𝑥2 𝑥2 2 ⋯ 𝑥2 𝑛−1 ⋮
{⋮}=| |{ ⋮ }
⋮
𝑦𝑛 1 𝑥𝑛 𝑥𝑛 2 𝑥𝑛 𝑛−1 𝑎𝑛−1
Although the problem is well defined, it is tedious to solve (𝑎0 ⋯ ⋯ 𝑎𝑛−1 ) from a large sized 𝑛 ×
𝑛 matrix.
𝑦̅ = 𝐴𝑛×𝑛 𝑎̅
The common methods to fit 𝑃𝑛−1 polynomial passing through ‘n’ data points, or 𝑃𝑛 through
(n + 1) data points:
Newton’s divided-difference scheme
Consider 2 data points
Let a line pass through (𝑋0 , 𝑌0 ) & (𝑋1 , 𝑌1 ):

65
 𝑌 − 𝑌0 𝑌1 − 𝑌0
=
𝑋 − 𝑋0 𝑋1 − 𝑋0
𝑌 −𝑌
Y1 or 𝑌 = 𝑌0 + 𝑋1 −𝑋0 (𝑋 − 𝑋0 )
1 0
Y 𝑜𝑟 𝑌 = 𝑎0 + 𝑎1 (𝑋 − 𝑋0 )
Y0 where a0 = Y0
Y −Y
a1 = X1−X0 = Y(X1 , X0 ) defined as
1 0
X
Newton First Divided
X0 X1
{ Difference(N1DD)
Similarly, define
𝑌[𝑋2 , 𝑋1 , 𝑋0 ] ≡ Newton 2nd divided difference

𝑌[𝑋2 , 𝑋1 ] − 𝑌[𝑋1 , 𝑋0 ]
= (𝑁2𝐷𝐷)
𝑋2 − 𝑋0

𝑌[𝑋𝑛 , 𝑋𝑛−1 , ⋯ 𝑋1 ] − 𝑌[𝑋𝑛−1 , 𝑋𝑛−2 , ⋯ 𝑋0 ]

𝑌[𝑋𝑛 , 𝑋𝑛−1 , ⋯ 𝑋0 ] = (𝑁𝑛𝐷𝐷)
𝑋𝑛 − 𝑋0
Take 3 data points (𝑋0, 𝑋1 , 𝑋2 ):
𝑌(𝑋) ≡ 𝑃2 (𝑋) = 𝑎0 + 𝑎1 (𝑋 − 𝑋) + 𝑎2 (𝑋 − 𝑋0 )(𝑋 − 𝑋1 )
By fitting 𝑃2 (𝑋) to three data points, the three coefficients
Y2
can be determined as
Y1
Yi 𝑌 −𝑌
𝑎0 = 𝑌0 , 𝑎1 = 𝑋1 −𝑋0 ≡ 𝑌[𝑋1 , 𝑋0 ]
1 0
Y0
X0 X1 X2
𝑌[𝑋2 ,𝑋1 ]−𝑌[𝑋1 ,𝑋0 ]
Xi and 𝑎2 = ≡ 𝑌[𝑋2 , 𝑋1 , 𝑋0 ]
𝑋2 −𝑋0

Similarly for (n+1) data

𝑌(𝑋) 𝑜𝑟 𝑃𝑛 (𝑋)
= 𝑎0 + 𝑎1 (𝑋 − 𝑋0 ) + 𝑎2 (𝑋 − 𝑋0 )(𝑋 − 𝑋1 ) + ⋯ 𝑎𝑛 (𝑋 − 𝑋0 )(𝑋 − 𝑋1 ) ⋯ (𝑋 − 𝑋𝑛−1 )
where 𝑎0 = 𝑌0
𝑌 −𝑌
𝑎1 = 𝑌[𝑋1 , 𝑋0 ] = 𝑋1 −𝑋0 (1st divided difference)
1 0

𝑎𝑛 = 𝑌[𝑋𝑛 , 𝑋𝑛−1 ⋯ 𝑋0 ] ( nth divided difference )

66
(Substitute x = x0 , x1 , x2 … xn in Pn (x) to explore the sequence of polynomial)
In the tabular form,
𝑖 𝑋𝑖 𝑌𝑖 1𝑠𝑡 2𝑛𝑑 ⋯ ⋯ 𝑛𝑡ℎ

0 𝑋0 𝑌0 𝑎𝑛 = (𝑌[𝑋𝑛 , ⋯ 𝑋1 ]
1 𝑋1 𝑌1 −𝑌[𝑋𝑛−1 , ⋯ 𝑋0 ])
2 𝑋2 𝑌2 𝑋𝑛 − 𝑋0
⋮
𝑛−1 𝑋𝑛−1 𝑌𝑛−1
{ 𝑛 𝑋𝑛 𝑌𝑛
Note: Top row (circles) provides all NDD coefficients, ao, a1, a2….
2. Lagrangian Interpolation
𝑦(𝑥) = ∑𝑛𝑘=0 𝑙𝑘 (𝑥) 𝑦𝑘 (𝑛 + 1 𝑑𝑎𝑡𝑎)

Lagrangian coefficient of the kth term

Function must be a linear combination of 𝑦𝑘 .

𝑙𝑘 (𝑥) must not depend on 𝑦𝑘
𝑙𝑘 (𝑥) is defined as 𝑙𝑗 (𝑥𝑖 ) = 𝛿𝑖𝑗 (Kronecker delta)
= 1 when i = j and 0 when i ≠ j
Propose, lj (𝑥) = 𝐶𝑗 (𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) ⋯ (𝑥 − 𝑥𝑗−1 )(𝑥 − 𝑥𝑗+1 ) ⋯ (𝑥 − 𝑥𝑛 ), and
=0 (𝑥 ≠ 𝑥𝑗 )
=1 (𝑥 = 𝑥𝑗 )
−1 −1 −1 −1 −1
Therefore, 𝐶𝑗 = (𝑥𝑗 − 𝑥0 ) (𝑥𝑗 − 𝑥1 ) ⋯ (𝑥𝑗 − 𝑥𝑗−1 ) (𝑥𝑗 − 𝑥𝑗+1 ) ⋯ (𝑥𝑗 − 𝑥𝑛 )
(𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) ⋯ (𝑥 − 𝑥𝑗−1 )(𝑥 − 𝑥𝑗+1 ) ⋯ (𝑥 − 𝑥𝑛 )
and, 𝑙𝑗 (𝑥) =
(𝑥𝑗 − 𝑥0 )(𝑥𝑗 − 𝑥1 ) ⋯ (𝑥𝑗 − 𝑥𝑗−1 )(𝑥𝑗 − 𝑥𝑗+1 ) ⋯ (𝑥𝑗 − 𝑥𝑛 )

The interpolating function is as follows:

𝑌𝑛 (𝑥) = ∑ 𝑙𝑘 (𝑥)𝑌𝑘 ≡ 𝑃𝑛
𝑘=0
where
𝑛
(𝑥 − 𝑥𝑖 )
𝑙𝑗 (𝑥) = ∏
𝑖=0
(𝑥𝑗 − 𝑥𝑖 )
(𝑖≠𝑗)

67
Suppose 2 data points: (𝑋0 , 𝑌0 ) and (𝑋1 , 𝑌1 )

(X,Y) (X1,Y1)
Y1 (x) = l0 (x)Y0 + l1 (x)Y1
x−x x−x
l0 = x −x1 , l1 = x −x0
0 1 1 0
x−x1 x−x0
or Y1 (x) = (x ) Y0 + (x ) Y1 (X0,Y0)
0 −x1 1 −x0

(Since 𝑌1 (𝑥) is unique, it cannot be different from Newton-Divided difference formula; only
forms are different)

Suppose there are 3 data points: (X0 , Y0 ), (X1 , Y1 ), and (X2 , Y2 )

𝑌2 (𝑥) = 𝑙0 (𝑥)𝑌0 + 𝑙1 (𝑥)𝑌1 + 𝑙2 (𝑥)𝑌2

(𝑥 − 𝑥1 )(𝑥 − 𝑥2 ) (𝑥 − 𝑥0 )(𝑥 − 𝑥2 )
𝑙0 (𝑥) = ; 𝑙1 (𝑥) = ,
(𝑥0 − 𝑥1 )(𝑥0 − 𝑥2 ) (𝑥1 − 𝑥0 )(𝑥1 − 𝑥2 )
(𝑥 − 𝑥1 )(𝑥 − 𝑥2 ) (𝑥 − 𝑥0 )(𝑥 − 𝑥2 ) (𝑥 − 𝑥0 )(𝑥 − 𝑥1 )
𝑌(𝑥) = 𝑌0 + 𝑌1 + 𝑌
(𝑥0 − 𝑥1 )(𝑥0 − 𝑥2 ) (𝑥1 − 𝑥0 )(𝑥1 − 𝑥2 ) (𝑥2 − 𝑥0 )(𝑥2 − 𝑥1 ) 2
≡ 𝑠𝑎𝑚𝑒 𝑎𝑠 𝑁𝐷𝐷𝐸 𝑜𝑟 𝑎𝑋 2 + 𝑏𝑋 2 + 𝑐

Ex. 𝑋1 𝑋2 𝑋 𝑋3
Depth (m) -1 0 0.5 1.0

𝑌(𝑡 0 𝑐) 10.1 11.3 ? 11.9

𝑌2 (𝑥) = 𝐶0 + 𝐶1 𝑥 + 𝐶2 𝑥 2 (unique)

3 data points and 3 eqns to solve Cs :

1 −1 1 𝐶0 10.1
[1 0 0] {𝐶1 } = {11.3}…….>Solve by any previously learnt method to obtain
1 1 1 𝐶2 11.9

𝐶0 = 11.3, 𝐶1 = 0.9, 𝑎𝑛𝑑 𝐶2 = −0.3 ⇒ 𝑌(0.5) = 11.675

𝑁𝐷𝐷 1𝑠𝑡𝐷𝐷: 𝑌[𝑋0 , 𝑋1 ] 2𝑛𝑑 𝐷𝐷: 𝑌[𝑋0 , 𝑋1 , 𝑋2 ]

11.3−10.1 1.2−0.6
𝑋0 −1 10.1 = 1.2 = −0.3
0+1 −1−1

11.9−11.3
𝑋1 0 11.3 = 0.6
1−0

𝑋2 1 11.9

68
Therefore, Y2 (x) = a0 + a1 (x − x0 ) + a2 (x − x0 )(x − x1 )

a0 = Y0 = 10.1, a1 = FDD = 1.2, a2 = SDD = −0.3

= 10.1 + 1.2(x + 1) − 0.3(x + 1)x

- (Same as before, check) Y(0.5) = 11.675

x−x
(2) Lagrange Y2 (x) = ∑2k=0 lk Yk ; lj (x) = ∏2i=0 (x −xi )
j i
i≠j

(x−x1 )(x−x2 ) (x−x0 )(x−x2 )

l0 (x) = (x ; l1 (x) = (x , etc.
0 −x1 )(x0 −x2 ) 1 −x0 )(x1 −x2 )

(x − 0)(x − 1) (x + 1)(x − 1) (x + 1)x

Y2 (x) = × 10.1 + × 11.3 + × 11.9
−1 × −2 1 × −1 2×1
(Same as before, check) Y(0.5) = 11.675

Note: ⇒ If ‘n’ is large, (> 4), the polynomial shows oscillation. In principle, one uses ‘piece-
wise’ interpolation for large # of data points.

⇒ Less smooth if data points are less or sparse.

⇒ ‘Problem’ at the end points.

⇒ NDD uses the previous calculations, if an extra point is to be included in finding a new
polynomial, unlike each coefficient has to be recalculated in ‘LI’. However, it is easy to program
‘LI’.

Error analysis: For (n+1) data points, 𝑃𝑛 (𝑜𝑟𝑑𝑒𝑟 𝑛) is the unique polynomial to pass through
(n+1) data points. This polynomial can be used to predict 𝑃𝑛 (𝑋) at a new data points ‘X’. How
much error can be expected in this prediction/interpolation? From engineering point of view,
how much confidence an user has in the interpolated functional value? Well, in such case one
should use an additional data (𝜉, 𝑌(𝜉)) over the range (𝑋0 ⋯ 𝑋𝑛 ) and check the convergence in
the re-interpolated value at 𝑃𝑛+1 (𝑋). Note, now you have (n+2) data points including 𝜉.

Theoretically, it can be shown using Taylor’s series that the error in the interpolation using
𝑃𝑛 (𝑋) for (n+1) data is

𝑦 𝑛+1 (𝜉)
(𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) ⋯ ⋯ (𝑥 − 𝑥𝑛 )
𝑛+1!

where (𝑥0 < 𝜉 < 𝑥𝑛 )

additional data pt.

69
You should have also noted that
𝑦(𝑥1 )−𝑦(𝑥0 )
𝑦[𝑥0 , 𝑥1 ] = = 𝑦′ (accurate to the 1st order)
𝑥1 −𝑥0

𝑦[𝑥0 , 𝑥1 , 𝑥2 ] = ⋯ ⋯ = 𝑦′′ (accurate to the 2nd order)

𝑌[𝜉,⋯𝑥𝑛−1 ]−𝑌[𝑥0 ,⋯𝑥𝑛 ]
and 𝑦[𝜉, 𝑥0 , 𝑥1 , ⋯ 𝑥𝑛 ] = = 𝑦 𝑛+1 (𝜉)
𝜉−𝑥𝑛

Therefore, the error calculated is nothing but 𝑅𝑛+1 (remainder) of the Taylor’s series. For more
on this, read the book by Ferziger.

70
 Lecture #13
Similar to the Newton’s Divided Difference and Lagrangian interpolation schemes, there is the
Newton’s interpolation formula that makes use of ′Δ′ operator (Forward Difference
Operator)

Δ𝑦(𝑥) = 𝑦(𝑥 + Δ𝑥) − y(𝑥) ∶ 𝐹𝑜𝑟𝑤𝑎𝑟𝑑 − 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑜𝑝𝑒𝑟𝑎𝑡𝑜𝑟

𝐸 𝑦(𝑥) = 𝑦(𝑥 + Δ𝑥) ∶ 𝑆ℎ𝑖𝑓𝑡 𝑂𝑝𝑒𝑟𝑎𝑡𝑜𝑟

2
{ 𝐸 𝑦(𝑥) = 𝐸[𝑦(𝑥 + Δ𝑥)] = 𝑦(𝑥 + 2Δ𝑥)
𝐸 𝑛 𝑦(𝑥) = 𝑦(𝑥 + 𝑛Δ𝑥)

Therefore, Δ𝑦(𝑥) = 𝐸𝑦(𝑥) − 𝑦(𝑥) = (𝐸 − 1)𝑦(𝑥)

𝑜𝑟 𝐸 = 1 + Δ ∶ 𝐸 & Δ operators

𝛼(𝛼 − 1) 2 𝛼(𝛼 − 1) ⋯ (𝛼 − 𝑛 + 1) n
𝑜𝑟 𝐸 𝛼 = (1 + Δ)𝛼 = 1 + 𝛼Δ + Δ +⋯ Δ +. ..
2! 𝑛!
𝑜𝑟 𝐸 𝛼 𝑦(𝑥0 ) = 𝑦(𝑥0 + 𝛼Δ𝑥) = (1 + Δ)𝛼 𝑦(𝑥0 )
𝛼(𝛼−1) 𝛼(𝛼−1)⋯(𝛼−𝑛+1)
= 𝑦0 + 𝛼Δ𝑦0 + Δ2 𝑦0 + ⋯ Δn 𝑦0 +
2! 𝑛!

𝑥−𝑥0
If 𝛼 = Δ𝑥

𝛼(𝛼−1) 𝛼(𝛼−1)⋯(𝛼−𝑛+1)
𝑦(𝑥) = 𝑦(𝑥0 + 𝛼Δ𝑥) = 𝑦0 + 𝛼(Δ𝑦0 ) + Δ2 𝑦0 + ⋯ Δn 𝑦0 + 𝑅
2! 𝑛!

𝛼(𝛼−1)(𝛼−𝑛)
𝑅= Δn+1 𝑦(𝜉) 𝑤ℎ𝑒𝑟𝑒 𝑥𝑜 < 𝜉 < 𝑥𝑛
(𝑛+1)!

Newton Forward-Difference Interpolation formula

In general
Δ𝑦𝑖 = y(𝑥𝑖 + Δ𝑥) − 𝑦(𝑥𝑖 ) = 𝑦𝑖+1 − 𝑦𝑖
}
Δ2 𝑦𝑖 = Δ(Δ𝑦𝑖 ) = 𝑦𝑖+2 − 2𝑦𝑖+1 + 𝑦𝑖
Similar to ′Δ′, ∇ has been defined as backward difference operator:
x−xn
∇𝑦𝑖 = 𝑦𝑖 − 𝑦𝑖−1 𝑜𝑟 ∇𝑦(𝑥) = 𝑦(𝑥𝑛 + 𝛼Δ𝑥), where α = Δx

And, therefore, there is another Newton’s interpolation formula! Again, note that all
interpolating functions are one and the same, but written in different forms. Considering that
the Newton’s interpolation formula resembles Taylor’s series, it can also be used for error
analysis. To this end, we have learnt how to approximate f(x) or y(x) with P n(x). We will
see later f’(x) or y’(x) and f’’(x) or y’’(x) will be approximated with P’ n(x) and Pn”(x),
respectively.

71
 Splines
We have earlier noted that a high order polynomial (𝑛 ≥ 4) often shows oscillation. Therefore,
a piece-wise polynomial (𝑛 ≤ 3) is usually preferred for most engineering applications.

→ ‘Spline’ is a piece-wise cubic polynomial (𝑦 = 𝑎𝑥 3 + 𝑏𝑥 2 + 𝑐𝑥 + 𝑑) fitting 2 data points in

segments or pieces. Is it unique? No. Because there are two extra degrees of freedom. In other
words, two extra constraints or conditions may be imposed.

→ Let us define the problem:

𝑆𝑖 = 𝑎𝑖 𝑥 3 + 𝑏𝑖 𝑥 2 + 𝑐𝑖 𝑥 + 𝑑𝑖 is the spline fitting the data 𝑥𝑖 𝑎𝑛𝑑 𝑥𝑖+1 . Thus, it is a higher

order polynomial for two data points.
Si-1 Si+1
Si
xi
xi+1
xi-1

Each spline (𝑆𝑖 ) is a piece-wise cubic.

Curve must pass through the points (𝑥𝑖 & 𝑥𝑖+1 )
𝑓𝑖′ 𝑎𝑛𝑑 𝑓𝑖 ′′ are continuous at all ‘knots’, i.e, curve is smooth and curvature is the same at all
nodes.
𝑆𝑖−1 (𝑥𝑖 ) = 𝑆𝑖 (𝑥𝑖 ) 𝑆+ = 𝑆−
or 𝑆′𝑖−1 (𝑥𝑖 ) = 𝑆𝑖 ′(𝑥𝑖 ) } ≡ 𝑆′+ = 𝑆 ′− }
𝑆𝑖−1 ′′(𝑥𝑖 ) = 𝑆𝑖 ′′(𝑥𝑖 ) 𝑆′′+ = 𝑆 ′′−

Therefore, for n data points there are (n-1) spline segments or 4(n-1) unknown quantities
(coefficients) to be determined:

Equations: 𝑦𝑖 = 𝑆𝑖 (𝑥𝑖 ) ∶ 𝑛

𝑆 + = 𝑆 − ∶ (𝑛 − 2) ∶ only interior pts.

𝑆 ′+ = 𝑆 ′− ∶ (𝑛 − 2)

𝑆 ′′+ = 𝑆 ′′− ∶ (𝑛 − 2)

4𝑛 − 6

You have two BCs specified at the end data points or knots.

Therefore, there are as many (4n-4) equations as # of unknowns.

The problem is now well defined:

72
 𝑆𝑖 (𝑥) = ∑3𝑖=0 𝑎𝑖 𝑥 𝑖 = ?

Note that 𝑆𝑖 "(𝑥) is a piece-wise linear and continuous (i. e. 𝑆𝑖 "(𝑥) = 6𝑎𝑖 𝑥 + 2𝑏𝑖 )

𝑆𝑖 "(𝑥)

𝑥𝑖 (linear) 𝑥𝑖+1

𝑆𝑖 (𝑥) ≡ 𝑐𝑢𝑏𝑖𝑐

Therefore, 𝑆𝑖 ”(𝑥) = ∑1𝑘=0 𝑙𝑘 (𝑥)𝑦𝑘 (𝐿𝑎𝑔𝑟𝑎𝑛𝑔𝑖𝑎𝑛 𝑖𝑛𝑡𝑒𝑟𝑝𝑜𝑙𝑎𝑡𝑖𝑜𝑛)

𝑥−𝑥𝑖+1 𝑥−𝑥𝑖
= 𝑆”(𝑥𝑖 ) + 𝑆”(𝑥𝑖+1 )
𝑥𝑖 −𝑥𝑖+1 𝑥𝑖+1 −𝑥𝑖

𝑥3 𝑥2 𝑆”(𝑥𝑖 ) 𝑥3 𝑥2 𝑆”(𝑥𝑖+1 )
On integrating twice, 𝑆𝑖 (𝑥) = ( 6 − 𝑥𝑖+1 ) + ( 6 − 𝑥𝑖 ) + 𝑐1 𝑥 + 𝑐2
2 Δ𝑖 2 Δ𝑖+1

Apply the conditions

𝑆𝑖 (𝑥𝑖 ) = 𝑦𝑖 and Si (𝑥𝑖+1 ) = 𝑦𝑖+1

𝑆 ′′ (𝑥𝑖 )(𝑥𝑖+1 − 𝑥)3 𝑆 ′′ (𝑥𝑖+1 )(𝑥 − 𝑥𝑖 )3 𝑦𝑖 Δ𝑖

→ 𝑆𝑖 (𝑥) = + + ( − 𝑆"𝑖 (𝑥𝑖 )) (𝑥𝑖+1 − 𝑥)
6Δ𝑖 6Δ𝑖 Δ𝑖 6
𝑦𝑖+1 Δ𝑖
+( − 𝑆"𝑖 (𝑥𝑖+1 )) (𝑥 − 𝑥𝑖 )
Δ𝑖 6

Apply another condition:

𝑆𝑖−1 ′(𝑥𝑖 ) = 𝑆𝑖′ (𝑥𝑖 ) on the following

𝑆 ′′ (𝑥𝑖 )(𝑥𝑖+1 − 𝑥)2 𝑆 ′′ (𝑥𝑖+1 )(𝑥 − 𝑥𝑖 )2 𝑦𝑖 Δ𝑖

𝑆𝑖′ (𝑥) = − + − ( − 𝑆𝑖′′ (𝑥𝑖 ))
2Δ𝑖 2Δ𝑖 Δ𝑖 6
𝑦𝑖+1 Δ𝑖
+ [( − 𝑆𝑖 ′′(𝑥𝑖+1 ))]
Δ𝑖 6

On re-arrangement it can be shown

Δ𝑖−1 ′′ Δ𝑖−1 + Δ𝑖 ′′ Δ𝑖
𝑆 (𝑥𝑖−1 ) + 𝑆 (𝑥𝑖 ) + 𝑆 ′′ (𝑥𝑖+1 )
6 3 6
𝑦𝑖+1 − 𝑦𝑖 𝑦𝑖 − 𝑦𝑖−1
= − ,
Δ𝑖 Δ𝑖−1

𝑖 = 2, 𝑛 − 1 (𝑖𝑛𝑡𝑒𝑟𝑖𝑜𝑟 𝑛𝑜𝑑𝑒𝑠)

73
 1 2 𝑥𝑖−1 𝑥𝑖 𝑥𝑖+1 𝑛 − 1 𝑛

If 𝑆 ′′ (𝑥1 ) 𝑎𝑛𝑑 𝑆 ′′ (𝑥𝑛 ) are pre-known as boundary conditions, we have

𝐴𝑋̅ = 𝑏̅ 𝑜𝑟 𝐴𝑆 ′′ (𝑥) = 𝑏̅ where A is the tridiagonal matrix

For a ‘natural’ spline, 𝑆 ′′ (𝑥1 ) = 𝑆 ′′ (𝑥𝑛 ) = 0 OR 𝑆 ′′ (1) = 𝑆 ′′ (𝑛) = 0

If Δ𝑖 = Δ𝑖−1 (𝑒𝑞𝑢𝑎𝑙𝑙𝑦 𝑠𝑝𝑎𝑐𝑒𝑑 𝑑𝑎𝑡𝑎)
1 𝑦𝑖+1 −2𝑦𝑖 +𝑦𝑖−1
** [𝑆 ′′ (𝑥𝑖−1 ) + 4𝑆 ′′ (𝑥𝑖 ) + 𝑆 ′′ (𝑥𝑖+1 )] = ( )
6 Δ2

1
Call subroutine Tridiag(𝑛 − 2, 1⁄6 , 2⁄3 , 6 , 𝑏̅)

𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 𝑅𝐻𝑆 ̅̅̅̅̅̅

𝑜𝑟 𝑏 𝑣𝑒𝑐𝑡𝑜𝑟
# 𝑜𝑓 𝑒𝑞 𝑛𝑠
𝑜𝑓 𝐴̅

𝑆𝑖 ′′ (𝑥𝑖 ), 𝑖 = 2, 𝑛 − 1 are now known on solving the set of eqns .

Thus, 𝑆𝑖 (𝑥) is determined from the boxed equation on the preceding page 3.

Example: -2.5 0 2.5

0 1.67 0

Fit a cubic natural spline.

Use ** above
1 2 1 0−2×1.67+0
𝑆 ′′ (𝑥1 ) + 3 𝑆 ′′ (𝑥2 ) + 6 𝑆 ′′ (𝑥3 ) =
6 2.52

(You have only one linear algebraic eqn)

𝑆 ′′ (𝑥2 ) = −0.8016 𝑖 = 2 (middle node only)

𝑆1 (𝑥) = −0.05344(𝑥 + 2.5)3 + 1.002(𝑥 + 2.5)

} 𝐴𝑛𝑠.
𝑆2 (𝑥) = 1.667 − 0.4𝑥 2 + 0.0533𝑥 3 )

74
 Numerical Differentiation

f(x)
or P2 approximated
y(xi) polynomial
f(x) ≈ Pn (x)

xi-1 xi xi+1

Given a continuous 𝑓(𝑥), numerical differentiation can be performed to determine gradient

′
𝑓 ′ (𝑥𝑖 ) 𝑎𝑡 𝑥𝑖 . Fit a polynomial (𝑢𝑠𝑢𝑎𝑙𝑙𝑦 𝑃1−3 ) and then determine 𝑃1−3 to estimate 𝑦𝑖′ ,
depending on desired order of accuracy or error. Taylor’s series can be conveniently used to
derive 𝑓 ′ 𝑜𝑟 𝑓′′ because the series terms contain 𝑓 ′ 𝑜𝑟 𝑓′′ .

Taylor’s series: 𝑓(𝑥 + ℎ) = 𝑓(𝑥) + ℎ𝑓 ′ (𝑥) + ℎ2 ⁄2 𝑓 ′′ (𝑥) + ⋯ (1)

𝑓(𝑥 − ℎ) = 𝑓(𝑥) − ℎ𝑓 ′ (𝑥) + ℎ2 ⁄2 𝑓 ′′ (𝑥) − ⋯ (2)

𝑓(𝑥𝑖+1 )−𝑓(𝑥𝑖 )
𝑓 ′ (𝑥𝑖 ) = + 0(ℎ2 ) ∶ 𝐹𝑜𝑟𝑤𝑎𝑟𝑑 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑆𝑐ℎ𝑒𝑚𝑒 (𝐹𝐷𝑆)
ℎ
𝑓(𝑥𝑖 )−𝑓(𝑥𝑖−1 )
𝑓 ′ (𝑥𝑖 ) = + 0(ℎ2 ) ∶ 𝐵𝑎𝑐𝑘𝑤𝑎𝑟𝑑 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑆𝑐ℎ𝑒𝑚𝑒 (𝐵𝐷𝑆)
ℎ
′ 𝑓(𝑥𝑖+1 )−𝑓(𝑥𝑖−1 )
{ 𝑓 (𝑥𝑖 ) = + 0(ℎ3 ) ∶ 𝐶𝑒𝑛𝑡𝑟𝑎𝑙 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑆𝑐ℎ𝑒𝑚𝑒 (𝐶𝐷𝑆)
2ℎ

Richardson’s extrapolation technique can be used to take 3 data pts.

𝑓(𝑥𝑖 + 2ℎ) = 𝑓(𝑥) + 2ℎ𝑓 ′ (𝑥) + 4 (ℎ2 ⁄2) 𝑓 ′′ (𝑥) + ⋯ ⋯ (3)

𝑓(𝑥𝑖 − 2ℎ) = 𝑓(𝑥) − 2ℎ𝑓 ′ (𝑥) + 4 (ℎ2 ⁄2) 𝑓 ′′ (𝑥) + ⋯ ⋯ (4)

xi-2 xi-1 xi xi+1 xi+2

Use 1 an 3 : Multiply eq(1) by 4 and subtract from 3 to eliminate 𝑓′′(𝑥𝑖 )

4𝑓(𝑥𝑖+1 ) − 3𝑓(𝑥𝑖 ) − 𝑓(𝑥𝑖+2 )

𝑓 ′ (𝑥𝑖 ) = + 0(ℎ3 ) ∶ 𝐹𝐷𝑆
3 2ℎ
𝑝𝑜𝑖𝑛𝑡𝑠 𝑆𝑖𝑚𝑖𝑙𝑎𝑟𝑙𝑦 2 𝑎𝑛𝑑 4 𝑤𝑖𝑙𝑙 𝑔𝑖𝑣𝑒
′ (𝑥 )
−4𝑓(𝑥𝑖−1 ) + 3𝑓(𝑥𝑖 ) + 𝑓(𝑥𝑖−2 )
{ 𝑓 𝑖 = + 0(ℎ3 ) ∶ 𝐵𝐷𝑆
2ℎ
You will see later in the course that although BDS and FDS are also 2nd order accurate like
CDS, such equations are used at the boundary points or end points only to discretize Neumann

75
 𝜕𝑇 𝜕𝐶 𝜕𝑣
or mixed or flux (gradient) terms: −𝑘 𝜕𝑥 𝑜𝑟 − 𝐷 𝜕𝑥 𝑜𝑟 − µ 𝜕𝑥, whereas CDS is invariably used at
interior nodes.

Higher order 𝑓 ′ (𝑥𝑖 ) can also be obtained as CDS:

4 −4𝑓(𝑥𝑖+2 ) − 8𝑓(𝑥𝑖−1 ) + 8𝑓(𝑥𝑖+1 ) + 4𝑓(𝑥𝑖+2 )

{𝑓 ′ (𝑥𝑖 ) = + 0(ℎ4 )
𝑑𝑎𝑡𝑎 𝑝𝑜𝑖𝑛𝑡𝑠 12ℎ

( See the book by Ferziger)

Similarly, use eqns 1 & 2 to obtain

𝑓(𝑥𝑖+1 )−2𝑓(𝑥𝑖 )+𝑓(𝑥𝑖−1 )
𝑓′′(𝑥𝑖 ) = + 0(ℎ3 )
ℎ2

= 𝑓 ′′ (𝑥𝑖−1 ) = 𝑓 ′′ (𝑥𝑖+1 ):

𝑊ℎ𝑦? (𝑌𝑜𝑢 ℎ𝑎𝑣𝑒 2𝑛𝑑 𝑜𝑟𝑑𝑒𝑟 𝑝𝑜𝑙𝑦𝑛𝑜𝑚𝑖𝑎𝑙 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 3 𝑑𝑎𝑡𝑎 𝑝𝑜𝑖𝑛𝑡𝑠)

You should note that similar to 𝑓 ′ (𝑥𝑖 ), FDS and BDS or formulae can also be derived for
𝑓′′(𝑥𝑖 ) using three points or manipulating Taylor’s series or equations 1-4. However, these
equations are seldom or in fact, never used for most of chemical engineering problems. Why
not? Because the N-S or species or thermal energy balance equations (PDEs) are 2nd order, with
the 1st order (flux) boundary condition. Therefore, CDS for 𝑓 ′′ (𝑥𝑖 ), applied at interior nodes,
suffices.

To this end, apply Richardson’s extrapolation technique to obtain even higher order 𝑓′′

−𝑓(𝑥𝑖−2 ) + 16𝑓(𝑥𝑖−1 ) − 30𝑓(𝑥𝑖 ) + 16𝑓(𝑥𝑖+1 ) − 𝑓(𝑥𝑖+2 )

𝑓′′(𝑥𝑖 ) = + 0(ℎ4 )
12ℎ2
(Five data points)

Note the notations:

𝒇(𝒙𝒊 ) 𝐚𝐧𝐝 𝒚𝒊 (𝒙𝒊 ); 𝐟 ′ 𝐚𝐧𝐝 𝐟 ′′ , 𝐚𝐧𝐝 𝐲 ′ 𝐚𝐧𝐝 𝐲 ′′ .

𝑺𝒖𝒎𝒎𝒂𝒓𝒚 𝒂𝒕 𝒂 𝒈𝒍𝒂𝒏𝒄𝒆:

𝒇(𝒙) ≃ 𝑷𝒏 (𝒙), 𝒇′(𝒙) ≃ 𝑷′𝒏 (𝒙), 𝒇′′(𝒙) ≃ 𝑷′′𝒏 (𝒙)

76
 Lecture #14-15
Numerical Integration
Similar to numerical differentiation, the integration of 𝑓(𝑥), i.e., ∫ 𝑓(𝑥)𝑑𝑥 may be performed by
approximating 𝑓(𝑥) ≈ 𝑃𝑛 (𝑥), usually 𝑛 ≤ 3. Therefore, first calculate 𝑦𝑖 (𝑥𝑖 ) from 𝑓(𝑥), then
fit a polynomial through the ′𝑛′ data points and determine∫ 𝑃𝑛 𝑑𝑥 𝑜𝑟 area under the
approximated curve. Considering usual oscillations set in the fitted polynomial for 𝑛 ≥ 4, piece-
wise integration is performed. Detailed method and analysis are as follows.
𝑓(𝑥) 𝑃𝑛 (𝑥) is the interpolating 𝑓 𝑛 between (𝑥0 , ⋯ 𝑥𝑛 )
𝑃𝑛 (𝑥)
It can be shown/proposed that
𝑏
∫𝑎 𝑓(𝑥)𝑑𝑥 = ∑𝑛𝑘=0 𝑤𝑘 𝑓(𝑥𝑘 ); 𝑓(𝑥𝑘 ) = 𝑦(𝑥𝑘 )

𝑎 𝑥𝑖 𝑏 where, 𝑛 is the number of segments between ‘a’ and ‘b’

and ‘𝑤𝑖 ’ is the respective weight at 𝑥𝑖 . It is also known as

Quadrature formula. Let us look at the proposed formula differently. Apply Lagrange
interpolation between (𝑛 + 1) data points or 𝑛 segments:

𝑃𝑛 (𝑥) = ∑𝑛𝑘=0 𝑙𝑘 (𝑥)𝑦𝑘 (𝑦𝑘 ≡ 𝑦(𝑥𝑘 ) ≡ 𝑦𝑘 (𝑥𝑘 ))

𝑏 𝑏 𝑏
∫𝑎 𝑓(𝑥)𝑑𝑥 = ∫𝑎 ∑𝑛𝑘=0 𝑙𝑘 (𝑥)𝑦𝑘 𝑑𝑥 = ∑𝑛𝑘=0 ∫𝑎 𝑙𝑘 (𝑥)𝑦𝑘 𝑑𝑥

1 𝑏
= (𝑏 − 𝑎) ∑𝑛𝑘=0 𝐶 𝑘 𝑦𝑘 ; 𝐶 𝑘 = (𝑏−𝑎) ∫𝑎 𝑙𝑘 (𝑥)𝑑𝑥

Cotes number

- It is the same as the Quadrature formula ; n ≡ parameter(degree of polynomial)

(It can be shown ∑𝑛𝑘=0 𝐶 𝑘 = 1 (𝑝𝑢𝑡 𝑓(𝑥) = 1))

Therefore, we can have a table as follows:

77
 Newton-Cotes Coefficients (ends points included)

𝑛 𝑁 𝑁𝐶0𝑛 𝑁𝐶1𝑛 𝑁𝐶2𝑛 ⋯ ⋯ 𝑁𝐶0𝑁 𝐸𝑟𝑟𝑜𝑟

1 2 1 1 . 0183Δ3 𝑓′′
2 6 1 4 1 . 0035Δ5 𝑓′′′′
3 8 1 3 3 1 . 00016Δ5 𝑓′′′′
4 90 7 32 12 32 7 ⋯
5 288 19 75 50 50 75 19 ⋯
These are called closed N-C formula because they use functions at the end points also (𝑎, 𝑏).
Note, 𝑛 or the number of segments is usually 1 or 2 or 3, or the degree of polynomial is 1 or 2
or 3, respectively. Readers may have recognized that n = 1 corresponds to Trapezoidal rule
with ‘1⁄2’ coefficients of 𝑦𝑘 ; 𝑛 = 2 corresponds to Simpson’s ‘1⁄3 𝑟𝑑’ rule with
(1⁄6 , 2⁄3 , 1⁄6) coefficients, etc. For more details, you can refer the book by Ferziger or SKG.
The error in the numerical integration can be calculated from the remainder term of the
polynomial:
𝑦 𝑛+1 (𝜉)
⇒ 𝐼 = ∫(𝑒𝑟𝑟𝑜𝑟)𝑑𝑥 ⇒ 𝑤ℎ𝑒𝑟𝑒 ′𝑒𝑟𝑟𝑜𝑟 ′ = (𝑥 − 𝑥0 )(𝑥 − 𝑥1 ) ⋯ (𝑥 − 𝑥𝑛 )
𝑛 + 1!

𝑓𝑜𝑟 (𝑛 + 1) 𝑑𝑎𝑡𝑎 𝑝𝑜𝑖𝑛𝑡𝑠

⇒ SKG’s book uses Newton’s Interpolation formula to determine the various

integration formulae:

𝛼(𝛼−1)(Δ2 𝑦0 ) 𝛼(𝛼−1)(𝛼−2)𝛼⋯(𝛼−𝑛)(Δ𝑛+1 𝑦0 )
𝑦 (𝑥) = 𝑦(𝑥0 + 𝛼Δ𝑥) = 𝑦0 + 𝛼(Δ𝑦0 ) + + ⋯⋯
2! 𝑛+1!

𝑅𝑛
𝑟𝑑
To obtain the different methods for integration, viz. Trapezoidal’s rule, Simpson’s 1⁄3 , etc,
two approaches must give the identical results because polynomial of the highest degree that
can fit to (𝑛 + 1) data points is unique.

Take 𝑛 = 1
𝑓(𝑥)
𝑏 𝑃1 (𝑥)
∫𝑎 𝑓(𝑥)𝑑𝑥 = (𝑏 − 𝑎) ∑𝑛𝑘=0 𝐶𝑘𝑛 𝑓(𝑥𝑘 ) 𝑓(𝑏)

1 1 𝑓(𝑥) 𝑓(𝑎)
= (𝑏 − 𝑎) [2 𝑦𝑎 + 2 𝑦𝑏 ]

(𝑏−𝑎)
= (𝑦𝑎 + 𝑦𝑏 )
2
𝑎 𝑥 𝑏

78
 (𝑏−𝑎)
[𝑜𝑟 = (𝑓(𝑎) + 𝑓(𝑏))]
2

(Trapezoidal rule)

𝑛=2
𝑓(𝑥) 𝑃2 (𝑥)

𝑏 (𝑏 − 𝑎) 𝑎+𝑏
∫ 𝑓(𝑥)𝑑𝑥 = [𝑓(𝑎) + 4𝑓 ( ) + 𝑓(𝑏)]
𝑎 6 2
𝑟𝑑
𝑆𝑖𝑚𝑝𝑠𝑜𝑛′ 𝑠 1⁄3 𝑓𝑜𝑟𝑚𝑢𝑙𝑎
(ℎ𝑖𝑔ℎ𝑒𝑟 𝑜𝑟𝑑𝑒𝑟 𝑎𝑐𝑐𝑢𝑟𝑎𝑡𝑒) ΔX ΔX
𝑎 (𝑎 + 𝑏)⁄ 𝑏
2
(𝑁𝑜𝑡𝑒: 𝑃2 ≡ (𝑎 + 𝑏𝑥 + 𝑎𝑥 2 ) ≡ 𝑝𝑎𝑟𝑎𝑏𝑜𝑙𝑎)
Δ𝑋 𝑎+𝑏 𝑏−𝑎
= [𝑓(𝑎) + 4𝑓 ( ) + 𝑓(𝑏)] Δ𝑋 =
3 2 2

𝑛=3

𝑏 (𝑏 − 𝑎)
∫ 𝑓(𝑥)𝑑𝑥 = [𝑓(𝑎) + 3𝑓(𝑐) + 3𝑓(𝑑) + 𝑓(𝑏)]
𝑎 8
3Δ𝑋 (𝑏−𝑎)
= [ − 𝑑𝑖𝑡𝑡𝑜 − ], Δ𝑥 =
8 3

𝑡ℎ
- Simpson’s 3⁄8 formula.
In actual practice, applying a single Cotes formula over the entire range is rarely done.
Considering large error for large ′Δ𝑋′, interval is broken into sub-intervals & then applying
piece-wise quadrature formula and summing it over.

𝑋 ℎ
∫𝑋 𝑖 𝑓(𝑥)𝑑𝑥 = 2 [𝑓(𝑋𝑖−1 ) + 𝑓(𝑋𝑖 )]
𝑖−1

(𝑏−𝑎)
ℎ= (𝑛 + 1 data points or 𝑛 segments)
𝑛

𝑎 𝑋𝑖−1 𝑋𝑖 𝑏
𝑋0 𝑋𝑛

79
 𝑏 ℎ
Therefore, ∫𝑎 𝑓(𝑥)𝑑𝑥 = ∑𝑛𝑖=1 2 [𝑓(𝑥𝑖−1 ) + 𝑓(𝑥𝑖 )]
𝑛−1
ℎ
[𝑓(𝑥0 ) + 2 ∑ 𝑓(𝑥𝑘 ) + 𝑓(𝑥𝑛 )]
2
𝑘=1
= 𝑛 ∗
(𝑏 − 𝑎) 1
[∑ 𝑓(𝑥𝑘 ) − [𝑓(𝑎) + 𝑓(𝑏)]]
𝑛 2 }
𝑘=0

* Trapezoidal rule
𝒓𝒅
𝑺𝒊𝒎𝒑𝒔𝒐𝒏’𝒔 𝟏⁄𝟑 𝒓𝒖𝒍𝒆: (intervals must be even or data points must be odd),

𝑛 𝑁 𝑁𝐶0𝑛 𝑁𝐶1𝑛 𝑁𝐶2𝑛

2 6 1 4 1
(𝑏 − 𝑎)⁄
ℎ= 2
𝑏−𝑎
𝑏 𝑓(𝑥)𝑑𝑥 = ( 6
) (𝑓(𝑎) + 4𝑓(𝑐) + 𝑓(𝑏))
∫𝑎 𝑏−𝑎
𝑎 𝑐 𝑏
ℎ= 2

(𝑏−𝑎) (𝑛 𝑒𝑣𝑒𝑛 𝑠𝑒𝑔𝑚𝑒𝑛𝑡𝑠 𝑜𝑟

ℎ ℎ ℎ=
𝑋𝑖 𝑋𝑖 𝑋𝑖+1 𝑛 𝑛 + 1 𝑜𝑑𝑑 𝑑𝑎𝑡𝑎 )
𝑎 𝑋 𝑋𝑋 𝑏
𝑖−1 𝑖 𝑖+1
(𝑋0 ) (𝑋𝑛 )
𝑏 ℎ
∫𝑎 𝑓(𝑥)𝑑𝑥 = ∑ 3 [𝑓(𝑋𝑖−1 ) + 4 ∑ 𝑓(𝑋𝑖 ) + 𝑓(𝑋𝑖+1 )]

ℎ
= 3 (𝑓(𝑋0 ) + 4 ∑𝑛−1 𝑛−2
𝑘=1,3,5 +2 ∑𝑘=2,4,6 +𝑓(𝑋𝑛 ))

(Note that all even nodes ‘𝑋𝑘 ’ are summed/counted twice in calculating total area)

What do you do if 𝑛 ≡ odd or data are even? Apply the method to one data less so that n =
even. Add the area corresponding to the left-over segment, calculated using Trapezoidal rule!

One open NC method is popular: Mid-point rule, the method is frequently applied when the
𝑓 𝑛 is discontinuous at end points: a or b or both.
𝑏 𝑎+𝑏
∫𝑎 𝑓(𝑥)𝑑𝑥 = (𝑏 − 𝑎) [𝑓 ( 2
)]
𝑎 (𝑎+𝑏) 𝑏
2
80
 1 1
ex. ∫0 𝑑𝑥 (Note 𝑓(1) → ∞)
1−𝑥 2

Example:
Tw R

r
z
𝑣(r) T(r) Δr

fully developed region (2)

for both flow and temperature
𝑟2
𝑣 = 2𝑣̅ (1 − 𝑅2 )

𝑇−𝑇𝑤 𝑟4 7
𝜃= = −4𝑧 − 𝑟 2 + + 24
𝑇𝑤 4

(𝑧, 𝑟 are dimensionless quantities: 𝑟 ⁄𝑅 𝑎𝑛𝑑 𝑧⁄𝐿)

Consider the classical heat transfer/transport problem of SS heat transfer at constant wall
temperature in the fully developed region of a tubular laminar parabolic flow, often discussed in
TP lectures. The dimensionless SS temperature 𝜃(𝑧, 𝑟) in the fluid is analytically derived. See
the solution above. Determine the ‘mix-cup’ temperature at z = 0.5.

Soln : In a flow-system, mix-cup temperature is the radially emerged temperature, taking into
consideration the total heat (cal) carried away by the flowing fluid in unit time across the cross-
section of the tube.

𝑅
∫0 𝜌𝐶𝑝 𝑣(𝑟 ′ )𝑇(2𝜋𝑟 ′ )𝑑𝑟′
𝑚3
(𝑁𝑜𝑡𝑒: 𝑑𝑞 = 𝑣2𝜋𝑟𝑑𝑟 )
𝑠
𝑘𝑔⁄𝑚3 𝑐𝑎𝑙⁄𝑘𝑔 − 𝐾 𝐾 𝑅
𝑇̅ = 𝑅
∫0 𝜌𝐶𝑝 𝑣(𝑟 ′ )2𝜋𝑟 ′ 𝑑𝑟′ 𝑄 = ∫ 𝑣2𝜋𝑟𝑑𝑟
0
at a ref. temperature
( 𝑐𝑎𝑙⁄𝑠 − 𝐾 )
In the dimensionless quantities,

1
∫0 𝜃(𝑟, 𝑧)(1 − 𝑟 2 )𝑟𝑑𝑟
𝜃̅(𝑧) = 1
∫0 (1 − 𝑟 2 )𝑟𝑑𝑟
Simpson’s 1⁄3 rule can be applied by taking odd number (5) of data points over 𝑟 ≡ (0, 1).
1−0
Therefore # of segment = 4, step size, ℎ = = 0.25, Δ𝑟 ≡ ℎ = 0.25; 𝑟𝑖 = 𝑖Δ𝑟
4

81
Numerical integration is performed for both numerator and denominator terms.
1 1
𝐼 = ∫0 𝑓(𝑟)𝑑𝑟⁄∫0 Φ(𝑟)𝑑𝑟 (𝑛)
𝑘=0 1 2 3 4
where 𝑓(𝑟) = 𝜃(𝑟, 𝑧)(1 − 𝑟 2 )𝑟 & Φ(𝑟) = (1 − 𝑟 2 )𝑟

For both functions,

𝐼 = ℎ⁄3 [𝑓(𝑋0 ) + 4 ∑𝑛−1 𝑛−2

𝑘=1,3,5 +2 ∑𝑘=2,4 + 𝑓(𝑋𝑛 )]

= ℎ⁄3 [𝑓(𝑋0 ) + 4 ∑1,3 +2 ∑2 +𝑓(𝑋4 )]

Best way is to prepare a table for calculations:

(1 − 𝑟𝑘2 )𝑟𝑘 𝜃𝑘 (1 − 𝑟𝑘2 )𝑟𝑘

𝑘 𝑟𝑘 𝜃𝑘
or Φ(𝑟) 𝑜𝑟 𝑓(𝑟)

0 0 −1.7083 0.0 0.0

1 0.25 −1.7698  
𝐹𝑖𝑙𝑙 − 𝑢𝑝 𝑡ℎ𝑒
2 0.50 −1.9427  
𝑡𝑎𝑏𝑙𝑒
3 0.75 −2.1917  
4 1.00 −2.4583 0.0 0.0
∑ Add ∑ Add

Show,
𝑇−𝑇𝑤
̅
𝜃𝑚𝑖𝑥−𝑐𝑢𝑝 (0.5) ≡ −1.9976 =
𝑇 𝑤

(wall is colder than bulk fluid)

𝑇𝑤

𝑇𝑓𝑙𝑢𝑖𝑑

𝑇𝑤
What does decide the # of data points ‘n’ or the step size ‘h’ ? It is the level of desired accuracy.
Repeat the entire calculations for n = 9 and check your requirement of an improved 𝜃𝑚𝑖𝑥−𝑐𝑢𝑝 ̅
from the previous value -1.9976. Sometimes you have to decrease ‘n’ or increase ‘h’ if you have
exceeded accuracy limit!

Error Analysis

Let us take the simplest Trapezoidal rule applied to one segment over (a,b).

Start from Newton’s Difference Interpolation formula:

82
 𝛼(𝛼 − 1) 2
𝑓(𝑥) = 𝑓(𝑥0 ) + 𝛼Δ𝑓(𝑥0 ) + Δ 𝑓(𝑥0 ) + ⋯ 𝑅𝑛
2!
𝛼(𝛼−1)⋯(𝛼−𝑛)
𝑅𝑛 = Δ𝑛+1 𝑓(𝜉) 𝑤ℎ𝑒𝑟𝑒 𝑥0 < 𝜉 < 𝑥
𝑛+1!

𝑥−𝑥0
𝛼= 𝑜𝑟 𝑑𝑥 = ℎ 𝑑𝛼 ; 𝑥0 = 𝑎,
ℎ

If there is only one segment, ℎ = (𝑏 − 𝑎)

𝑏 1 𝛼(𝛼−1)
𝐼 = ∫𝑎 𝑓(𝑥)𝑑𝑥 = ℎ ∫0 (𝑓(𝑥0 ) + 𝛼Δ𝑓(𝑥0 ) + Δ2 𝑓(𝜉)) 𝑑𝑥
2!
𝑃1 = (𝑎𝑥 + 𝑏)
1 𝛼3 𝛼2 1 2
= ℎ [𝑓(𝑎) + (𝑓(𝑏) − 𝑓(𝑎)) + ( − ) ∫ ℎ 𝑓′′(𝜉)]
2 6 4 0

𝑓(𝑎)+𝑓(𝑏) 1 Δ2 𝑓(𝜉)
= ℎ[ ] − 12 ℎ3 𝑓 ′′ (𝜉) (𝑓 ′′ (𝜉) = )
2 2ℎ

Therefore, error on one segment is 1⁄2 ℎ3 𝑓 ′′ (𝜉) 𝑜𝑟 0(ℎ3 )

If there are n segments between (a & b) error should also be summed up to

1 1
ℎ3 ∑𝑛 𝑓 ′′ (𝜉) = 12 ℎ2 𝑓 ′′ (𝜉 ̅)(𝑏 − 𝑎) 𝑜𝑟 0(ℎ2 )
12

∑ 𝑓 ′′ (𝜉) ℎ ∑ 𝑓 ′′ (𝜉)
𝑁𝑜𝑡𝑒: 𝑓 ′′ (𝜉 ̅) = = .
𝑛 (𝑏−𝑎)

Similarly, for the error estimation of Simpson’s 1⁄3 rule start with 𝑃2 = 𝑎𝑥 2 + 𝑏𝑥 + 𝑐, 𝑜𝑟

𝛼(𝛼−1)(𝛼−2)Δ3 𝑓(𝜉)
𝑅𝑛 = over one segment or 3 data pts.
3!

to show that error ≡ 0(ℎ5 )𝑓 ′′′′ (𝜉) and if summed over ‘n’ segments error = 0(ℎ4 )𝑓 ′′′′ (𝜉 ̅).

error(one segment) ≡ 0(h5 )

Simpson’s 3⁄8 𝑟𝑢𝑙𝑒:
error( segments) ≡ 0(h4 )

A common question is asked: Show that Simpson’s 1⁄3 𝑟𝑢𝑙𝑒 gives the error corresponding to
a cubic interpolating function, although it is based on a parabola or a quadratic polynomial. See
Chhapra and Cannale's book for the details.

Mid-Term

83
 Lecture #15-16
1st order ODE
A set of ODEs takes the following form:
𝑑𝑦1
= 𝑓1 (𝑡, 𝑦1 , 𝑦2 ⋯ 𝑦𝑛 )
𝑑𝑡 𝑑𝑦̅
⋮ } 𝑜𝑟 ̅ 𝑦̅)
= 𝑓(𝑡,
𝑑𝑦𝑛 𝑑𝑡
= 𝑓𝑛 (𝑡, 𝑦1 , 𝑦2 ⋯ 𝑦𝑛 )
𝑑𝑡

Requires initial conditions to solve 𝑦̅ .

𝑡 = 0, 𝑦1 = 𝑦10 … . . 𝑦𝑛 = 𝑦𝑛0 , 𝑒𝑡𝑐 𝑜𝑟 ̅𝑦 = 𝑦̅0

𝑑𝑦̅
̅ 𝑦̅ ) with 𝑥 = 0, 𝑦̅ = 𝑦̅0 .
Note that ODEs can also be1st order on x (space) & 𝑑𝑥 = 𝑓(𝑥,

⇒ We explore numerical solution to ODEs starting with Euler’s forward and backward
methods. These methods are basics, and also set the tone for the stability analysis. Let us begin
with one ODE.

Euler’s Method
𝑑𝑦
= 𝑓(𝑥, 𝑦) ; 𝑦(𝑥0 ) = 𝑦0
𝑑𝑥
𝑦0 𝑦1 𝑦𝑛 𝑦𝑛+1
𝑦𝑛+1 −𝑦𝑛
𝑜𝑟 , = 𝑓(𝑥𝑛 , 𝑦𝑛 ) + 0(ℎ2 )
ℎ 𝑥0 𝑥1 𝑥𝑛 𝑥𝑛+1
(This is the general ‘discretization’ of a derivative at the ‘n’ th grid. Recall the lecture on
numerical differentiation. We have used FDS.)

or 𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑓(𝑥𝑛 , 𝑦𝑛 ) + 0(ℎ2 )

𝑦𝑛′

Note: The RHS terms are known to be 1st order accurate or have 2nd order error.

From the previous calculations, start from 𝑦(𝑥 = 0 ) = 𝑦0 → 𝑦1 → 𝑦2 ⋯ 𝑒𝑡𝑐. Therefore, you
march forward and the method is called Euler’s forward or explicit method. Think carefully.
The method is all about taking a linear slope/tangent at
(xn , yn ) and moving on to (xn+1 , yn+1 ).

84
Graphically,

It is all about
calculating true derivatives
Numerical (gradients) at xn as
calculations
you march forward:
Theoretical y(x) y0′ = f(x0 , y0 )
𝑌 y1′ = f(x1 , y1 )
⋮
𝑦0 } yn′ = f(xn , yn )
𝑥0 𝑥1 𝑥2 𝑋

It is clear that smaller is ‘h’ (step-size), less is the error, or accuracy is higher: 0(ℎ2 ). However,
there is a price for long CPU time. Also, stability may become an issue. Let us look at the
stability aspect of the method.
𝑑𝑦
= 𝑓(𝑥, 𝑦) = 𝑓(𝑥0 + Δ𝑥, 𝑦0 + Δ𝑦)
𝑑𝑥

𝑑𝑓 𝑑𝑓
= 𝑓(𝑥0 , 𝑦0 ) + (𝑦 − 𝑦0 ) | + (𝑥 − 𝑥0 ) | + 0(Δ𝑥 2 , Δ𝑦 2 )
𝑑𝑦 𝑥 𝑑𝑥 𝑥0 ,𝑦0
0 ,𝑦0

Multi-variable Taylor series can be applied to estimate gradient 𝑦 ′ (𝑥0 , 𝑦0 ):

= 𝛼𝑦 + 𝛽𝑥 + 𝜈
term const.
gives exponential
for linear
solution solution

Model equation for testing stability.

𝑑𝑦
= 𝛼𝑦 ∶ (Note: Stability is governed by the
𝑑𝑥
homogeneous part of the solution)

Let us look at differently: Introduce perturbation or find true value, 𝑦 .

𝑑𝑦
Therefore, = 𝛼𝑦̅ + 𝛽𝑥 + 𝜈
𝑑𝑥

𝑑𝜉
𝑜𝑟, = 𝛼(𝑦 − 𝑦) = 𝛼𝜉 (𝜉 = 𝑦 − 𝑦)
𝑑𝑥

We have the same inference: error satisfies homogeneous part of the solution. The equation
describes how ‘𝜉’ can grow because of a small perturbation.

At 𝑥 = 𝑥0 , 𝑦 = 𝑦0 ⇒ exact solution: 𝑦 = 𝑦0 𝑒 𝛼𝑥 𝐼𝑓 𝛼 > 0, f n grows, α < 0, f n is bounded.

85
Now Apply Euler’s method:

𝑦𝑛+1 − 𝑦𝑛 = ℎ𝑓(𝑥𝑛 , 𝑦𝑛 ) = ℎ𝑦𝑛′ = ℎ(𝛼 𝑦𝑛 )

𝑦𝑛+1 = 𝑦𝑛 (1 + 𝛼ℎ)

𝑦1 = 𝑦0 (1 + 𝛼ℎ), 𝑦2 = 𝑦0 (1 + 𝛼ℎ)2

𝑦𝑛 = 𝑦0 (1 + 𝛼ℎ)𝑛 ⇒ Numerical solution

If α > 0 function grows or unbounded

]
If α < 0 function may be bounded

Let us write, 𝛼 = 𝛼𝑟 + 𝑖𝛼𝐼

𝐼(𝛼𝑖 ℎ)
𝑦𝑛 = 𝑦0 (1 + 𝛼𝑟 ℎ + 𝑖𝛼𝐼 ℎ)𝑛

For function to be bounded

|(1 + 𝛼𝑟 ℎ) + 𝑖𝛼𝐼 ℎ| < 1

−2 −1 0 𝑅
(1 + 𝛼𝑟 ℎ) + 2
𝛼𝑖2 ℎ2 <1 (𝛼𝑟 ℎ)

0 > 𝛼𝑟 ℎ > −2 if αi = 0 for stability Stable region

(depends on ℎ and αr )
Note: 𝛼 = −𝑣𝑒, the exact solution is always bounded! Therefore, Euler’s forward/explicit
numerical method is conditionally stable. Example:
𝑑𝑦
= −𝑦 (𝑥 = 0, 𝑦 = 1) ⇒ 𝑦 = 𝑒 −𝑥
𝑑𝑥

Numerical soln : exact solution: 1.0 ′bounded (theroteical)solution′

𝑌
𝛼 = −1; 𝛼𝑟 = −1, 𝛼𝐼 = 0
𝑋
0 > 𝛼𝑟 ℎ > −2 for stability

or ℎ < +2

⇒ 𝑦𝑛+1 − 𝑦𝑛 = ℎ𝑓(𝑥𝑛 , 𝑦𝑛 ) = −ℎ𝑦𝑛 ⇒ 𝑦𝑛 = (1 − ℎ)𝑛

Plot to see that oscillations start creeping in h > 2

ℎ>5
exact
𝑙𝑜𝑔 1.0 2.0
(bounded)
𝑌 ℎ=1
ℎ = 0.001
1.0
error increases ℎ = 0.1 ⇒
ℎ = 0.8 0
−2
10 −7 0 6 −1
2 4 𝑋 (unbounded) 𝑋
86
Backward or Implicit Euler:
𝑑𝑦
= 𝑓(𝑥, 𝑦), 𝑦(𝑥0 ) = 𝑦0
𝑑𝑥

𝑦𝑛+1 −𝑦𝑛 (Note this BDS may require iteration)

= 𝑓(𝑥𝑛+1 , 𝑦𝑛+1 ) + 0(ℎ2 )
ℎ , maybe NR method; you can not march!

Test it on the model eqn

𝑦𝑛+1 − 𝑦𝑛 = ℎ𝑓(𝑥𝑛+1 , 𝑦𝑛+1 ) ⇒ dy
= αy for the stability analysis
dx

𝑦𝑛+1 = 𝑦𝑛 + ℎ(𝛼𝑦𝑛+1 )

𝑦𝑛+1 = 𝑦𝑛 ⁄(1 − 𝛼ℎ)

= 𝑦0 (1 − 𝛼𝑟 ℎ − 𝑖𝛼𝑖 ℎ)−𝑛

For 𝑓 𝑛 to be bounded |(1 − 𝛼𝑟 ℎ − 𝑖𝛼𝑖 ℎ)| > 1

𝑜𝑟 (1 − 𝛼𝑟 ℎ)2 + 𝛼𝑖2 ℎ2 > 1

𝑜𝑟 |(1 − 𝛼𝑟 ℎ)| > 1 if 𝛼𝑖 = 0

Clearly, 0 > 𝛼𝑟 ℎ > 2

Draw the region:

(𝛼𝑖 ℎ)
For all (αr h) in the left
half of the plane αr < 0 𝑎𝑛𝑑
⇒ {
2 the method is unconditionally
0 1 (𝛼𝑟 ℎ) stable
For all ℎ outside the circle
the method is stable (or αr h > 2)

It is clear the stability region for the implicit Euler is much more larger than that of the explicit
Euler.

Notes: Error decreases with decreasing step sizes (h): 0(ℎ𝑛 ), however, at the expense of CPU
time. This is not surprising. The actual ‘issue’ is with increasing step sizes. There is a tendency
to speed up computation by taking large step-size. An explicit method like Euler Forward may
not permit or allow you to do so, even if the level of desired accuracy may be acceptable.
Instability may set in at large ‘h’. On the other hand, Euler Backward or an implicit method
permits you to use relatively larger ‘h’ without instability, yielding a bounded solution.

87
⇒ ‘Error’ must not be mixed up with the ‘instability’. A method may be relatively higher order
accurate, but may show instability at small step-size. In general, implicit methods are more
stable than explicit methods, but require iterations to compute 𝑦𝑛+1 . Examples:
𝑑𝑦
= 𝑓(𝑥, 𝑦) = 𝛼𝑦 2
𝑑𝑥

Explicit: 𝑦𝑛+1 − 𝑦𝑛 = ℎ𝛼𝑦𝑛2

2
Implicit: 𝑦𝑛+1 − 𝑦𝑛 = ℎ 𝛼𝑦𝑛+1 (requires iterations or NR method? )

⇒ " A method is stable if it produces a bounded solution when it is supposed to and is unstable
𝑑𝑦
if it does not." Therefore, while solving = −𝑦, if the numerical method produces a stable
𝑑𝑥
𝑑𝑦
solution, it is stable, else unstable. On the other hand, there is no instability ‘issue’ with =
𝑑𝑥
+𝑦 because the exact solution itself is unbounded. Therefore, both Euler Forward(explicit) or
Backward(implicit) will produce an unbounded solutions!
𝑑𝑦
Get back to 𝑑𝑥 = −𝑦(𝛼 = −1) and 𝑦𝑛 = (1 + ℎ)−𝑛

Error increases with increasing ‘h’ but the

1.0 error
h=1 solution remains stable. Compare to Forward Euler
y
exact 0.2 when 𝛼 < 2 for a stable solution.
0 0.1
X
⇒ Trapezoidal rule (0(ℎ2 )):
ℎ
𝑦𝑛+1 − 𝑦𝑛 = 2 [𝑓(𝑥𝑛 , 𝑦𝑛 ) + 𝑓(𝑥𝑛+1 , 𝑦𝑛+1 )]

(Implicit method)

It can also be tested on the model equation:

𝑑𝑦
= 𝛼𝑦
𝑑𝑥

𝛼ℎ
𝑜𝑟 𝑦𝑛+1 − 𝑦𝑛 = [𝑦 + 𝑦𝑛+1 ]
2 𝑛
(1+𝛼ℎ⁄2)
𝑦𝑛+1 = 𝑦𝑛 (1−𝛼ℎ⁄2)

1 + 𝛼ℎ⁄2 𝑛
𝑜𝑟 𝑦𝑛 = 𝑦0 ( )
1 − 𝛼ℎ⁄2)

88
 |𝑧| < 1
(1+𝛼ℎ⁄2)
For stability |(1−𝛼ℎ⁄2)| < 1 𝑜𝑟

Each method has a stability region for 𝛼𝑟 ℎ

selecting the step-size, ‘h’ for calculation without instability.

See the textbooks for details. Stability region

89
 Lecture #17
Single Step Methods
Forward or Backward Euler is the simplest single step method to solve ODEs. The methods
are 1st order accurate. A general higher order method can be derived considering that it is all
about choosing a suitable gradient at (𝑥𝑛 , 𝑦𝑛 ) 𝑜𝑟 (𝑡𝑛 , 𝑦𝑛 ) to calculate 𝑦𝑛+1 (𝑥𝑛+1 ):
𝑑𝑦
= 𝑓(𝑡, 𝑦); 𝑦(0) = 𝑦0
𝑑𝑡

𝑦𝑛+1 = 𝑦𝑛 + ℎ𝜙(𝑡𝑛 , 𝑦𝑛 , ℎ) (ℎ = 𝑠𝑡𝑒𝑝 𝑠𝑖𝑧𝑒 = Δ𝑡)

𝑑𝑦
It is all about finding: 𝜙(𝑡𝑛 , 𝑦𝑛 , ℎ) = ( 𝑑𝑡 ) =?
𝑡𝑛 ,𝑦𝑛

Propose as (𝑠𝑙𝑜𝑝𝑒)

𝑦𝑛+1 = 𝑦𝑛 + ℎ[𝑎𝑘1 + 𝑏𝑘2 ] 1 ⇒ 𝑔𝑒𝑛𝑒𝑟𝑎𝑙 𝑚𝑒𝑡ℎ𝑜𝑑.

Here, 𝑘1 and 𝑘2 are the slopes with weights a & b respectively ( Forward Euler may be
considered to assume 𝑎 = 1, 𝑏 = 0 and 𝑘1 = 𝑦𝑡′𝑛,𝑦𝑛 , whereas Backward Euler assumes 𝑎 =
0, 𝑏 = 1, 𝑘2 = 𝑦𝑡′𝑛+1 ,𝑦𝑛+1 )

𝑘1 = 𝑓(𝑡𝑛 , 𝑦𝑛 ) Thus, the model has

Here, }
𝑘2 = 𝑓(𝑡𝑛 + 𝑝ℎ, 𝑦𝑛 + 𝑞(ℎ𝑘1 ) 4 parameters; a, b, p, q.
𝑦
Thus k 2 (slope)is calculated
𝑦𝑛+1
at (t n + ph), fraction less than
𝑘2 ℎ𝑘1
𝑦𝑛 + 𝑞(ℎ𝑘1 )
(t n + h) or t n+1 , as x − coordinate,
𝑦𝑛
𝑘1 𝑞(ℎ𝑘1 ) and yn + q(hk1 ), fraction less
ℎ than (yn + hk1 ) or yn+1 , as y −
𝑡𝑛 𝑡𝑛+1
{ coordinate.
(𝑡𝑛 + 𝑝ℎ)

𝑘2 = 𝑓(𝑡𝑛 , 𝑦𝑛 ) + 𝑓𝑡 (𝑡𝑛 , 𝑦𝑛 ). 𝑝ℎ + 𝑓𝑌 (𝑡𝑛 , 𝑦𝑛 )𝑞ℎ𝑘1 + 0(ℎ2 )

-Taylor multi-variable series applied on 𝑘2 (𝑡, 𝑦) with Δ𝑡 = 𝑝ℎ, Δ𝑦 = 𝑞ℎ𝑘1

Substituting in (1)

𝑦𝑛+1 = 𝑦𝑛 + ℎ[𝑎𝑓(𝑡𝑛 , 𝑦𝑛 ) + 𝑏𝑓(𝑡𝑛 , 𝑦𝑛 )] + ℎ2 [𝑏𝑝𝑓𝑡 (𝑡𝑛 , 𝑦𝑛 ) + 𝑏𝑞𝑓𝑓𝑦 (𝑡𝑛 , 𝑦𝑛 )] + 0(ℎ3 ) − (2)

90
 𝑑𝑦
Also, restart from = 𝑓(𝑡, 𝑦(𝑡)) as a total derivative on t.
𝑑𝑡

ℎ2
or 𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑓 + 𝑓 ′ + 0(ℎ3 ): (Single variable(t) Taylor − Series)
2!

𝑑𝑓 𝜕𝑓 𝜕𝑓 𝑑𝑦
where, 𝑓 ′ = (𝑡𝑛 , 𝑦𝑛 ) = [ + 𝜕𝑦 ∙ 𝑑𝑡 ] = [𝑓𝑡 + 𝑓𝑦 𝑓]
𝑑𝑡 𝜕𝑡 𝑡𝑛 ,𝑦𝑛

On substitution,

ℎ2 ℎ2
𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑓 + 𝑓𝑡 (𝑡𝑛 , 𝑦𝑛 ) + 𝑓𝑦 𝑓 + 0(ℎ3 ) (3)
2 2

Thus, model equation (2) can be equated with the fundamental equation (3) derived for 𝑦𝑛+1 ;

𝑎+𝑏 =1
1
( 𝑏𝑝 = ⁄2 )
𝑏𝑞 = 1⁄2

Case 1: 𝑎 = 1⁄2 , 𝑏 = 1⁄2 , 𝑝 = 𝑞 = 1

ℎ
𝑦𝑛+1 = 𝑦𝑛 + 2 [𝑓(𝑡𝑛 , 𝑦𝑛 ) + 𝑓(𝑡𝑛 + ℎ, 𝑦𝑛 + ℎ𝑦𝑛′ )] + 0(ℎ3 )
𝑘1 𝑘2
Write differently,
∗
𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑓(𝑥𝑛 , 𝑦𝑛 ) predictor full step
ℎ ∗ )]
𝑦𝑛+1 = 𝑦𝑛 + 2 [𝑓(𝑡𝑛 , 𝑦𝑛 ) + 𝑓(𝑡𝑛 + ℎ, 𝑦𝑛+1
corrector full step
𝑘1 𝑘2 }

Look at graphically,

𝑦
∗ 𝑘2
𝑦𝑛+1
𝑦𝑛+1 (𝑘1 + 𝑘2 )⁄
2=𝑘
𝑦𝑛 𝑘1

𝑡𝑛 𝑡𝑛+1
𝑡
∗
Therefore, first predict 𝑦𝑛+1 and then correct as 𝑦𝑛+1 by taking the average of the slopes,
𝑘1 and 𝑘2 ∶ 𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑘̅

91
This method is called Heun’s improved method or modified Euler method or Runga-Kutta (RK-
2) method.

Case 2: 𝑎 = 0 , 𝑏 = 1, 𝑝 = 𝑞 = 1⁄2

ℎ ℎ
𝑦𝑛+1 = 𝑦𝑛 + ℎ [𝑓 (𝑡𝑛 + 2 , 𝑦𝑛 + 2 𝑦𝑛′ )]

Write differently,
∗
𝑦𝑛+ 1 ′
1⁄ ≡ 𝑦𝑛 + ⁄2 ℎ𝑦𝑛 (Euler predictor 1⁄2 step)
2

𝑦𝑛+1 ≡ 𝑦𝑛 + ℎ [𝑓 (𝑡𝑛 + ℎ⁄2 , 𝑦𝑛+

∗
1⁄ )] : Mid − point corrector full step
2

Graphically,
𝑦 (1) Start with slope 1
𝑌𝑛+1 Reach half step.
∗
𝑌𝑛+ 1⁄
(2) Calculate slope 2
2 1 (3)Re − start with the slope
𝑦𝑛 }
2 2 to reach full step.

𝑡𝑛+1 𝑡
𝑡𝑛 𝑡𝑛+1⁄
2
(𝑡𝑛 + ℎ)

- This method is called midpoint method.

3. RK-4
ℎ
𝑦𝑛+1 = 𝑦𝑛 + 6 [𝑘1 + 2𝑘2 + 2𝑘3 + 𝑘4 ] + 0(ℎ5 )

𝑘̅ = avg of four slopes with

more weights at mid points.

𝑘1 = 𝑓(𝑥𝑛 , 𝑦𝑛 )

𝑘2 = 𝑓(𝑥𝑛 + ℎ⁄2 , 𝑦𝑛 + 𝑘1 ℎ⁄2) 𝑜𝑟 ∗

𝑓(𝑥𝑛+1⁄ , 𝑦𝑛+ 1⁄ )
2 2

∗
𝑦𝑛+ 1
1⁄ ∶ 𝐸𝑢𝑙𝑒𝑟 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑜𝑟 ⁄2 𝑠𝑡𝑒𝑝
2

𝑘3 = 𝑓(𝑥𝑛 + ℎ⁄2 , 𝑦𝑛 + 𝑘2 ℎ⁄2) 𝑜𝑟 ∗∗

𝑓(𝑥𝑛+1⁄ , 𝑦𝑛+ 1⁄ )
2 2

∗∗
𝑦𝑛+ 1⁄ ∶ 𝐵𝑎𝑐𝑘𝑤𝑎𝑟𝑑 𝐸𝑢𝑙𝑒𝑟
1⁄ 𝑠𝑡𝑒𝑝 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑖𝑜𝑛
2 2

92
𝑘4 = 𝑓(𝑥𝑛 + ℎ, 𝑦𝑛 + 𝑘3 ℎ): 𝑀𝑖𝑑 𝑝𝑜𝑖𝑛𝑡 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑜𝑟 𝑓𝑢𝑙𝑙 𝑠𝑡𝑒𝑝
∗∗∗
𝑦𝑛+1

Write differently,

∗
𝑦𝑛+ ℎ 1
1⁄ = 𝑦𝑛 + ⁄2 𝑓(𝑥𝑛 , 𝑦𝑛 ) ∶ 𝐸𝑃 ⁄2 𝑠𝑡𝑒𝑝
2

∗∗
𝑦𝑛+ ℎ ∗ 1
1⁄ = 𝑦𝑛 + ⁄2 𝑓 (𝑥𝑛+1⁄ , 𝑦𝑛+1⁄ ) ∶ 𝐵𝐸 ⁄2 𝑠𝑡𝑒𝑝 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑜𝑟
2 2 2

∗∗∗ ∗∗
𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑓 (𝑥𝑛+1⁄ , 𝑦𝑛+ 1⁄ ) : 𝑀𝑖𝑑 𝑝𝑜𝑖𝑛𝑡 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑜𝑟 𝑓𝑢𝑙𝑙 𝑠𝑡𝑒𝑝.
2 2

ℎ ∗ ∗∗ ∗∗∗
𝑦𝑛+1 = 𝑦𝑛 + [𝑓(𝑥𝑛 , 𝑦𝑛 ) + 2𝑓(𝑥𝑛+1⁄ , 𝑦𝑛+ 1⁄ ) + 2𝑓 (𝑥𝑛+1⁄ , 𝑦𝑛+1⁄ ) + 𝑓(𝑥𝑛+1 , 𝑦𝑛+1 )]
6 2 2 2 2
∶ 𝑆𝑖𝑚𝑝𝑠𝑜𝑛 𝑓𝑢𝑙𝑙 𝑠𝑡𝑒𝑝 𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑜𝑟.

Graphically,
𝑌
∗∗∗
𝑌𝑛+1 𝑘4

𝑦𝑛+1 𝑘3
𝑘̅
∗ 𝑘2
𝑌𝑛+ 1⁄
2
∗∗
𝑌𝑛+ 1⁄ 𝑘3
2
𝑘2
𝑦𝑛 𝑘1

𝑥𝑛 ℎ⁄ 𝑥𝑛+1⁄ ℎ⁄ 𝑥𝑛+1 𝑋
2 2 2

𝑘̅ = (𝑘1 + 2𝑘2 + 2𝑘3 + 𝑘4 )⁄6

Notes: (1) RK-4 is a 4th order accurate method!

(2) The method is explicit (does not require iterations to calculate 𝑦𝑛+1 ). You can march ahead
from 𝑥𝑛 𝑡𝑜 𝑥𝑛+1 . Yet, considering that it makes use of the value 𝑦𝑛+1 at 𝑥𝑛+1 ahead of 𝑥𝑛 via a
predictor step, the method possesses some ‘characteristics’ of an implicit method. Therefore, the
method also produces a stable solution at a relatively larger step size.

(3) Coding (program) is simple.

(4) Even if two or more number of simultaneous ODEs are coupled, the method is explicitly
applied w/o requiring iterations,

93
(5) Considering that the method is explicit, non-linearity is not an issue.

(6) Considering that the method is explicit, the implementation is identical for a system of
ODEs.

See example below:

𝑑𝑦1 with choose

Solve: = −100𝑦1
𝑑𝑡 𝑦1 (0) = 2 ℎ = Δ𝑡 = 0.02
𝑑𝑦2
= 2𝑦1 − 𝑦2 𝑦2 (0) = 1
𝑑𝑡

𝑑𝑦
Write alternatively, solve = 𝑓(𝑥, 𝑦) ; 𝑦(𝑡, 0) = 𝑦0
𝑑𝑡

𝑦′ −100 0 𝑦1 𝑦1 2
or { 1′ } = [ ]{ 𝑦 }; { 𝑦 } ={ }
𝑦2 2 −1 2 2 0 1

or

𝑦′ −100𝑦1 𝑦1 2
{ 1′ } = { }; { 𝑦 } ={ }
𝑦2 2𝑦1 − 𝑦2 2 0 1

𝑘̅1 = 𝑓(𝑡
̅ 𝑛 , 𝑦𝑛 )
𝑘′ −100 × 2 −200
𝑘1 = [ ′′1 ] =[ ]=[ ]
𝑘1 2×2−1 3

𝑘2′ ̅ ̅
1
̅
𝑘2 = [ ′′ ] ⇒ 𝑘2 = 𝑓 (𝑡𝑛+1⁄2 , 𝑦𝑛 + 𝑘1 ℎ)
𝑘2 2
1
0.01, 2 + (−200) × 0.02
2 0.01, 0.00 0.00
= 𝑓[ ] = 𝑓[ ]=[ ]
1 0.01, 1.03 −1.03
0.01, 1 + (3) × 0.02
2

𝑘3′ ̅
1
̅
𝑘3 = [ ′′ ] ⇒ 𝑘3 = 𝑓(𝑡𝑛+1⁄2 , 𝑦𝑛 + 𝑘2 ℎ)
𝑘3 2
. 02
0.01, 2+0×
= 𝑓[ 2 ] = 𝑓 [0.01, 2.000
]=[
−100 × 2
]
0.02 0.01, 0.9897 2 × 2 − 0.9897
0.01, 1 − 1.03 ×
2
−200
=[ ]
3.0103

𝑘4′
𝑘4 = [ ] ⇒ 𝑘̅4 = 𝑓(𝑡𝑛+1 , 𝑦𝑛 + ℎ𝑘̅3 )
𝑘4′′

94
 0.02, 2 − 0.02 × 200 0.02, −2
= 𝑓[ ] = 𝑓[ ]
0.02, 1 + 0.02 × 3.0103 0.02, 1.060206

−100 × −2 200
=[ ]=[ ]
2 × −2 − 1.060206 −5.060206

𝑦1 2 0.02 −200 + 2 × 0 − 2 × 200 + 200

{𝑦 } = ⌊1⌋ + 6 [ ]
2 3 − 2 × 1.03 + 2 × 3.0103 − 5.060206

0.66667
=[ ] 𝐴𝑛𝑠.
1.00633
⇒ How to choose h? Choose a reasonable ‘h’ depending on the physics of the problem. Solve
and re-do the calculation for ℎ⁄2 and check the convergence. Sometimes, one may have to
increase the step-size if the accuracy can be relaxed.

⇒ There are R-K-Mersen or R-K-Fehlberg automatic step-size control method where ‘h’ is
adjusted (either h → 2h or h → h⁄2 ) depending upon the error in the RK-4 solution relative
to that from error in a higher order method proposed by Mersen or Fehlberg.

In summary, the single step methods to solve 1st order ODE(s) are all about finding a
relatively “accurate” slope or tanӨ to reach the target at “h” step forward. The difference
between the methods is the computation of the slopes, viz.,

(t n , yn ) or (t n+1 , yn+1 ) or (t n+1/2 , yn+1/2 ) or a combintation of these?

Thus, the methods are called differently (explicit/implicit) and have different orders
(1st/2nd/4th, etc) of accuracy and different stability regions (smaller or bigger):

𝑦𝑛+1 = 𝑦𝑛 + ℎ 𝜙(𝑡𝑛 , 𝑦𝑛 ): 𝐹𝑜𝑟𝑤𝑎𝑟𝑑 𝐸𝑢𝑙𝑒𝑟

𝑦𝑛+1 = 𝑦𝑛 + ℎ 𝜙(𝑡𝑛+1 , 𝑦𝑛+1 ): 𝐵𝑎𝑐𝑘𝑤𝑎𝑟𝑑 𝐸𝑢𝑙𝑒𝑟

𝑦𝑛+1 = 𝑦𝑛 + ℎ [𝜙(𝑡𝑛 , 𝑦𝑛 ) + (𝑡𝑛+1 , 𝑦𝑛+1 )]/2: 𝑇𝑟𝑎𝑝𝑒𝑧𝑜𝑖𝑑𝑎𝑙 𝑅𝑢𝑙𝑒

𝑦𝑛+1 = 𝑦𝑛 + ℎ 𝑘: RK-4 predictor-corrector

𝑤ℎ𝑒𝑟𝑒, 𝑘 𝑖𝑠 𝑠𝑜𝑚𝑒 𝑤𝑒𝑖𝑔ℎ𝑡𝑎𝑔𝑒 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑜𝑓 𝑠𝑙𝑜𝑝𝑒𝑠 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑎𝑡 𝑠𝑜𝑚𝑒 𝑖𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑎𝑖𝑡𝑒 𝑙𝑜𝑐𝑎𝑡𝑖𝑜𝑛𝑠.

Note: All methods predict/estimate the functional value, 𝑦𝑛+1 linearly starting with initial
condition, 𝑦𝑛 at 𝑥𝑛.

95
 Lecture #18
Example 1 (RK-4)
Consider a special (diameter = 10 cm) ball made of steel k = 40
𝑊 ⁄𝑚 − 𝐾, 𝜌 = 8000 𝑘𝑔⁄𝑚3 , 𝐶𝑝 = 400 𝐽⁄𝑘𝑔 − 𝑘). The initial temperature of the ball is
300 K. It is immersed in a large oil tank at 400 K. The convective heat transfer coefficient, h at
the sphere surface is 3000 𝑊 ⁄𝑚2 − 𝑘. Assume that there is no radial temperature gradient
inside the ball. Use the R-K-4 numerical technique to solve the governing first order energy
balance equation for predicting the temperature of the sphere at 𝑡 = 1 𝑚𝑖𝑛 after it was
immersed in the oil tank. Use Δ𝑡 = 10 𝑠. Repeat calculations for Δ𝑡 = 5 𝑠. Draw graphical
depiction of the slopes comprising all stages of the RK method. Do calculations up to 4 digits
after decimal. Calculate the rate of temperature-decrease at 𝑡 = 0, 0.5 and 1.0 min.

ℎ
𝑅 = 5.0 𝑐𝑚
𝑡 = 0 𝑇 = 300 𝐾 𝑇(𝑡) 𝑡
𝑇𝑜 = 400 𝐾
𝑅

The energy balance (transient/unsteady equation):

𝑑𝑇
𝜌𝐶𝑝 𝑑𝑡 = −ℎ𝑎(𝑇 − 𝑇0 )
↓ 𝐴𝑠
3 𝐾 2 4𝜋𝑅 2 𝑚2
(𝑘𝑔⁄𝑚 . 𝐽⁄𝑘𝑔 − 𝑘 . 𝑠 ) = 𝐽⁄𝑠 − 𝑚 − 𝑘 (400 − 𝑇) 𝐾 (4 𝜋𝑅3 ) 𝑚3
⁄3

𝑑𝑇 3 𝑉𝑠
8000 × 400 × 𝑑𝑡 = +3000 × 0.05 (400 − 𝑇)

𝑑𝑇
= +0.05625(400 − 𝑇) = 𝑓(𝑡, 𝑇)
𝑑𝑡

Δ𝑇 = ℎ = 10 or 5 𝑠

𝑘1 = 𝑓(0, 300) = +0.05625(400 − 300) = +5.6250

𝑘2 = 𝑓(5, 300 + 5.625 × 5) = +0.05625(400 − 328.125) = 4.0429

328.125
𝑘3 = 𝑓(5, 300 + 4.0429 × 5) = 0.05625(400 − 320.2145) = 4.4879
320.2145
96
𝑘4 = 𝑓(10, 300 + 4.4879 × 10) = 0.05625(400 − 344.879) = 3.1005
344.8790
𝑘 = 1⁄6 (5.6206 + 2 × 4.0429 + 2 × 4.4879 + 3.1005) = 4.2979

𝑇 = 300 + 4.2979 × 10 = 𝟑𝟒𝟐. 𝟗𝟕𝟗𝟓 𝑲 𝑨𝒏𝒔.

March for the next time steps 𝑡 = 20, 30, 40, 50, 60 𝑠, each time using the calculated values
from the previous time steps (viz. 342.9795, etc). Repeat the entire calculations for Δ𝑡 = 5𝑠.

⇒ In a 20-25 min quiz or examination, such problem can be solved for one time step using
calculator. One requires to write a code to solve such problem for a long time or several steps
calculations.

⇒ Rate at 𝑡 = 0, 0.5 and 1.0 min

𝑑𝑇
| = 0.05625(400 − 𝑇) = 0.05625 × 100 = 5.6250 𝑘⁄𝑠
𝑑𝑡 0

(Note: Once the functional values or temperatures are determined or known, the derivatives can
also be calculated using the formulae you learnt! But, here is the analytical solution.)

Plot the graph as done in the previous lecture and approximately draw the steps for
∗ ∗∗ ∗∗∗
𝑘1 , 𝑘2 , 𝑘3 , 𝑘4 , 𝑘, and functional values (𝑇𝑛 , 𝑇𝑛+ 1⁄ , 𝑇𝑛+1⁄ , 𝑇𝑛+1 , 𝑇𝑛+1 ) on y-axis.
2 2

⇒ From the chemical engineering point of view, temperature-gradient in (steel) sphere was
neglected considering a large thermal conductivity of the material, 𝑘 ≫ ℎ′𝑅′ or small Biot #.
Therefore, the lump body approach was used to determine the sphere temperature, and the
resulting energy balance equation is derived to be a 1st order ODE, else.

Ex2. Consider a steady-state 1D reactive flow of a gaseous species, A in a long quartz tubular
reactor (ID = D, length = L) (Re > 5000) radiated by UV light. The average velocity in the
tube is 𝑉. The species 𝐴 is converted to B by the 1st order homogeneous reaction (𝑟 = 𝑘𝐶𝐴 ),
as it flows in the tube. The inlet concentration of A is 𝐶𝐴𝑜 . Determine the axial concentration
profiles of A in the reactor. The reaction is exothermic: (−Δ𝐻; 𝑐𝑎𝑙⁄𝑚𝑜𝑙 ).

Inlet temperature is 𝑇𝑜

𝑇𝑜 𝐶𝐴 (𝑥) =?
𝑉̅
𝐶𝐴𝑜 (𝑟 = 𝑘𝐶𝐴 )
𝑥=𝐿

97
- There are two commonly used multi-step methods for solving ODEs (Note: RK-4 is a single
step multi-stage method): Adams-Bashforth and Adams-Moulten. These methods use a
polynomial on the derivatives of a function, passing through the single or multi-points.

Adams-Bashforth (Explicit)
𝑑𝑦
= 𝑓(𝑦, 𝑡) ≡ 𝑓(𝑦(𝑡), 𝑡)
𝑑𝑡
𝑡 = 0, 𝑦(0) = 𝑦0
Solve for y(t) = ?
𝑡0 𝑡1 𝑡𝑛 𝑡𝑛+1
𝑦
{ 0 𝑦1 𝑦𝑛 𝑦𝑛+1 : solve
𝑦0′ 𝑦1′ 𝑦𝑛′ ′
𝑦𝑛+1 : gradient
We have,
𝑡
𝑦𝑛+1 = 𝑦𝑛 + ∫𝑡 𝑛+1 𝑦′𝑑𝑡
𝑛

extrapolated value between t n and t n+1

𝑡−𝑡𝑛 1
Define 𝛼 = ℎ
:⇒ 𝑦𝑛+1 = 𝑦𝑛 + ℎ ∫0 𝑦 ′ (𝛼)𝑑𝛼 (0 ≤ 𝛼 ≤ 1) − (1)

Fit a jth order polynomial of 𝑦𝑛′ through 𝑋𝑛 and 𝑋𝑛−𝑗 (𝑗 + 1 pts preceding 𝑋𝑛 ) using backward
difference formula (∇𝑦𝑖 = 𝑦𝑖 − 𝑦𝑖−1 ).

𝛼(𝛼 + 1) 2 𝛼(𝛼 + 1) ⋯ (𝛼 + 𝑗 − 1) 𝑗 ′
𝑦 ′ (𝛼) = [1 + 𝛼∇ + ∇ +⋯ ∇ ] 𝑦𝑛 + 𝑅(𝜉)
2! 𝑗!

(Thus, 𝑦′(𝛼) is the polynomial between (𝑋𝑛 , 𝑋𝑛−1 , 𝑋𝑛−2 ⋯ 𝑋𝑛−𝑗 ), substituting in (1)

1
1 5 3
𝑦𝑛+1 = 𝑦𝑛 + ℎ [1 + ∇ + ∇2 + ∇3 + ⋯ 𝐶𝑗 ∇j ] 𝑦𝑛′ + ℎ ∫ 𝑅(𝜉)𝑑𝑥
2 12 8 0

𝑗 = 0 ∶ you have fitted a constant polynomial of the derivative (𝑦′)

′
𝑦𝑛′ 𝑦𝑛+1 𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑦𝑛′ +
ℎ 2 ′′
𝑦 (𝜉) (𝑦𝑛 < 𝜉 < 𝑦𝑛+1 )
2

𝑛 𝑛+1 (Recognize it as the Euler 1st order formula/explicit)

𝑗 = 1, you have fitted a line (linear polynomial) between 𝑋𝑛 𝑎𝑛𝑑 𝑋𝑛−1

′
′ 𝑦𝑛+1
′ 𝑦𝑛
𝑦𝑛−1
𝑛+1
𝑛−1 𝑛 1 5
𝑦𝑛+1 = 𝑦𝑛 + ℎ [𝑦𝑛′ + ∇𝑦𝑛′ ] + h3 𝑓′′(𝜉)
2 12

99
 ℎ
= 𝑦𝑛 + 2 [3𝑓(𝑦𝑛 ) − 𝑓(𝑦𝑛−1 )] − 2𝑛𝑑 𝑜𝑟𝑑𝑒𝑟 (𝐴𝑑𝑎𝑚 − 𝐵𝑎𝑠ℎ𝑓𝑜𝑟𝑡ℎ)

𝑗 = 2, you have a quadratic polynomial between (X n , X n−1 , X n−2 )

′
𝑦𝑛+1
𝑦𝑛′
′
𝑦𝑛−1
′
𝑦𝑛−2

𝑛−2 𝑛−1 𝑛 𝑛+1

Therefore, by taking different numbers of points preceding 𝑡𝑛 𝑜𝑟 𝑋𝑛 one can get a set of working
formulae having different 0(ℎ𝑛 )

𝑦𝑛+1 = 𝑦𝑛 + ℎ ∑𝑘𝑚=1 𝛽𝑘𝑚 𝑓𝑛+1−𝑚

𝑚=1 2 3 4 ⋯
𝛽1𝑚 1 1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟
2𝛽2𝑚 3 −1 2𝑛𝑑 𝑜𝑟𝑑𝑒𝑟
12𝛽3𝑚 23 −16 5 3𝑟𝑑 𝑜𝑟𝑑𝑒𝑟

Graphically,
𝑗 = 1 (𝑙𝑖𝑛𝑒𝑎𝑟) (𝑞𝑢𝑎𝑑𝑟𝑎𝑡𝑖𝑐)
𝑗=2
𝑗 = 0 (𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡)

𝑦𝑛′ 𝑦𝑛′
⇒

𝑡𝑛−1 𝑡𝑛 𝑡𝑛+1 𝑡𝑛−2 𝑡𝑛−1 𝑡𝑛 𝑡𝑛+1

⇒ From the above description, it is clear that the method has a ‘starting’ problem because only one
condition is known: 𝑦(0) = 𝑦0 . In practice one starts with the Euler’s forward/explicit approach to
calculate 𝑦1 , then use 2nd order method to predict 𝑦2 , and then use (𝑦0 , 𝑦1 , 𝑦2 ) to calculate 𝑦3 . Use of the
preceding ‘3’ data-points gives 3rd order accurate method.

⇒ Similar to the Adams-Bashforth (explicit), there is Adams-Moulton (Implicit) method based on

forward difference interpolation formula. For details, refer the textbooks.

100
 Lecture #19
Instability and stiffness for a system of ODEs
In the previous lectures, we have shown that a system of ODEs (IV𝑃𝑠 ) 𝑦𝑖′ = 𝑓𝑖 (𝑦1 ⋯ 𝑦𝑚 ) 𝑖 =
′
1,2, ⋯ 𝑚 with 𝑦𝑖 (0) = 𝑦0 can be written as the coupled ODEs: 𝑦𝑖 = 𝐴𝑦, where A is the
Jacobian matrix.

It is also clear that for the 𝑚 × 𝑚 matrix A, there are m eigenvalues: 𝐴𝜙𝑘 = 𝜆𝑘 𝜙𝑘 , where 𝜙𝑘 is
the eigenvector corresponding to 𝜆𝑘 eigenvalue. Now, make use of linear transformation: one
can construct a matrix, S whose columns are the eigenvectors of A such that 𝑆 −1 𝐴𝑆 = Λ, where
𝜆1
Λ is the diagonal matrix of the elements which are 𝜆𝑘 of A, as [ ⋱ ]. Then one can
𝜆𝑘
write,

𝑧̅′ (defintion) = 𝑆 −1 𝑦̅ ′ = 𝑆 −1 𝐴𝑦̅ = (𝑆 −1 𝐴𝑆)𝑆 −1 𝑦̅ = Λz̅

or 𝑧𝑘′ = 𝜆k 𝑧𝑘 , 𝑘 = 1,2, ⋯ 𝑛 − cannonical form.

where, 𝑧𝑘′ are uncoupled linear ODEs. Therefore, all (stability) characteristics of single ODE
will also be shown by the system of ODEs.

We have also seen earlier shown that a general solution to the system of ODEs can be written
as 𝐶1 𝑒 𝜆1 𝑡 { } + 𝐶2 𝑒 𝜆2 𝑡 { } +
⋯ , where the terms in the parenthesis are the eigenvectors corresponding to the eignevalues.
In such case, the growth or decay of the function depends on 𝜆𝑚𝑖𝑛 − 𝜆𝑚𝑎𝑥 . A large numerical
value of 𝜆𝑚𝑎𝑥 will cause the function to grow or decay at a faster rate than that corresponding
to

𝜆𝑚𝑖𝑛
𝑦

𝜆𝑚𝑎𝑥 𝑡
101
𝜆𝑚𝑖𝑛 . In other words, small change in ‘t’ will cause a large change/variation in 𝑦(𝜆𝑚𝑎𝑥 ).
Solving such type of system of ODEs is numerically not trivial because a fine grid or step-size
Δ𝑡 𝑜𝑟 ℎ is required to accurately predict change in 𝑦(𝜆𝑚𝑎𝑥 ), whereas a coarse grid can be used
to predict 𝑦(𝜆𝑚𝑖𝑛 ).

- Such equations are called stiff ODEs. In general,

𝜆𝑚𝑎𝑥
Stiff ratio(𝑆𝑅) = > 10 represents a system
𝜆𝑚𝑖𝑛
of stiff ODEs.
- Examples:
𝑦 (1) A common example is the boundary layer problem
in hydrodynamics or heat or mass transport.
Δ𝑥2 Functional value changes sharply within the
boundary layer, from that at the surface to that in
the potential region far from the surface.
x
Therefore, a numerical computation of the
Δ𝑥1 𝑓𝑜𝑟 𝑏𝑜𝑡ℎ 𝑟𝑒𝑔𝑖𝑜𝑛𝑠 functional values in the boundary layer (bottom
region), in principle, requires fine grid size, Δ𝑥1 ,
while that in the potential region (top region)
requires coarse grid size, Δ𝑥2 . However, the sets of
the conservation equations in two regions are
coupled through some interfacial boundary
conditions, and therefore, must be solved
simultaneously using Δ𝑥1 , which is CPU-wise
extensive!
(2) A combination of series and parallel chemical reactions with low and high rate constants:
𝑑𝑦 −100 0
ex: =[ ] 𝑦 ; 𝑦(0) = [2 1]𝑇
𝑑𝑡 2 −1

on coefficient matrix ‘A’: (−100 − 𝜆)(−1 − 𝜆) = 0 ⇒ 𝜆1 = −100, 𝜆2 = −1

−100 0 𝑌1 100 𝑌1
𝜆1 = −100 ∶ [ ]{ } = { } (𝑁𝑜𝑡𝑒: 𝑆𝑅 = 100)
2 −1 𝑌2 𝑌2

1
2𝑋1 = −99𝑋2 ⇒ 𝑌 (1) = {− 2⁄ }
99

−100 0 𝑌1 −1 𝑌1
𝜆2 = −1 ∶ [ ]{ } = { }
2 −1 𝑌2 𝑌2

0
−100𝑌1 = −𝑌1 and 2𝑌1 − 𝑌2 = −𝑌2 ⇒ 𝑌 (2) = { }
1
1 0
𝑌 = 𝐶1 𝑒 −100𝑡 {− 2 + 𝐶2 𝑒 −𝑡 { }
⁄99}
GS:
1

102
Gear’s Technique: Used to solve stiff ODEs, based on multiple steps but using predictor-corrector
approach:

Predictor
𝑦 = 𝛼1 𝑦𝑛 + 𝛼2 𝑦𝑛−1 + ⋯ 𝛼𝑘 𝑦𝑛−(𝑘−1) + ℎ𝛽0 𝑦𝑛′
(𝑒𝑥𝑝𝑙𝑖𝑐𝑖𝑡): 𝑛+1

Corrector ′
: 𝑦 = 𝛼1 𝑦𝑛 + 𝛼2 𝑦𝑛−1 + ⋯ 𝛼𝑘 𝑦𝑛−(𝑘−1) + ℎ𝛽0 𝑦𝑛+1
(𝑖𝑚𝑝𝑙𝑖𝑐𝑖𝑡) 𝑛+1

You should note that this method also has a "starting" problem, similar to the previously discussed
multi-step methods.

Predictor Table

𝑘 𝛼1 𝛼2 𝛼3 ⋯ 𝛼6 𝛽0

1 1 0 0 ⋯ 0 1
2 0 1 0 ⋯ 0 2
3 −3⁄ 6⁄ −1⁄ ⋯ 0 6⁄
2 2 2 2
⋮
⋮
6

Corrector Table

𝑘 𝛼1 𝛼2 𝛼3 ⋯ 𝛼6 𝛽0

1 0 0 ⋯ 1
1 4⁄ −1⁄ 0 2⁄
2 3 3 0 ⋯ 0 3
3 18⁄ −9⁄ 2⁄ ⋯ 0 6⁄
11 11 11 11
⋮
⋮
6

The first row (k = 1) of each table represents j = 0 (no previous data in the interpolating function); the
second row (k = 2) represents j = 1 (one preceding data in the interpolating function or a linear function),
and the third row (k = 3) represents j =2 (two preceding data in the interpolating function or a quadratic
function or parabola), etc.

Quiz II

104
 Lecture #21
(continued…..from the previous lecture)
𝑑2 𝐶𝐴 2𝐷𝑝 𝑑𝐶𝐴
𝑜𝑟 𝐷𝑝 + −𝑘 =0
𝑑𝑟 2 𝑟 𝑑𝑟

𝑑2 𝑦 2 𝑑𝑦 −𝑘
𝑜𝑟 + 𝑟 𝑑𝑟 + 𝐴 = 0 ; 𝐴 = (𝐷 )
𝑑𝑟 2 𝑝

rp
1 𝑁−1 𝑁
(h = )
0 2 𝑖−1 𝑖 𝑖+1 N
𝑑𝐶𝐴
BCs. 𝑟=0 = 0 (symmetric)
𝑑𝑟
𝑑𝐶𝐴
𝑟 = 𝑟𝑝 − 𝐷𝑝 = 𝑘𝑚 (𝐶𝐴 − 𝐶𝑏 )
𝑑𝑟

Discretized eqn:
𝑦𝑖+1 −2𝑦𝑖 +𝑦𝑖−1 2 𝑦𝑖+1 −𝑦𝑖−1
+𝑟 +𝐴 =0 − (1)
ℎ2 𝑖 2ℎ

(𝑟𝑖 = 𝑖ℎ)

Discretized BCs:
𝑦𝑁+1 −𝑦𝑁−1
𝑦1 −𝑦−1
=0 𝑟𝑖 = 𝑟𝑝 − 𝐷𝑝 = 𝑘𝑚 (𝑦𝑁 − 𝐶𝑏 )
2ℎ
𝑟𝑖 = 0 2ℎ (𝑖 = 𝑁)
(𝑖 = 0) 2ℎ 2ℎ𝑘𝑚
𝑦−1 = 𝑦1 𝑦𝑁+1 = (𝑦𝑁−1 − 𝐷 𝑘𝑚 𝑦𝑛 + 𝐶𝑏 )
𝑝 𝐷𝑝

Before you re-arrange the terms, you should realize that there is a clear discontinuity at 𝑟𝑖 =
0 (or 1st node). Therefore, you cannot proceed without removing discontinuity. One way of
1 𝑑𝑦 𝑑2 𝑦 because ∇y = 0 at r = 0
doing this is to approximate 𝑟 𝑑𝑟 | as 𝑑𝑟 2 |
𝑖=0 𝑖=0

𝑦+1 −2𝑦0 +𝑦−1 𝑦1 −2𝑦0 +𝑦−1

or +2 +𝐴=0
ℎ2 ℎ2

𝑑2 𝑦
or (3 𝑑𝑟 2 + 𝐴) = 0

𝑦1 −2𝑦0 +𝑦−1
or 3 + 𝐴 = 0 at 𝑖 = 0
ℎ2

Eq. (1) is applied over the next nodes, i.e., 𝑖 = 1 ⋯ 𝑁 without any discontinuity

110
 Arrange the terms now,
2
𝑦−1 + (𝐴 ℎ ⁄3 − 2) 𝑦0 + 𝑦1 = −𝐴 𝑟=0
1 1 2 1 1
(ℎ2 − 𝑟 ℎ) 𝑦0 − ℎ2 𝑦1 + (ℎ2 + 𝑟 ℎ) 𝑦2 = −𝐴 𝑟 = 1
1 1
1 1 2 1 1
(ℎ2 − 𝑟 ℎ) 𝑦1 − ℎ2 𝑦2 + (ℎ2 + 𝑟 ℎ) 𝑦3 = −𝐴 𝑟 = 2
2 2
⋮
1 1 2 1 1
{(ℎ2 − 𝑟𝑁 ℎ) 𝑦𝑁−1 − ℎ2 𝑦𝑁 + (ℎ2 + 𝑟𝑁 ℎ) 𝑦𝑁+1 = −𝐴 𝑟 = 𝑁

Apply the BCs (𝑖 = 0 and 𝑖 = 𝑁)

2
1𝑠𝑡 𝑟𝑜𝑤: (𝐴 ℎ ⁄3 − 2) 𝑦0 + 2𝑦1 = −𝐴
2 2 2ℎ𝑘𝑚 1 1 2ℎ𝑘𝑚 1 1
𝐿𝑎𝑠𝑡 𝑟𝑜𝑤: 𝑦𝑁−1 − (ℎ2 + × (ℎ2 + 𝑟 ℎ)) 𝑌𝑁 = −𝐴 − (ℎ2 + 𝑟 ℎ) 𝐶𝑏
ℎ2 𝐷𝑝 𝑁 𝐷𝑝 𝑁

From the ‘preparation of tridiagonal matrix’ point of view,

2
𝑎0 = 1, 𝑏0 = (𝐴 ℎ ⁄3 − 2) , 𝑐0 = 1, 𝑑0 = −𝐴

and
1 −2 1 1
𝑎𝑖 = 1⁄ℎ2 − 𝑟 ℎ , 𝑏𝑖 = , 𝑐𝑖 = ℎ2 + ; 𝑑𝑖 = −𝐴 ; 𝑖 = 1 − 𝑁
𝑖 ℎ2 𝑟𝑖 ℎ

However, on the substitution of BCs,

𝑐0 = 𝑐0 + 𝑎0 (1st row @ i = 0)
2ℎ𝑘𝑚
𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁 , 𝑏𝑁 = 𝑏𝑁 − 𝑐𝑁 ,
𝐷𝑝
2ℎ𝑘𝑚 } 𝑙𝑎𝑠𝑡 𝑟𝑜𝑤
𝑑𝑁 = 𝑑𝑁 − 𝑐𝑁 . . 𝐶𝑏
𝐷𝑝

Therefore, 𝐴𝑋̅ = 𝐵. There are (𝑁 + 1) equations to solve (𝑁 + 1) unknown (𝑌0 ⋯ 𝑌𝑁 ) and A is

the tridiagonal matrix.

𝑏0 𝑐0 0
call Tridiagonal (𝑵 + 𝟏, 𝒂, 𝒃, 𝒄, 𝒅, 𝒚) 𝑖 = 0 ⋯ 𝑁 | |
0 𝑎𝑁 𝑏𝑁

Write a programming code to solve the SS profiles of CA (r) for different rp = 0.25,
2
0.5 and 1 mm, k m = 0.01, 0.1, 0.5 m⁄s and Dp = 10−7 , 10−9 , 10−11 m ⁄s.
Plot the profiles and interpret the results.

111
 for increasing
Dp

Recall the course on chemical reaction engineering. Once SS CA(r) is calculated, you can
calculate the (total) effectiveness factor considering both inter- and intraphase diffusion
resistances by computing the total rate of the consumption of the species, A within the catalyst
and that corresponding to the bulk (gas) phase concentration. The computations may require
numerical integration (maybe, Simpson’s 1/3rd method) to calculate the volume-average
quantities and/or numerical differentiation (may be, 2nd order BDS) to calculate the flux at the
surface of the sphere.

Ex. Consider the steady-state 1D reactive flow (Re > 5000) of a gaseous species A in a 1ong
quartz tubular reactor (1D = D, length = L) radiated by UV light. Velocity in the tube is 𝑉.

The species A is converted into B by the 1st order homogeneous reaction (𝑟 = 𝑘𝐶𝐴 ), 𝑎𝑠 it flows
in the tube. The inlet concentration of A is 𝐶𝐴𝑜 . Determine the axial concentration profiles
𝐶𝐴 (𝑋) =?. The reaction takes place under isothermal conditions and Δ𝐻 ≈ 0.

Δ𝑋
CA (𝑋) =? 𝑋
𝑉̅
𝐶𝐴𝑜 (r = kCA )
𝑋
𝐿

𝜕𝐶𝐴
Soln: + 𝑉. ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 + (𝑟𝐴 )
𝜕𝑡

𝑑𝐶 𝑑 𝐶 2
𝑆𝑆 𝑉̅ 𝑑𝑋𝐴 = 𝐷 𝑑𝑋 2𝐴 − 𝑘𝐶𝐴

𝑑𝐶𝐴
BC. 𝑋=0 𝐶𝐴 = 𝐶𝐴𝑜 , 𝑋 = 𝐿 ∇𝐶𝐴 = 0 or = 0 (long tube approximation)
𝑑𝑋

𝑑𝐶𝐴 𝑑2 𝐶𝐴 𝑘
+𝐵 + 𝐶𝐶𝐴 = 0, 𝐵 = − 𝐷⁄ ̅ ; 𝐶 = 𝑉̅ , Δℎ = 𝐿⁄𝑁
𝑑𝑋 𝑑𝑋 2 𝑉

0 1 2 𝑖−1 𝑖 𝑖+1 𝑁−1 𝑁

112
 𝑦𝑖+1 − 𝑦𝑖−1 𝑦𝑖+1 − 2𝑦𝑖 + 𝑦𝑖−1
+𝐵 + 𝐶𝑦𝑖 = 0, 𝑖 = 1, 𝑁
2ℎ ℎ2

𝑦𝑁+1 −𝑦𝑁−1
Discretize BCs 𝑦0 = 𝐶𝐴𝑜 ; = 0 ⇒ 𝑦𝑁+1 = 𝑦𝑁−1
2ℎ

Collect the terms

𝐵 1 2𝐵 𝐵 1
(ℎ2 − 2ℎ) 𝑦𝑖−1 − ( ℎ − 𝐶) 𝑦𝑖 + (ℎ2 + 2ℎ) 𝑦𝑖+1 = 0, 𝑖 = 1, 𝑁

Prepare the tridiagonal matrix:

𝐵 1 2𝐵 𝐵 1
𝑎𝑖 = (ℎ2 − 2ℎ), , 𝑏𝑖 = ( ℎ − 𝐶) , 𝑐𝑖 = (ℎ2 + 2ℎ), 𝑑𝑖 = 0; 𝑖 = 1, 𝑁

On substitution of BCs,

𝑑1 = 𝑑1 − 𝑎1 𝐶𝐴𝑜 (i = 1, 1st row)

𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁 (i = N, last row)

You have 𝐴𝑦̅ = 𝑏̅, where A is the tridiagonal matrix.

𝑦 = 1, 𝑁

call Tridiagonal (𝑵, 𝒂, 𝒃, 𝒄, 𝒅, 𝒚)

Plot (numerical solutions using reasonable values for 𝐷, 𝑉̅ , 𝑘, 𝐶𝐴𝑜 , 𝐿)

𝐶𝐴𝑜

0 𝐿 𝑋

113
 Lecture #20
BVP (Boundary Value Problems)/2nd order ODEs
In this lecture, we will learn how to apply finite difference (direct) methods to solve BVPs. Such
problems are common in 1 D steady-state heat and mass transport in solid or fluid. The general
form of the BVP assumes a 2nd order ODE:

𝑑2 𝑦
= 𝑓(𝑥, 𝑦, 𝑦 ′ ) with two boundary conditions
𝑑𝑥 2

required at two ends of the domain (𝑥 = 0 and 𝑥 = 𝐿 𝑜𝑟 (0,1)).

A general form of BVP can also be written as:

𝑑2 𝑦 𝑑𝑦
𝐴(𝑥) 𝑑𝑥 2 + 𝐵(𝑥) 𝑑𝑥 + 𝐶(𝑥)𝑦 + 𝐷(𝑥) + 𝐸 = 0; (𝐴(𝑥) ≠ 0)

The required boundary conditions at two ends of the domain can assume any form: Drichlet or
Danckwarts or Neumann. That is, either functional value or gradient or mixed condition can be
specified:

𝑦 = 𝑐, or 𝑦 ′ = c, or 𝑎𝑦 ′ + 𝑏𝑦 = 𝑐.

Let us solve a general BVP equation:

𝑑2 𝑦 𝑑𝑦
𝐴(𝑥) 𝑑𝑥 2 + 𝐵(𝑥) 𝑑𝑥 + 𝐶(𝑥)𝑦 + 𝐷(𝑥) + 𝐸 = 0;

0 1 2 𝑖−1 𝑖 𝑖+1 𝑁−1 𝑁

Step1: divide the domain into N equal steps/grids or grids. Therefore, there are (N+1) nodes.

Step 2: discretize the equation or each terms of the equation at ‘ith’ grid. In this course, we will
discretize the terms using 2nd order accurate method (recall numerical differentiation).

𝑦𝑖−1 − 2𝑦𝑖 + 𝑦𝑖+1 𝑦𝑖+1 − 𝑦𝑖−1

𝐴(𝑥𝑖 ) 2
+ 𝐵(𝑥𝑖 ) + 𝐶(𝑥𝑖 )𝑦𝑖 + 𝐷(𝑥𝑖 ) + 𝐸 = 0
ℎ 2ℎ
0(ℎ2 ) 0(ℎ2 )
Step3: Arrange the terms:

𝐴(𝑥𝑖 ) 𝐵(𝑥𝑖 ) 2𝐴(𝑥𝑖 ) 𝐴(𝑥𝑖 ) 𝐵(𝑥𝑖 )

( − ) 𝑦𝑖−1 − ( − 𝐶(𝑥𝑖 )) 𝑦𝑖 + ( + ) 𝑦𝑖+1 = −𝐷(𝑥𝑖 ) − 𝐸
ℎ2 2ℎ ℎ2 ℎ2 2ℎ

or 𝑎𝑖 𝑦𝑖−1 + 𝑏𝑖 𝑦𝑖 + 𝑐𝑖 𝑦𝑖+1 = 𝑑𝑖 ← discretized form of equation

105
In principle, this equation should be valid for 𝑖 = 0 ⋯ 𝑁. However, there may be problem (!) at
the boundary nodes 𝑖 = 0 and 𝑁, because 𝑦−1 and 𝑦𝑁+1 may not be known or defined or
calculated. You should carefully inspect the given boundary conditions and decide to apply the
discretized equation over (𝑖 = 0 ⋯ 𝑁) 𝑜𝑟 (𝑖 = 1 ⋯ 𝑁)𝑜𝑟 (𝑖 = 1 ⋯ 𝑁 − 1)𝑜𝑟 (𝑖 = 0 ⋯ 𝑁 − 1) so
that the number of equations is the same as that many unknowns (𝑦𝑖 ). For the moment, let us
take the simplest boundary conditions, 𝑦(0) = 𝑦0 and 𝑦(𝑁) = 𝑦𝑁 .

Apply the discretized equation over 𝑖 = 1 ⋯ 𝑁 − 1 as

𝑖 = 1: 𝑎1 𝑦0 + 𝑏1 𝑦1 + 𝑐1 𝑦2 = 𝑑1
𝑎2 𝑦1 + 𝑏2 𝑦2 + 𝑐2 𝑦3 = 𝑑2
⋮
𝑎𝑁−2 𝑦𝑁−3 + 𝑏𝑁−2 𝑦𝑁−2 + 𝑐𝑁−2 𝑦𝑁−1 = 𝑑𝑁−2
𝑁 − 1: 𝑎𝑁−1 𝑦𝑁−2 + 𝑏𝑁−1 𝑦𝑁−1 + 𝑐𝑁−1 𝑦𝑁 = 𝑑𝑁−1 }

Step 4: Apply the BCs and bring the terms containing y0 and yN to RHS:
𝑏1 𝑦1 + 𝑐1 𝑦2 = (𝑑1 − 𝑎1 𝑦0 ) = 𝑑1′
𝑎2 𝑦1 + 𝑏2 𝑦2 + 𝑐2 𝑦3 = 𝑑2
⋮
𝑎𝑁−2 𝑦𝑁−3 + 𝑏𝑁−2 𝑦𝑁−2 + 𝑐𝑁−2 𝑌𝑁−1 = 𝑑𝑁−2
′
𝑎𝑁−1 𝑦𝑁−2 + 𝑏𝑁−1 𝑦𝑁−1 = (𝑑𝑁−1 − 𝑐𝑁−1 𝑌𝑁 ) = 𝑑𝑁−1 }

There are exactly (N-1) unknown and (N-1) equations. Notably, set of equations can be written as
𝑏1 𝑐1 0
𝑎 𝑏2 𝑐2 0 ⋯
𝐴𝑦 = 𝑏 where 𝐴 = 2
0 0 𝑎𝑁−2 𝑏𝑁−2 𝑐𝑁−2
[ 0 𝑎𝑁−1 𝑏𝑁−1 ]
is the tridiagonal matrix.

Step 5: Call Thomas Algorithm (learnt earlier) to invert such matrix:

Tridiagonal (𝑁 − 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑌)
𝑜𝑢𝑡𝑝𝑢𝑡 (𝑖 = 1, ⋯ 𝑁 − 1)
# 𝑜𝑓 𝑒𝑞 𝑛𝑠
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠
𝑜𝑓 𝑚𝑎𝑡𝑟𝑖𝑥 𝐴
You should be able to clearly recognize that while preparing to call tridiagonal matrix, the
following coefficients were changed:
𝑑′1 = 𝑑1 − 𝑎1 𝑌0 ∶ 1st row (𝑖 = 1)

The other coefficients remained the same: (i = 2, N-2)

𝐴(𝑥𝑖 ) 𝐵(𝑥𝑖 ) 2𝐴(𝑥𝑖 )
𝑎𝑖 = 2 − ; 𝑏𝑖 = 𝐶(𝑥𝑖 ) −
ℎ 2ℎ ℎ2

106
 𝐴(𝑥𝑖 ) 𝐵(𝑥𝑖 )
𝑐𝑖 = + ; 𝑑𝑖 = −𝐷(𝑥𝑖 ) − 𝐸 ;
ℎ2 2ℎ

Also, dN−1 was changed as d′N−1 = dN−1 − cN−1 YN ∶ last row (i = N − 1)

⇒ Let us consider another example where the BVP equation is the same. At boundary, 𝑦(0) =
𝑦0 (also same as before). However, at the other boundary, gradient is specified instead of functional
value: 𝑦𝑁′ = 0.
𝑌𝑁+1 −𝑌𝑁−1
The boundary condition 𝑦𝑁′ can be discretized as 2ℎ
= 0 (0(ℎ2 )) 𝑜𝑟 𝑌𝑁+1 = 𝑌𝑁−1

Considering that YN is NOT known in this case, the coefficients (𝑎𝑖 , 𝑏𝑖 , 𝑐𝑖 , 𝑑𝑖 ) are extended to the rows:
𝑖 = 1 … . . 𝑁, 𝑖. 𝑒., the matrix A will have ‘N’ rows and the 1st & last row will be modified as

𝑑1 = 𝑑1 − 𝑎1 𝑌0 (same as before)
and 𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁 ; why ?

because the Nth equation

aN YN−1 + bN YN + cN YN+1 = dN is modified as

aN YN−1 + bN YN + cN YN−1 = dN on substitution of the bc.

Finally, call Tridiagonal (𝑁, 𝑎, 𝑏, 𝑐, 𝑑, 𝑌); i = 1, N

⇒ It is recommended that you should do your best to keep consistency between the order of methods
used to discretize the equation and boundary conditions. Also note that, 2nd order discretization method
is reasonably good/accurate for engineering calculations. If you use a higher order method, the
coefficient matrix A will no more be tridiagonal that is easy and fast to invert.

Ex: Consider the steady-state one dimensional heat conduction in a fine (length L, cross-section A,
perimeter P), whose one end is at a constant temperature 𝑇𝑓 and the other end is insulated. The ambient
temperature is 𝑇𝑎 and the convective heat transfer coefficient at the fix surface is ℎ.

Solve the steady-state temperature profiles in the fin.

ℎ𝑓
Δ𝑥 𝑇𝑎

𝑇𝑓 𝑇(𝑥) =? 𝐴
𝑥 𝑃
L

An energy balance over ′𝐴Δ𝑥′ control volume will result in the following differential equation:
∂T cal⁄
ρCp ( ∂t + V. ∇T) = k∇2 T + S ∶
s − m3 𝑚2
SS Solid −h(T − Ta ) (PΔx)
cal⁄ .
(A Δx)
s − m2
𝑚3 107
Note that there is no source or sink of heat as such in the fin. However, considering 1D axial heat
transfer, heat dissipated from the fin surface is homogenously distributed within the 𝐶𝑉: 𝑆 =
−ℎ(𝑇 − 𝑇𝑎 )𝑎𝑠 where 𝑎𝑠 is the surface area per unit volume of the fin.

𝑑2𝑇 ℎ(𝑇−𝑇𝑎 )𝑃
⇒𝑘 − =0
𝑑𝑥 2 𝐴

d2 T −hP hP
or + BT + C = 0 where B = ; C= T and choose Δh = L⁄10
dx2 kA kA a

𝑑𝑇
BC: 𝑇(𝑥 = 0) = 𝑇𝑓 ; −𝑘∇𝑇 (𝑥 = 𝐿) = 0 𝑜𝑟 𝑑𝑥 = 0
(𝑖𝑛𝑠𝑢𝑙𝑎𝑡𝑖𝑜𝑛)
Discretize the equation using the 2nd order method.

𝑑2 𝑇
Soln: ( + 𝐵𝑇 + 𝐶 = 0)
𝑑𝑥 2

0 1 𝑖−1 𝑖 𝑖+1𝑁−1 𝑁
𝑇𝑖−1 −2𝑇𝑖 +𝑇𝑖+1
ℎ2
+ 𝐵𝑇𝑖 + 𝐶 = 0

𝑇𝑖−1 + (𝐵ℎ2 − 2)𝑇𝑖 + 𝑇𝑖+1 = −𝐶ℎ2 𝑖 = 1, 𝑁

Prepare the tridiagonal matrix

𝑎𝑖 = 1, 𝑏𝑖 = (𝐵ℎ2 − 2), 𝑐𝑖 = 1, 𝑑𝑖 = −𝐶ℎ2 , 𝑖 = 1, 𝑁
Apply BC/discretize BC:
𝑇0 = 𝑇𝑓 (𝑖 = 0), 𝑇𝑁+1 = 𝑇𝑁−1 (𝑖 = 𝑁)

Therefore, 1st & last row will be modified as

𝑑1 = 𝑑1 − 𝑎1 𝑇𝑓
𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁
How? 1st row is modified as (𝐵ℎ2 − 2)𝑇1 + 𝑇2 = −𝐶ℎ2 − 𝑇𝑓
Last row is modified as 𝑇𝑁−1 + (𝐵ℎ2 − 2)𝑇𝑁 + 𝑇𝑁−1 = −𝐶ℎ2
Therefore, you have the tridiagonal matrix A. Call the subroutine as:

Tridiagonal (𝑁, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦) (Note that the total number of equations is N)

As an example, take the fin-problem of heat transfer: L = 1 cm, width of the fin = 0.1 cm, 𝑇𝑓 =
1000 𝐶, 𝑇𝑎 = 300 𝐶, ℎ = 100 𝑤⁄𝑚2 𝐾, 𝑘 = 20 𝑊 ⁄𝑚 − 𝐾. Take Δℎ = 0.1 𝑐𝑚, write a programming
code to obtain the solution and plot the temperature profiles:
100𝑜 𝐶

𝑇
30𝑜 𝐶 𝑥 𝐿 = 1 𝑐𝑚

108
 𝜕𝐶𝐴
𝑋 = 𝐿, ∇𝐶𝐴 = 0 (long − tube approximation) 𝑜𝑟 =0
𝜕𝑋

𝜕𝐶𝐴 𝜕𝐶𝐴 𝜕2 𝐶𝐴
+ 𝑉𝑋 =𝐷 ; 𝑉𝑋 = 𝑉̅
𝜕𝑡 𝜕𝑋 𝜕𝑋 2

𝑖=0 𝑁 𝑡+1 ′Δ𝑡′

𝑖=0 𝑖−1 𝑖 𝑖+1 𝑁 𝑡
ℎ

𝐶𝐴𝑡+1
𝑖
− 𝐶𝐴𝑡𝑖 1 𝐶𝐴,𝑖+1 − 2𝐶𝐴,𝑖 + 𝐶𝐴,𝑖−1 𝑡+1 𝐶𝐴,𝑖+1 − 2𝐶𝐴,𝑖 + 𝐶𝐴,𝑖−1 𝑡
= [𝐷 ( ) +𝐷( )
Δ𝑡 2 ℎ2 ℎ2
𝐶𝐴,𝑖+1 − 𝐶𝐴,𝑖−1 𝑡+1 𝐶𝐴,𝑖+1 − 𝐶𝐴,𝑖−1 𝑡
̅
−𝑉( ) ̅
−𝑉( )]
2ℎ 2ℎ

𝐷 𝑉̅ 1 𝐷 𝐷 𝑉̅
( 2 + ) 𝐶𝐴,𝑖−1 𝑡+1 − ( + 2 ) 𝐶𝐴,𝑖 𝑡+1 + ( 2 − ) 𝐶𝐴,𝑖+1 𝑡+1
2ℎ 4ℎ Δ𝑡 ℎ 2ℎ 4ℎ
𝐷 ̅
𝑉 1 𝐷 𝐷 𝑉̅
= − ( 2 + ) 𝐶𝐴,𝑖−1 𝑡 − ( − 2 ) 𝐶𝐴,𝑖 𝑡 − ( 2 − ) 𝐶𝐴,𝑖+1 𝑡
2ℎ 4ℎ Δ𝑡 ℎ 2ℎ 4ℎ

𝑖 = 1, 𝑁
𝑡 𝑡+1
𝑖 = 0: 𝐶𝐴,0 = 𝐶𝐴,0 = 𝐶𝐴,𝑖𝑛 (Inlet BC)

𝑡+1 𝑡+1 𝑡 𝑡
𝑖 = 𝑁: 𝐶𝐴,𝑁+1 = 𝐶𝐴,𝑁−1 and 𝐶𝐴,𝑁+1 = 𝐶𝐴,𝑁−1

* Prepare the tridiagonal matrix

𝐷 𝑉̅ 𝐷 1 𝐷 𝑉
𝑎𝑖 = ( 2 + ), 𝑏𝑖 = − ( 2 + ) , 𝑐𝑖 = ( 2 − ) , 𝑑𝑖
2ℎ 4ℎ ℎ Δ𝑡 2ℎ 4ℎ
𝐷 𝑉 𝑡 1 𝐷 𝑡 𝐷 𝑉̅ 𝑡
= − ( 2 + ) 𝐶𝐴,𝑖−1 − ( − 2 ) 𝐶𝐴,𝑖 − ( 2 − ) 𝐶𝐴,𝑖+1
2ℎ 4ℎ Δ𝑡 ℎ 2ℎ 4ℎ
Apply BCs:

𝑑1 = 𝑑1 − 𝑎1 𝐶𝐴,𝑖𝑛 : Note that 𝐶𝐴,𝑖𝑛 is substituted for 𝐶𝐴,𝑖−1 𝑡 in 𝑑1 .

𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁
1 𝐷 𝑡 𝑡𝐷
𝑑𝑁 = − (Δ𝑡 − ℎ2 ) 𝐶𝐴,𝑁 − (ℎ2 ) 𝐶𝐴,𝑁−1

You have 𝐴𝑦 (𝑡+1) = 𝑑 (𝑡) ; A ≡ Tridiagonal matrix with i = 1, N rows

118
Call Tridiagonal (N, a, b, c, d, y)

Start with 𝑡 = 0: 𝐶𝐴 (𝑖 = 0, 𝑁) = 0 (pure solvent) or use a small concentration, 𝐶𝐴𝑖 ≪ 𝐶𝐴,𝑖𝑛𝑙𝑒𝑡 .

(1)
Solve for 𝑦 (1) 𝑜𝑟 𝐶𝐴 𝑎𝑡 𝑡 = 1, recalculate RHS 𝑑 (2) and keep on marching on time till a
steady-state solution or at 𝑡 = 𝑡𝑓𝑖𝑛𝑎𝑙 is reached.

̅ = 0.1 cm⁄s , D = 0.01 cm2 ⁄s)

Plot (use L = 10 cm, CA,i = 0.001, CA,inlet = 1.0, V

1.0 𝑡

𝜕𝐶
}⇒ = 0|
𝜕𝑋 𝛼=𝐿

0.001

0 𝑥 𝐿

Note that

 Tridiagonal matrix is called at every time-step.

 ̅ vector is updated with the recent most 𝐶 values.
𝑏 𝐴
 gradient at the exit must be flat (consistent with BC)
 There are two steps (ℎ & Δ𝑡). How do you choose them?
Δ𝑋 Δ𝑋 2
 No one knows! In general Δ𝑡 < , . Why? ?
𝑉 𝐷

One should choose a fixed Δ𝑋, and then refine ′Δ𝑡′ till there is a convergence in the solution.
Again, use Δ𝑋 = Δ𝑋⁄2, adjust ′Δ𝑡′ till the solution converges, and so forth!

Ex. Repeat the previous problem, assuming that some solute is irreversibly adsorbed at the
tube-wall @ 𝑘𝐶𝐴 rate.

r = -kCA
𝐶𝐴,𝑖𝑛 X
𝑉̅ 𝐶𝐴 (𝑡, 𝑋) =?
Δ𝑋 L
Species balance equation:
𝜕𝐶𝐴
+ 𝑉. ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 − (𝑘𝐶𝐴 )𝑎, where a is the specific surface area per unit
𝜕𝑡
volume of the tube.

𝜕𝐶𝐴 𝜕𝐶 𝜕 𝐶 2
or 𝜕𝑡
+ 𝑉̅ 𝜕𝑋𝐴 = 𝐷 𝜕𝑋 2𝐴 − 𝑘′𝐶𝐴 ;

119
 𝐼𝐶: 𝑡 = 0 𝐶𝐴 = 𝐶𝐴,𝑖 ,
𝜕𝐶𝐴
𝐵𝐶 0+ , 𝑋 = 0; 𝐶𝐴 = 𝐶𝐴,𝑖𝑛 , 𝑋 = 𝐿, ∇𝐶𝐴 = 0 ⇒ =0
𝜕𝑋

0 𝑁 𝑡+1
𝑖−1 𝑖+1
𝑖=0 1 𝑖−1 𝑖 𝑖+1 𝑁−1 𝑡
𝑁
𝐶𝐴𝑡+1 −𝐶𝐴𝑡 1 𝐶𝐴,𝑖+1 −2𝐶𝐴,𝑖 +𝐶𝐴,𝑖−1 𝑡+1 𝐶𝐴,𝑖+1 −2𝐶𝐴,𝑖 +𝐶𝐴,𝑖−1 𝑡 𝐶 −𝐶 𝑡+1
𝑖 𝑖
= 2 [𝐷 ( ) +𝐷( ) − 𝑉̅ ( 𝐴,𝑖+12ℎ 𝐴,𝑖−1 ) −
Δ𝑡 ℎ2 ℎ2
𝐶 −𝐶 𝑡
𝑉̅ ( 𝐴,𝑖+12ℎ 𝐴,𝑖−1 ) − (𝑘′𝐶𝐴,𝑖
𝑡+1 𝑡
+ 𝑘′𝐶𝐴,𝑖 )] Note: This is an extra term.

Re-arrange,

𝐷 𝑉̅ 𝑡+1 1 𝐷 𝑘′ 𝑡+1 𝐷 𝑉̅ 𝑡+1

( 2 + ) 𝐶𝐴,𝑖−1 − ( + 2 + ) 𝐶𝐴,𝑖 + ( 2 − ) 𝐶𝐴,𝑖+1
2ℎ 4ℎ Δ𝑡 ℎ 2 2ℎ 4ℎ
𝐷 𝑉̅ 𝑡 1 𝐷 𝑘′ 𝑡 𝐷 𝑉̅ 𝑡
= − ( 2 + ) 𝐶𝐴,𝑖−1 − ( − 2 − ) 𝐶𝐴,𝑖 − ( 2 − ) 𝐶𝐴,𝑖+1
2ℎ 4ℎ Δ𝑡 ℎ 2 2ℎ 4ℎ

𝑖 = 1, 𝑁
𝑡 𝑡+1
𝑖 = 0, 𝐶𝐴,𝑜 = 𝐶𝐴,𝑜 = 𝐶𝐴,𝑖𝑛 (for all time steps)
𝐵𝐶𝑠: { 𝑡+1 𝑡+1 𝑡 𝑡
𝑖 = 𝑁, 𝐶𝐴,𝑁+1 = 𝐶𝐴,𝑁−1 𝑎𝑛𝑑 𝐶𝐴,𝑁+1 = 𝐶𝐴,𝑁−1

Prepare tridiagonal matrix:

𝐷 𝑉̅ 1 𝐷 𝑘′ 𝐷 𝑉̅
𝑎𝑖 = ( 2 + ), 𝑏𝑖 = − ( + 2 + ) , 𝑐𝑖 = ( 2 − ) , 𝑑𝑖
2ℎ 4ℎ Δ𝑡 ℎ 2 2ℎ 4ℎ
𝐷 𝑉̅ 𝑡 1 𝐷 𝑘′ 𝑡 𝐷 𝑉̅ 𝑡
= − ( 2 + ) 𝐶𝐴,𝑖−1 − ( − 2 − ) 𝐶𝐴,𝑖 − ( 2 − ) 𝐶𝐴,𝑖+1
2ℎ 4ℎ Δ𝑡 ℎ 2 2ℎ 4ℎ

Apply BCs.
𝐷 ̅
𝑉 1
𝑡 𝐷 𝑘′ 𝑡 𝐷 ̅
𝑉
𝑑1 = − (ℎ2 + 2ℎ) 𝐶𝐴,𝑖𝑛 − (Δ𝑡 − ℎ2 − 2 ) 𝐶𝐴,1 − (2ℎ2 − 4ℎ) 𝐶𝐴,2 : i =1 (1st row)

(Note that this step is the same as 𝑑1 = 𝑑1 − 𝑎1 𝐶𝐴,𝑖𝑛 )

𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁
𝑡 𝐷 𝑡1 𝐷 𝑘′ } 𝑖=𝑁 (𝑙𝑎𝑠𝑡 𝑟𝑜𝑤)
𝑑𝑁 = − ℎ2 𝐶𝐴,𝑁−1 − (Δ𝑡 − ℎ2 − 2 ) 𝐶𝐴,𝑁

𝐴𝑦 (𝑡+1) = 𝑑(𝑡) ; A is the tridiagonal matrix.

120
Proceed as before with initial condition CA (i = 1, N) = CA,i . Evaluate d(t) , solve for
y (t+1) , evaluate d(t+1) , solve for y (t+2) ⋯ etc. Subroutine Tridiagonal (N, a, b, c, d, y) is called
at every time step. Plot qualitatively for the same conditions as before, use k = 0.1 s−1 .

The axial concentration profiles for different ′t s ′ will be the similar as before; however increase
in concentration with ‘t’ will be less than before because of the reactive/adsorptive wall.

𝐶𝐴,𝑖𝑛
𝑡2

𝐶𝐴 (non − reactive wall)

𝑡2 𝑡1
𝑡1 (reactive wall)

121
 Lecture #23
Example: Consider a spherical (𝑑𝑝 = 0.1 cm) steel pellet (𝑘 = 40 W⁄m − K , ρ =
8000 kg⁄m3 , Cp = 400 J⁄kg − K. The initial temperature of the pellet is 300 𝐾. It is
immersed in a large oil tank at 400 K. The corrective heat transfer coefficient, ℎ𝑓 at the sphere
surface is 3000 𝑊 ⁄𝑚2 𝐾. Solve for unsteady- state temperature profiles in the sphere. Choose
Δ𝑡 = 1 𝑠. Determine the average temperature in the sphere and the rate of which the surface
temperature decreases at 𝑡 = 20 𝑠.

* Recall: There is a discontinuity at 𝑖 = 0 (𝑟 = 𝑖ℎ)

1 𝜕𝑇 𝜕2 𝑇
However, the grad ∇𝑇 = 0 at 𝑟 = 0. Therefore, 𝑟 𝜕𝑟 ≈ 𝜕𝑟 2

𝑑𝑞
𝑡
400 (𝑇𝑓 )
𝑟

hf Δ𝑟
𝑡=0 300 𝐾
rp

Energy balance equation:

𝜕𝑇
𝜌𝐶𝑝 𝜕𝑡 = 𝑘∇2 𝑇 + 𝑆 (𝛼 = 𝑘⁄𝜌𝐶 )
𝑝

𝜕𝑇 𝛼 𝜕 𝜕𝑇
= (𝑟 2 ) 𝑟𝑝 > 𝑟 ≥ 0
𝜕𝑡 𝑟 2 𝜕𝑟 𝜕𝑟

𝐼𝐶: 𝑡 = 0 𝑇 = 300 𝐾(𝑇𝑜 ) for all 𝑟𝑝 ≥ 𝑟 ≥ 0

𝜕𝑇
𝐵𝐶: 0+ 𝑟=0 ∇𝑇 = 0 𝑜𝑟 =0
𝜕𝑟

𝜕𝑇
𝑟 = 𝑟𝑝 − 𝑘 𝜕𝑟 = ℎ(𝑇 − 𝑇𝑓 )

𝜕𝑇
or 𝑘 𝜕𝑟 = ℎ(𝑇𝑓 − 𝑇)

122
 𝜕𝑇 𝜕2 𝑇 2𝛼 𝜕𝑇
The equation is re-written as = 𝛼 𝜕𝑟 2 +
𝜕𝑡 𝑟 𝜕𝑟

𝑡+1
0 1 𝑖−1 𝑖 𝑖+1 𝑁−1 𝑁
0 1 𝑖−1 𝑖 𝑖+1 𝑁 (𝑟𝑖 = 𝑖ℎ)
𝑁−1

Apply Crank-Nicholson discretization scheme:

𝑇𝑖𝑡+1 − 𝑇𝑖𝑡 1 𝑇𝑖+1 − 2𝑇𝑖 + 𝑇𝑖−1 𝑡+1 𝑇𝑖+1 − 2𝑇𝑖 + 𝑇𝑖−1 𝑡 2𝛼 𝑇𝑖+1 − 𝑇𝑖−1 𝑡+1
= [𝛼 ( ) +𝛼( ) + ( )
Δ𝑡 2 ℎ2 ℎ2 𝑟𝑖 2ℎ
2𝛼 𝑇𝑖+1 − 𝑇𝑖−1 𝑡
+ ( )]
𝑟𝑖 2ℎ
𝛼 𝑡+1 𝛼 1 𝛼 𝛼 𝛼 𝛼 𝛼
Arrange: (2ℎ2 − 2𝑟 ℎ) 𝑇𝑖−1 − (Δ𝑡 + ℎ2 ) 𝑇𝑖𝑡+1 + (2ℎ2 + 2𝑟 ℎ) 𝑇𝑖+1
𝑡+1 𝑡
= (2𝑟 ℎ − 2ℎ2 ) 𝑇𝑖−1 +
𝑖 𝑖 𝑖
𝛼 1 𝛼 𝛼
(ℎ2 − Δ𝑡) 𝑇𝑖𝑡 − (2ℎ2 − 2𝑟 ℎ) 𝑇𝑖+1
𝑡
𝑖 = 1, 𝑁
𝑖

𝜕𝑇 𝜕2 𝑇
Note 𝑖 = 0 requires a modified equation as (why?): = 3𝛼 𝜕𝑟 2
𝜕𝑡

The modified equation is discretized:

𝑇0𝑡+1 −𝑇0𝑡 3𝛼 𝑇1 −2𝑇0 +𝑇−1 𝑡+1 𝑇1 −2𝑇0 +𝑇−1 𝑡

= [( ) +( ) ]: i = 0
Δ𝑡 2 ℎ2 ℎ2

3𝛼 𝑡+1 3𝛼 1 3𝛼 3𝛼 3𝛼 1 3𝛼
(2ℎ2 ) 𝑇−1 − (ℎ2 + Δ𝑡) 𝑇0𝑡+1 + (2ℎ2 ) 𝑇1𝑡+1 = − (2ℎ2 ) 𝑇−1
𝑡
+ ( ℎ2 − Δ𝑡) 𝑇0𝑡 − (2ℎ2 ) 𝑇1𝑡 ; i =0

Prepare tridiagonal matrix:

3𝛼 3𝛼 1 3𝛼
2
𝑎0 =
, 𝑏0 = − ( 2 + ) , 𝑐0 = 2
2ℎ ℎ Δ𝑡 2ℎ } 𝑖 = 0
3𝛼 𝑡
3𝛼 1 3𝛼
𝑑0 = − ( 2 ) 𝑇−1 + ( 2 − ) 𝑇0𝑡 − ( 2 ) 𝑇1𝑡
2ℎ ℎ Δ𝑡 2ℎ
𝛼 𝛼 𝛼 1 𝛼 𝛼
𝑎𝑖 = ( 2 − ), 𝑏𝑖 = − ( 2 + ) , 𝑐𝑖 = ( 2 + ) , 𝑑𝑖
2ℎ 2𝑟𝑖 ℎ ℎ Δ𝑡 2ℎ 2𝑟𝑖 ℎ
𝛼 𝛼 𝑡 𝛼 1 𝛼 𝛼
=( − 2 ) 𝑇𝑖−1 + ( 2 − ) 𝑇𝑖𝑡 − ( 2 + 𝑡
) 𝑇𝑖+1
2𝑟𝑖 ℎ 2ℎ ℎ Δ𝑡 2ℎ 2𝑟𝑖 ℎ
𝑖 = 1, 𝑁
𝜕𝑇
Discretize BCs (1) 𝑟 = 0 = 0 ⇒ 𝑖 = 0; 𝑇−1 = 𝑇+1
𝜕𝑟

𝜕𝑇
(2) 𝑟 = 𝑟𝑝 𝑘 𝜕𝑟 = ℎ𝑓 (𝑇𝑓 − 𝑇) 𝑖 = 𝑁;

123
 𝑇𝑁+1 −𝑇𝑁−1 ℎ𝑓
= (𝑇𝑓 − 𝑇𝑁 )
2h 𝑘

2ℎ𝑓 ℎ
or, 𝑇𝑁+1 = 𝑇𝑁−1 + (𝑇𝑓 − 𝑇𝑁 )
𝐾

Apply BC

𝑐0 = 𝑐0 + 𝑎0
3𝛼 1 3𝛼 } 𝑖=0
𝑑0 = ( ℎ2 − Δ𝑡) 𝑇0𝑡 − ℎ2 𝑇1𝑡

𝑎𝑁 = 𝑎𝑁 + 𝑐𝑁

𝑡 𝛼 𝛼 1 𝛼 𝛼 2ℎ𝑓 ℎ 𝛼 𝛼 2ℎ𝑓 ℎ
𝑑𝑁 = − ℎ2 𝑇𝑁−1 + (ℎ2 − Δ𝑡) 𝑇𝑁𝑡 + (2ℎ2 − 2𝑟 ℎ) 𝑇𝑁𝑡 − (2ℎ2 + 2𝑟 ℎ) 𝑇𝑓
𝑖 𝑘 𝑖 𝑘

Call Tridiagonal(N+1, a,b,c,d,y)

Solve 𝐴𝑦 (𝑡+1) = 𝑑(𝑡) , 𝑖 = 0, 𝑁 with initial condition: 𝑇 = 𝑇𝑖𝑛𝑖 = 300

𝐴𝑦 (1) = 𝑑(0) ⇒ 𝐴𝑦 (2) = 𝑑 (1) ⇒ 𝐴𝑦 (3) = 𝑑 (2) , 𝑒𝑡𝑐.

Once the functional values or y(t,r) or yit, i = 0, N are computed, the average temperature of
the sphere at any time, t can be calculated as the volume-average quantity (why?), as follows:

3 ∑yiri2∆r/R3, i = 0, N and ri = ih (assuming the constant or average thermophysical properties

i,.e., ρ, Cp of the material). You can use the Simpson’s 1/3rd or Trapezoidal rule of integration.

𝜕𝑇
Similarly, the rate of decrease of the surface temperature, viz. 𝜕𝑡 @ 𝑟 = 𝑅 at any
𝑦(𝑡𝑖+1 )−𝑦(𝑡𝑖−1 ) 4𝑦(𝑡𝑖+1 )−3𝑦(𝑡𝑖 )−𝑦(𝑡𝑖+2 )
time, t can be calculated using CDS as or BDS as or FDS
2∆𝑡 2∆𝑡
−4𝑦(𝑡𝑖−1 )+3𝑦(𝑡𝑖 )+𝑦(𝑡𝑖−2 )
as , depending upon the time, t. Therefore, you must use FDS at the
2∆𝑡
initial time, t = 0, etc.

124
Example: Consider the catalytic oxidation of 𝑆𝑂2 into 𝑆𝑂3 over a spherical Pt-dispersed porous
carbon catalyst (radius = 𝑟𝑝 ). The atmospheric concentration of 𝑆𝑂2 is 1%. The rate of
reaction is 𝑟 = 𝑘 (= 0.1) 𝐶𝐴 mole⁄s − m3 . The film mass transfer coefficient is 𝑘𝑚 (𝑚⁄𝑠).
Determine the time-development of concentration profiles within the catalyst. The pore
diffusion coefficient of 𝑆𝑂2 is estimated to be 𝐷 𝑚2 ⁄𝑠.

𝐶𝑏 = 0.01
𝐶(𝑡, 𝑟) =?
𝑆𝑆

𝑘𝑚

𝑟𝑝

𝜕𝐶𝐴 𝐷 𝜕 𝜕𝐶𝐴
= 𝑟 2 𝜕𝑟 (𝑟 2 ) − 𝑘𝐶𝐴 (1)
𝜕𝑡 𝜕𝑟

𝑡 = 0 𝐶𝐴 = 𝐶𝐴𝑖 ≪< 𝐶𝑏 for all 𝑟𝑝 ≥ 𝑟 ≥ 0

0+ ∇𝐶𝐴 = 0 @ 𝑟 = 0
𝜕𝐶𝐴
−𝐷 = 𝑘𝑚 (𝐶𝐴 − 𝐶𝑏 ) @ 𝑟 = 𝑟𝑝
𝜕𝑟

Before you solve, two clarifications follow:

(1) The steady-state solution of eq(1) is the same as that of the equation solved in the
previous lectures for BVP. Therefore, one way of checking the present code for this
unsteady-state 1D parabolic equation is to compare its SS solution with that from the
BVP-code. In the latter lectures you will see that very often it is better to artificially
𝜕𝑇 𝜕𝐶 𝜕𝑉
introduce the transient term (𝑣𝑖𝑧. 𝜕𝑡 , 𝜕𝑡 , 𝜕𝑡 ) in the BVPs and solve the entire time-
profiles, even when one is asked to solve the SS solution only. This strategy is often
followed for solving elliptic PDE.
(2) You should also compare this mass transport example with the previous example on heat
transfer. But for the non-homogenous term (𝑘𝐶𝐴 ), two equations are identically the same,
including the IC and BCs.

Recall:

125
 𝜕𝑇 𝛼 𝜕 𝜕𝑇 𝜕𝐶𝐴 𝐷 𝜕 𝜕𝐶𝐴
= 𝑟 2 𝜕𝑟 (𝑟 2 𝜕𝑟 ) = 𝑟 2 𝜕𝑟 (𝑟 2 )
𝜕𝑡 𝜕𝑡 𝜕𝑟
𝐼𝐶 ∶ 𝑡 = 0 𝑇 = 𝑇𝑜 𝐶𝐴 = 𝐶𝐴𝑖 for all 𝑟𝑝 > 𝑟 > 0
𝜕𝑇 𝜕𝐶𝐴
𝐵𝐶 1 ∶ 𝑟 = 0 =0 =0
𝜕𝑟 𝜕𝑟
𝜕𝑇 𝜕𝐶𝐴
2 ∶ 𝑟 = 𝑟𝑝 − 𝑘 𝜕𝑟 = ℎ𝑓 (𝑇 − 𝑇𝑓 ) −𝐷 = 𝑘𝑚 (𝐶𝐴 − 𝐶𝑏 )
𝜕𝑟
All it means is that their non-dimensionalized forms are identically the same:

𝜕𝜃 1 𝜕 𝜕𝜃
= 2 (𝜉 2 )
𝜕𝜏 𝜉 𝜕𝜉 𝜕𝜉

where,

𝑇 − 𝑇𝑜 𝐶𝐴 − 𝐶𝐴𝑖
𝜃= ,
𝑇𝑓 − 𝑇𝑜 𝐶𝐴,𝑏 − 𝐶𝐴𝑖
𝑟
𝜉 = ⁄𝑟𝑝

𝑡 𝑟𝑝2 𝑟𝑝2
{ 𝜏 = ⁄𝑡𝑐 ; 𝑡𝑐 = 𝛼 , 𝐷

𝐼𝐶. 𝜏=0 𝜃=0

𝜕𝜃
𝐵𝐶 1. 𝜉=0 =0
𝜕𝜉

𝜕𝜃
2. 𝜉 = 1 − 𝜕𝜉 = 𝐴(𝜃 − 1) , where A = Nu or Sh

Therefore, it is clear that their (non-dimensional) solutions will also be the same.

It also follows that one should non-dimensionalize the transport equations, as a good practice,
before solving the equations. Apart from learning about transport phenomena from the analogy
between heat, mass and momentum, there is a possibility that the non-dimensional forms of the
conservation equations & the respective bcs being the same, the non-dimensionalized solutions
will also be the same. In such cases or similar cases, the programming code written for one
problem will be the same or similar, requiring small modifications.

Let us continue with the non-dimensionalized form of the present mass transport problem:
𝜕𝜃 1 𝜕 𝜕𝜃 𝐶𝐴𝑖
= 𝜉2 𝜕𝜉 (𝜉 2 𝜕𝜉 ) − 𝐵𝜃 − 𝐶 , where 𝐵 = 𝑘𝑡𝑐 , 𝐶 = 𝑘𝑡𝑐 𝐶
𝜕𝜏 𝐴,𝑏 −𝐶𝐴𝑖

𝑁
𝜏+1
𝜏
𝑖=0 𝑖−1 𝑖 𝑖+1 𝑁

Discretize:

126
 𝜃𝑖𝜏+1 − 𝜃𝑖𝜏 1 𝜃𝑖+1 − 2𝜃𝑖 + 𝜃𝑖−1 𝜏+1 𝜃𝑖+1 − 2𝜃𝑖 + 𝜃𝑖−1 𝜏 2 𝜃𝑖+1 − 𝜃𝑖−1 𝜏+1
= [( ) +( ) + ( )
Δ𝜏 2 Δ𝜉 2 Δ𝜉 2 𝜉𝑖 2Δ𝜉
2 𝜃𝑖+1 − 𝜃𝑖−1 𝜏
+ ( ) − 𝐵(𝜃𝑖𝜏+1 + 𝜃𝑖𝜏 )] − 𝐶
𝜉𝑖 2Δ𝜉

𝑖 = 0, 𝑁

@𝑖 = 0 there is a discontinuity in the equation. Therefore,

𝜃0𝜏+1 −𝜃0𝜏 3 𝜃1 −2𝜃0 +𝜃−1 𝜏+1 𝜃1 −2𝜃0 +𝜃−1 𝜏 𝐵

= 2 [( ) +( ) + 2 (𝜃0𝜏+1 + 𝜃0𝜏 )] − 𝐶 for i = 0
Δ𝜏 Δ𝜉 2 Δ𝜉 2

Apply BCs

𝑖=0 𝜃−1 = 𝜃1
𝜃𝑁+1 −𝜃𝑁−1
𝑖=𝑁 − = 𝐴(𝜃𝑁 − 1) }
2Δ𝜉
𝑜𝑟 𝜃𝑁+1 = 𝜃𝑁−1 − 2𝐴Δ𝜉(𝜃𝑁 − 1)

1 3 𝐵 3 1 3 𝐵 3
𝑖 = 0: ( + 2 − ) 𝜃0𝜏+1 − 2 𝜃1𝜏+1 = ( − 2 + ) 𝜃0𝜏 + 𝜃𝜏 − 𝐶
Δ𝜏 Δ𝜉 2 Δ𝜉 Δ𝜏 Δ𝜉 2 2Δ𝜉 2 1

1 𝜏+1
1 𝐵 1 𝜏+1
1 𝜏
1 1 𝐵 𝜏
𝑖 = 𝑁: ( ) 𝜃𝑁−1 − ( + + ) 𝜃𝑁 = − ( ) 𝜃𝑁−1 − ( − − )𝜃 + 𝐶
Δ𝜉 2 Δ𝜉 2 2 Δ𝜏 Δ𝜉 2 Δ𝜏 Δ𝜉 2 2 𝑁

1 1 𝜏+1 1 1 𝐵 𝜏+1 1 1 𝜏+1

𝑖 = 1, 𝑁 − 1 ( 2
− ) 𝜃𝑖−1 − ( + 2
− ) 𝜃𝑖 + ( 2
− ) 𝜃𝑖+1
2Δ𝜉 2Δ𝜉𝜉𝑖 Δ𝜏 Δ𝜉 2 2Δ𝜉 2𝜉Δ𝜉
1 1 𝜏 1 1 𝐵 𝜏 1 1
= (− + ) 𝜃 − ( − − ) 𝜃 + (− − )𝜃 + 𝐶
2Δ𝜉 2 2Δ𝜉𝜉 𝑖−1
Δ𝜏 Δ𝜉 2 2 𝑖
2Δ𝜉 2 2Δ𝜉𝜉 𝑖+1

𝐴𝜃 𝜏+1 = 𝑑𝜏 , 𝑖 = 0, 𝑁 with 𝐼𝐶 𝜃 = 0 @ 𝜏 = 0
Solve
(same as before, 𝐴 ≡ Tridiagonal matrix)

Here, we skipped the steps for preparing tridiagonal matrix and directly substituted discretized
BCs into the discretized equations! As an exercise, prepare tridiagonal matrix and see if you get
the same discretized equations post substitution of BCs, as before.

Quiz III

127
 Lecture #24-25
Elliptic PDE (Method of Lines)
The model equations may be recognized by the following examples:

𝜕 𝑇2 𝜕 𝑇 2 2
(1) 𝑘 (𝜕𝑋 2 + 𝜕𝑦 2 ) + 𝑆 = 0 ; 𝑆 ≡ 𝐼 𝑅⁄∀ 𝑐𝑎𝑙⁄𝑠 − 𝑚3
𝜕2 𝑇 1 𝜕 𝜕𝑇
(2) 𝑘 (𝜕𝑋 2 + 𝑟 𝜕𝑟 (𝑟 𝜕𝑟 )) + 𝑆 = 0

--- SS 2D temperature profiles in a (1) rectangular plate and (2) cylindrical wire because of
uniform heating

𝜕𝐶 𝜕2 𝐶 𝐷 𝜕 𝜕𝐶
(3) 𝑉𝑋 𝜕𝑋 = 𝐷 𝜕𝑋 2 + 𝑟 (𝜕𝑟 (𝑟 𝜕𝑟 )) − 𝑘𝐶
---- SS 2D concentration distributions of a solute in the reactive flow in a tube

𝜕𝑇 𝜕2 𝑇 𝜕2 𝑇
(4) 𝑉𝑋 𝜕𝑋 = 𝛼 𝜕𝑋 2 + 𝛼 𝜕𝑦 2
- SS 2D temperature distributions in a flow through rectangular channel.

𝜕𝑉𝑋 𝜕2 𝑉 1 𝜕 𝜕𝑉𝑋 𝑑𝑝
(5) 𝜌𝑉𝑋 = 𝜇 ( 𝜕𝑋 2𝑋 + 𝑟 𝜕𝑟 (𝑟 )) − 𝑑𝑥
𝜕𝑋 𝜕𝑟

- SS 2D velocity profiles of an incompressible NF in a pressure-driven horizontal flow.

By now, you must have realized that we are referring to an elliptic PDE which describes a SS
2D (no time-dependent term) heat/mass/momentum transport. Why is the equation called
elliptic? Because one has to solve the entire 2D space. Considering that there is the SS
consideration, there is no ‘marching’ on time as such and the updating of the solutions from the
previous time-step, as we earlier discussed for the parabolic PDEs. The solution in this case
must be sought one-time only under SS conditions. You will see later that solving elliptic
equations is computational extensive requiring iterations.

- The best way of understanding the numerical technique for solving an elliptic PDE is to
directly take the example (1) above.

128
 Ex: A rectangular plate (𝐿 × 𝑤 × 𝑡ℎ ) fabricated from stainless steel (𝑘 = 40 𝑊 ⁄𝑚 − 𝑘 , 𝜌 =
8000 𝑘𝑔⁄𝑚3 , 𝐶𝑝 = 400 𝐽⁄𝑘𝑔 𝐾) is uniformly heated using an electric power source (100 𝑊).
The top and bottom ends are insulated, whereas the side surfaces are exposed to atmosphere
(𝑇𝑎 = 300 𝐶, ℎ = 100 𝑊 ⁄𝑚2 𝑘). Determine the SS temperature profiles in the plate.

𝑦
𝜕𝑇
−𝑘 =0
𝜕𝑦
𝑀 𝐿 100 𝑊
ℎ𝑓 S (𝑊 ⁄𝑚3 ) = ( )
𝐿 × 𝑊 × 𝑡ℎ
(ℎ𝑒𝑎𝑡 𝑠𝑜𝑢𝑟𝑐𝑒)
(𝑖, 𝑗 + 1) Δ𝑥

𝑤 Δ𝑦 ℎ𝑓
ℎ𝑓 (𝑖 + 1, 𝑗)
(𝑇𝑎 )300 𝐶 (𝑖 − 1, 𝑗) (𝑖, 𝑗)
30 0 𝐶(𝑇𝑎 )
(𝑖, 𝑗 − 1)

𝑂 N X

𝜕𝑇
−𝑘 =0
𝜕𝑦

2D energy balance over ′Δ𝑥Δ𝑦𝑡ℎ ′ 𝐶𝑉 under SS:

𝜕𝑇
𝜌𝐶𝑝 ( 𝜕𝑡 + 𝑉. ∇𝑇) = 𝑘∇2 𝑇 + 𝑆: 𝑊 ⁄𝑚3

𝑆𝑆 0, 𝑠𝑜𝑙𝑖𝑑

𝜕2 𝑇 𝜕2 𝑇
𝑘 (𝜕𝑥 2 + 𝜕𝑦 2 ) + 𝑆 = 0; S = 100 W/m3 of the plate volume

𝜕𝑇
BCs: 𝑥 = 0 𝑓or 𝑤 > 𝑦 > 0 ; −𝑘 𝜕𝑥 = −ℎ𝑓 (𝑇 − 𝑇𝑎 )

𝜕𝑇
= 𝐿 for 𝑤 > 𝑦 > 0 ; −𝑘 𝜕𝑥 = ℎ𝑓 (𝑇 − 𝑇𝑎 )

𝜕𝑇
𝑦 = 0 and 𝑤 for 𝐿 > 𝑥 > 0, − 𝑘 𝜕𝑦 = 0(insulation)

129
Computational molecule:
(𝑗 + 1) Δ𝑥 = 𝐿⁄𝑁
Δ𝑦 = 𝑊⁄𝑀

𝑖 = 0… 𝑁
(𝑖 − 1) (𝑖, 𝑗) (𝑖 + 1) }
𝑗 = 0… 𝑀

(𝑗 − 1)

Discretize equations over (𝑖, 𝑗). Note that it is a 2D problem. Therefore, one will have to
discretize in both directions (𝑥, 𝑦) using steps Δ𝑥 and Δ𝑦, respectively. They need not be equal.

𝜕2 𝑇 𝜕2 𝑇
𝑘 [(𝜕𝑋 2 ) + (𝜕𝑦 2 ) ] + 𝑆𝑖,𝑗 = 0
𝑖,𝑗 𝑖,𝑗

𝑇𝑖−1,𝑗 − 2𝑇𝑖,𝑗 + 𝑇𝑖+1,𝑗 𝑇𝑖,𝑗−1 − 2𝑇𝑖,𝑗 + 𝑇𝑖,𝑗+1 𝑆𝑖,𝑗

( 2
)+( 2
)+ =0
Δ𝑥 Δ𝑦 𝑘
𝜕2 𝑇 𝜕2 𝑇
(As expected, when discretizing 𝜕𝑋 2 , 𝑗 is constant, and when discretizing 𝜕𝑦 2 , 𝑖 is constant)

There are two ways to arrange the discretized terms to

𝐴𝑦̅ = 𝑏̅ form ∶
1 1 𝑆𝑖,𝑗
(1)Δ𝑥 2 (𝑇𝑖−1,𝑗 − 2𝑇𝑖,𝑗 + 𝑇𝑖+1,𝑗 ) = − Δ𝑦 2 (𝑇𝑖,𝑗−1 − 2𝑇𝑖,𝑗 + 𝑇𝑖,𝑗+1 ) − 𝑘

1 1 𝑆𝑖,𝑗
(2) Δ𝑦 2 (𝑇𝑖,𝑗−1 − 2𝑇𝑖,𝑗 + 𝑇𝑖,𝑗+1 ) = − Δ𝑥 2 (𝑇𝑖−1,𝑗 − 2𝑇𝑖,𝑗 + 𝑇𝑖+1,𝑗 ) − 𝑘

Notes: 1. Although the linear algebraic equations have taken the form of 𝐴𝑦̅ = 𝑏̅ in both cases,
𝑏̅ or RHS terms are not known. Therefore, in principle 𝑦̅ 𝑜𝑟 𝑇̅ cannot be solved by inverting
the ‘matrix’ formed on LHS, the way we did in the previous cases. In other words, there is no
way one can march in 𝑥 𝑜𝑟 𝑦 direction and solve the unknown variables, because marching in
either direction (𝑖 𝑜𝑟 𝑗) creates unknown variables in the other direction (𝑗 𝑜𝑟 𝑖). In fact, one
has to solve the ‘entire space’ at one time (without marching). This is the problem in solving an
elliptic PDE!

(2) The only way to solve an elliptic PDE or 𝐴𝑦̅ = 𝑏̅ , where 𝐴 is the tridiagonal matrix, is by
making guess for all variables (𝑇𝑖,𝑗 ) to start with, so that 𝑏̅ (RHS term) is known. Then, 𝑦̅
(which is similar to̅̅̅
𝑏 ) can be solved. Next, compare the newly calculated values with the guess
values and iterate till there is the convergence:

130
 b0 = initial guess for (Ti,j )
𝐴𝑦 𝑘+1 = 𝑏 𝑘
k = # of iterations.
(3) Either of the two discretized schemes (1) & (2) can be used to solve 𝑇𝑖,𝑗 . In general, one
should sweep the discretized set of equations in the direction the expected solution (functional
value) is lesser stiff than in the other direction. In the present example, 𝑤 ≪ 𝐿 and the y-ends
are insulated. One should use scheme (1). The convergence will be relatively faster.
1 1 𝑔 𝑆𝑖,𝑗
(𝑇𝑖−1,𝑗 − 2𝑇𝑖,𝑗 + 𝑇𝑖+1,𝑗 ) = − Δ𝑦 2 (𝑇𝑖,𝑗−1 − 2𝑇𝑖,𝑗 + 𝑇𝑖,𝑗+1 ) − (1)
Δ𝑥 2 𝐾

Prepare tridiagonal matrix:

1 2 1
𝑎(𝑖, 𝑗) = 2
; 𝑏(𝑖, 𝑗) = − 2 , 𝑐(𝑖, 𝑗) =
Δ𝑥 Δ𝑥 Δ𝑥 2 𝑖 = 0, 𝑁
1 𝑔 𝑆𝑖,𝑗 𝑗 = 0, 𝑀
𝑑(𝑖, 𝑗) = − 2 (𝑇𝑖,𝑗−1 − 2𝑇𝑖,𝑗 + 𝑇𝑖,𝑗+1 ) −
Δ𝑦 𝑘 }

The superscript "g" stands for the guess values. Once the initial guess is made, one can now
march along ‘𝑗’ direction, solving 𝑇𝑖,𝑗 at every ‘𝑗𝑡ℎ ’ step as 𝐴𝑦̅𝑗 = 𝑏, where 𝐴 is the
tridiagonal matrix containing the discretized ‘𝑦’ along ‘𝑖’ direction for a fixed 𝑗, and 𝑏̅ is known
from the guess.

Discretize BCs

𝑇1,𝑗 − 𝑇−1,𝑗
(𝑖 = 0), 𝑗: − 𝑘 = −ℎ𝑓 (𝑇0,𝑗 − 𝑇𝑎 )
2Δ𝑥
2ℎ𝑓 Δ𝑥
or 𝑇−1,𝑗 = 𝑇1,𝑗 − (𝑇0,𝑗 − 𝑇𝑎 )
𝑘
2ℎ𝑓 Δ𝑥
(𝑖
{ = 𝑁), 𝑗: 𝑇𝑁+1,𝑗 = 𝑇𝑁−1,𝑗 −
𝑘
(𝑇𝑁,𝑗 − 𝑇𝑎 )

(𝑗 = 0), 𝑖: 𝑇𝑖,−1 = 𝑇𝑖,1

(𝑗 = 𝑀), 𝑖: 𝑇𝑖,𝑀+1 = 𝑇𝑖,𝑀−1

Apply BCs

𝑗=0

𝑖=0 1 𝑖−1 𝑖 𝑖+1 𝑁−1 𝑁

131
 𝑐(0,0) = 𝑐(0,0) + 𝑎(0,0)
2ℎ𝑓 Δ𝑥
𝑏(0,0) = 𝑏(0,0) − 𝑎(0,0)
𝑘 𝑖=0
2 𝑔 𝑆0,0 2ℎ𝑓 Δ𝑥
𝑑(0,0) = 2 (𝑇0,0 − 𝑇0,1 ) − − . 𝑇𝑎 . 𝑎(0,0)
Δ𝑦 𝑘 𝑘 }
2 𝑔 𝑆𝑖,0
𝑑(𝑖, 0) = Δ𝑦 2 (𝑇𝑖,0 − 𝑇𝑖,1 ) − ; 𝑖 = 1, 𝑁 − 1
𝑘

𝑎(𝑁, 0) = 𝑎(𝑁, 0) + 𝑐(𝑁, 0)

2ℎ𝑓 Δ𝑥
𝑏(𝑁, 0) = 𝑏(𝑁, 0) − 𝑐(𝑁, 0)
𝑘 𝑖=𝑁
2 𝑔 𝑆𝑁,0 2ℎ𝑓 Δ𝑥
𝑑(𝑁, 0) = 2 (𝑇𝑁,0 − 𝑇𝑁,1 ) − − 𝑇𝑎 𝑎(𝑁, 0)
Δ𝑦 𝑘 𝑘 }

Tridiag(𝑁 + 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑇𝑖,0 ) (𝑖 = 0, 𝑁)

𝑗 = 1, 𝑀 − 1

𝑐(0, 𝑗) = 𝑐(0, 𝑗) + 𝑎(0, 𝑗)

2ℎ𝑓 Δ𝑥
𝑏(0, 𝑗) = 𝑏(0, 𝑗) − 𝑎(0, 𝑗)
𝑘 𝑖=0
2ℎ𝑓 Δ𝑥
𝑑(0, 𝑗) = 𝑑(0, 𝑗) − ( ) 𝑇𝑎 𝑎(0, 𝑗)
𝑘 }

⇒ No coefficients will change for the rows, 𝒊 = 𝟏, 𝑵 − 𝟏, because all of these correspond
to interior nodes:

𝑀
𝑀−1

𝑁−1
0,0 𝑖 𝑁

132
 𝑎(𝑁, 𝑗) = 𝑎(𝑁, 𝑗) + 𝑐(𝑁, 𝑗)
2ℎ𝑓 Δ𝑥
𝑏(𝑁, 𝑗) = 𝑏(𝑁, 𝑗) − ( ) 𝑐(𝑁, 𝑗)
𝑘 𝑖=𝑁
2ℎ𝑓 Δ𝑥
𝑑(𝑁, 𝑗) = 𝑑(𝑁, 𝑗) − ( ) 𝑇𝑎 𝑐(𝑁, 𝑗)
𝑘 }

Tridiag(𝑁 + 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑇𝑖,𝑗 ) (𝑖 = 0, 𝑁)

𝑗=𝑀

𝑐(0, 𝑗) = 𝑐(0, 𝑗) + 𝑎(0, 𝑗)

2ℎ𝑓 Δ𝑥
𝑏(0, 𝑗) = 𝑏(0, 𝑗) − 𝑎(0, 𝑗)
𝑘 𝑖=0
2 𝑔 𝑆0,𝑀 2ℎ𝑓 Δ𝑥
𝑑(0, 𝑗) = 2 (−𝑇0,𝑀−1 + 𝑇0,𝑀 ) − −( ) 𝑇𝑎 𝑎(0, 𝑗)
Δ𝑦 𝑘 𝑘 }
2 𝑔 𝑆𝑖,𝑀
𝑑(𝑖, 𝑗) = − Δ𝑦 2 (−𝑇𝑖,𝑀−1 + 𝑇𝑖,𝑀 ) − ; 𝑖 = 1, 𝑁 − 1
𝑘

𝑎(𝑁, 𝑗) = 𝑎(𝑁, 𝑗) + 𝑐(𝑁, 𝑗)

2ℎ𝑓 Δ𝑥
𝑏(𝑁, 𝑗) = 𝑏(𝑁, 𝑗) − 𝑐(𝑁, 𝑗)
𝑘 𝑖=𝑁
2 𝑔 𝑆𝑁,𝑀 2ℎ𝑓 Δ𝑥
𝑑(𝑁, 𝑗) = 2 (−𝑇𝑁,𝑀−1 + 𝑇𝑁,𝑀 ) − − 𝑇𝑎 𝑐(𝑁, 𝑗)
Δ𝑦 𝑘 𝑘 }

Tridiag(𝑁 + 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑇𝑖,𝑗 ) (𝑖 = 0, 𝑁)

You have, therefore, called Tridiagonal subroutine ‘𝑀 + 1’ times as you marched along ‘𝑗’
direction. At the end, entire ‘𝑥 − 𝑦’ grids have been solved for 𝑇𝑖,𝑗 (𝑖 = 0, 𝑁; 𝑗 = 0, 𝑀).
Compare solved values with the guess values and keep iterating till there is a convergence.

⇒ Solving an elliptic PDE is indeed computational extensive because there is a convergence

issue using the guesses.

⇒ There is another twist in solving an elliptic PDE. Do you update your guess values of 𝑏̅
(RHS terms) at ‘𝑗’ with the ones solved at ‘𝑗 − 1’ as you march in y-direction, or do you wait till
you have solved till ‘𝑀’ (the last boundary)? This question should remind you of the G-S and
Jacobi iterations. Choice is yours. A fast convergence is the criterion. Also, the relaxation
factor, ‘𝑤’ can be used to update the functional values in either case: 𝑇𝑖,𝑗 = 𝑤𝑇′𝑖,𝑗 + (1 − 𝑤)𝑇𝑖,𝑗
, where T’ and T are the new and old values, respectively.

133
 Lecture #25-26
Ex 2: Consider the fully developed SS flow (Re = 200) of an incompressible NF in a long
horizontal tube (L, D). The inlet temperature of the liquid is 𝑇𝑜 . The heat is supplied to the
flowing liquid at constant flux 𝑞𝑤 (𝑊 ⁄𝑚2 ) through the tube walls. Determine the 2D (𝑟, 𝑥) SS
temperature profiles in the tube.

𝐿 𝑞𝑤 𝑊 ⁄𝑚2

𝑅 Δ𝑟
T(r, X) = ? 𝑟
𝑇𝑜 𝑋 𝑟

Δ𝑋 𝑞𝑤 𝑊 ⁄𝑚2 Δ𝑟
2
𝑟
𝑢(𝑟) = 𝑈𝑚𝑎𝑥 (1 − ) (𝑅𝑒 = 200); 𝜌 , 𝐶𝑝 , 𝑘 ≡ 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑅2

′
Energy balance over 2𝜋𝑟Δ𝑟Δ𝑋 ′ 𝐶𝑉:

𝜌𝐶𝑝 𝑉𝑋 . ∇𝑇 = 𝑘∇2 𝑇 + 𝑆 ( 𝐽 𝑜𝑟 𝑐𝑎𝑙 ⁄𝑠 − 𝑚3 )

𝜕𝑇 𝜕2 𝑇 1 𝜕 𝜕𝑇 ̅
𝑘
𝑉(𝑟) 𝜕𝑋 = 𝛼 (𝜕𝑋 2 + 𝑟 𝜕𝑟 (𝑟 𝜕𝑟 )) 𝛼 = 𝜌̅𝐶̅ ;

BC. 𝑋 = 0 𝑇(𝑟) = 𝑇𝑜 𝑅 ≥ 𝑟 ≥ 0

𝜕𝑇
𝑋=𝐿 + 𝜕𝑋 = 0 (long tube approximation)

𝜕𝑇
𝑟=0 = 0 (symmetric BC)
𝜕𝑟
𝜕𝑇
𝑟=𝑅 −𝑘 = −𝑞𝑊 (𝑊 ⁄𝑚2 ) 𝐿≥𝑧≥0
𝜕𝑟
𝜕𝑇
𝑜𝑟 𝑘 𝜕𝑟 = 𝑞𝑊 }
Let us non-dimensionalize the equation & BC
𝑇 𝑋 Umax R r
𝜃= , 𝑧= 𝑤here Lc = RPe where Pe (radial) = , ξ=
To Lc α R

𝑈𝑚𝑎𝑥(1 − 𝜉 2 ) 𝜕𝜃 1 𝜕 2 𝜃 1 𝜕 2 𝜃 1 1 𝜕𝜃
= α( 2 2 + 2 2 + 2 )
Lc ∂z Lc ∂z R ∂ξ R ξ ∂ξ

134
 or
𝜕𝜃 1 𝜕 2 𝜃 𝜕 2 𝜃 1 𝜕𝜃
(1 − 𝜉 2 ) = 2 2+ 2+
∂z 𝑃𝑒 ∂z ∂ξ ξ ∂ξ

BCs:

𝑧=0 𝜃=1
+ 𝜕𝑧 = 0} 1 > 𝜉 > 0
𝜕𝜃
𝑧 = 𝐿⁄𝐿
𝑐

𝜕𝜃
𝜉=0 =0
𝜕𝜉

𝜕𝜃 𝑞𝑊 𝑅
𝜉=1 =( ) = 𝐻(constant) for 𝐿⁄𝐿 > 𝑧 > 0
𝜕𝜉 𝑘𝑇𝑜 𝑐

Before solving, let us consider another example on mass transport.

Ex. Consider the fully developed SS flow (𝑅𝑒 = 200) of an incompressible NF in a long
horizontal tube (𝐿, 𝐷). The inlet concentration of the species A in the liquid is 𝐶𝐴𝑜 .The species
are catalytically destroyed at the tube walls by the zeroth order chemical reaction
(𝑘 𝑚𝑜𝑙𝑒⁄𝑠 − 𝑚3 ). Determine the 2D (𝑟, 𝑋)𝑆𝑆 concentration profiles in the tube.

𝑅 𝐶(𝑟, 𝑋) 𝑟 = −𝑘 Δ𝑟
𝐶𝐴𝑜 𝑟 𝑋

𝑟2
𝑢(𝑟) = 𝑈𝑚𝑎𝑥 (1 − 2 ) Δ𝑋
𝑅

Species balance over (2𝜋𝑟Δ𝑟Δ𝑋) 𝐶, 𝑉.

𝜕𝐶𝐴 𝜕 2 𝐶𝐴 1 𝜕 𝜕𝐶𝐴
𝑉𝑋 = 𝐷( 2 + (𝑟 )) moles⁄s − m3
𝜕𝑋 𝜕𝑋 𝑟 𝜕𝑟 𝜕𝑟

𝑟 2 𝜕𝐶𝐴 𝜕 2 𝐶𝐴 1 𝜕 𝜕𝐶𝐴
𝑈𝑚𝑎𝑥 (1 − 2
) = 𝐷 ( 2
+ (𝑟 ))
𝑅 𝜕𝑋 𝜕𝑋 𝑟 𝜕𝑟 𝜕𝑟

𝜕2 𝐶 𝜕2 𝐶𝐴 1 𝜕𝐶𝐴
= 𝐷 ( 𝜕𝑋 2𝐴 + +𝑟 )
𝜕𝑟 2 𝜕𝑟

𝜕𝐶𝐴
BCs. 𝑋 = 0 𝐶𝐴 = 𝐶𝐴𝑜 , 𝑋 = 𝐿 = 0 (𝑅 > 𝑟 > 0)
𝜕𝑥

𝜕𝐶𝐴 𝜕𝐶𝐴
𝑟=0 =0 𝑟=𝑅 , −𝐷 =𝑘 (𝐿 > 𝑋 > 0)
𝜕𝑟 𝜕𝑟

Let us non-dimensionalize the equation & BCs:

135
 𝐶𝐴 𝑥 𝑟 𝑈𝑚𝑎𝑥 𝑅
𝜃= ⁄𝐶 , 𝑧= , 𝜉= ; 𝐿𝑐 = 𝑅𝑃𝑒 , where Pe (radial) =
𝐴𝑜 𝐿𝑐 𝑅 𝐷

(Note that this ′𝑃𝑒 ′ is based on mass transport)

2)
𝜕𝜃 1 𝜕 2 𝜃 𝜕 2 𝜃 1 𝜕𝜃
(1 − 𝜉 = + +
𝜕𝑧 𝑃𝑒2 𝜕𝑧 2 𝜕𝜉 2 𝜉 𝜕𝜉
𝜕𝜃
BC. 𝑧 = 0, 𝜃=1 ; 𝑧 = 𝐿⁄𝐿 , =0
𝑐 𝜕𝑧

𝜕𝜃 𝜕𝜃 𝑘𝑅
𝜉 = 0, =0 ; 𝜉=1 − 𝜕𝜉 = (𝐷𝐶 ) = 𝑀′ (constant)
𝜕𝜉 𝐴𝑜

⇒ It is clear that the non-dimensionalized equations and BCs of this (mass transport) and
previous (heat transport) examples are the same. Therefore, you need to solve only one of the
two. The dimensionless solutions will be the same. This situation should remind you of the
recommendation that one should non-dimensionalize the equation before solving it. Several
such analogous heat, mass, and momentum transport scenarios exist in chemical engineering
applications.

⇒In most cases, radial Pe (mass or heat) number in laminar flow regime is of the order of 10
or higher. In such cases, the axial diffusion term can be neglected. In other words,

𝜕2 𝑇 𝜕𝑇 𝜕2 𝐶𝐴 𝜕𝐶𝐴
𝛼 𝜕𝑋 2 << 𝑉𝑋 𝜕𝑋 𝑜𝑟 𝐷 << 𝑉𝑋 (See BSL book)
𝜕𝑋 2 𝜕𝑋

If you neglect the axial diffusion term, the 2D elliptic PDE is modified/simplified to 1D
parabolic PDE:
𝜕𝑇 𝛼 𝜕 𝜕𝑇 𝜕𝐶𝐴 𝐷 𝜕 𝜕𝐶𝐴
𝑉𝑋 𝜕𝑋 = (𝑟 𝜕𝑟 ) 𝑜𝑟 𝑉𝑋 = (𝑟 )
𝑟 𝜕𝑟 𝜕𝑋 𝑟 𝜕𝑟 𝜕𝑟

This simplified PDE can be solved using the Crank-Nicholson technique described in the
preceding lecture. In other words, one can numerically march in ‘X’ direction and solve in ‘r’
direction using Thomas Algorithm for the tridiagonal matrix built from the discretized ‘r’
terms. For now, let us revert to the original (full) non-dimensionalized elliptic PDE:

𝜕𝜃 1 𝜕 2 𝜃 𝜕 2 𝜃 1 𝜕𝜃
(1 − 𝜉 2 ) = + +
𝜕𝑧 𝑃𝑒2 𝜕𝑧 2 𝜕𝜉 2 𝜉 𝜕𝜉

At 𝜉 = 0, there is a discontinuity and you are solving an approximation:

𝜕𝜃 1 𝜕 2 𝜃 2𝜕 2 𝜃
(1 − 𝜉 2 ) = 2 2+
𝜕𝑧 𝑃𝑒 𝜕𝑧 𝜕𝜉 2

136
Discretize the main conservation equation:
𝜃𝑖+1,𝑗 −𝜃𝑖−1,𝑗 1 𝜃𝑖+1,𝑗 −2𝜃𝑖,𝑗 +𝜃𝑖−1,𝑗 𝜃𝑖,𝑗+1 −2𝜃𝑖,𝑗 +𝜃𝑖,𝑗−1 1 𝜃𝑖,𝑗+1 −𝜃𝑖,𝑗−1
(1 − 𝜉𝑗 2 ) = 𝑃2 + +𝜉
2Δ𝑧 𝑒 Δ𝑧 2 Δ𝜉 2 𝑗 2Δ𝜉
or

(1−𝜉𝑗 2 ) 1 𝑖,𝑗2𝜃 (1−𝜉𝑗 2 ) 1 1 1

( + 𝑃2 Δ𝑧 2 ) 𝜃𝑖−1,𝑗 − 𝑃2 Δ𝑧 2
−( − 𝑃2 Δ𝑧 2 ) 𝜃𝑖+1,𝑗 = (2Δ𝜉𝜉 − Δ𝜉2 ) 𝜃𝑖,𝑗−1 +
2Δ𝑧 𝑒 𝑒 2Δ𝑧 𝑒 𝑗
2𝜃𝑖,𝑗 1 1
− (2Δ𝜉𝜉 + Δ𝜉2 ) 𝜃𝑖,𝑗+1
Δ𝜉 2 𝑗

- (1)
0 Δ𝑧 2 i−1 i i+1 N 𝑖 = 1, 𝑁
𝑗 = 1, 𝑀
Discretize the approximated conservation equation at the line of symmetry:

(1−𝜉𝑗 2 ) 1 𝑖,02𝜃 (1−𝜉𝑗 2 ) 1 2𝜃𝑖,−1 4𝜃𝑖,0 2𝜃𝑖,+1

( + 𝑃2 Δ𝑧 2 ) 𝜃𝑖−1,0 − 𝑃2 Δ𝑧 2 −( − 𝑃2 Δ𝑧 2 ) 𝜃𝑖+1,0 = − + −
2Δ𝑧 𝑒 𝑒 2Δ𝑧 𝑒 Δ𝜉 2 Δ𝜉 2 Δ𝜉 2

𝑖 = 1, 𝑁
𝑗 = 0 (𝑛𝑜𝑡𝑒 𝜉0 = 0)
𝐿 ⁄𝐿 𝑐
Δ𝑧 = 1⁄𝑁 , Δ𝜉 =
𝑀

Note that RHS terms for all rows contain guess values for all 𝜃𝑖,𝑗 to begin with

Discretize BCs:

𝑖 = 0, 𝑗 = 0, 𝑀; 𝜃0,𝑗 = 1; 𝑖 = 𝑁, 𝑗 = 0, 𝑀; 𝜃𝑁+1,𝑗 = 𝜃𝑁−1,𝑗

𝑗 = 0, 𝑖 = 1, 𝑁; 𝜃𝑖,−1 = 𝜃𝑖,1 ; 𝑗 = 𝑀, 𝑖 = 0, 𝑁; 𝜃𝑖,𝑀+1 = 𝜃𝑖,𝑀−1 − 2𝑀′Δ𝜉

𝑗=0

0 1 i−1 i i+1 N−1 N

1st row of Tridiagonal matrix,

2𝜃1,0 1 − 𝜉02 1 4𝜃1,0 4𝜃1,0 1 − 𝜉02 1

𝑖=1 − 2 2−( − 2 2 ) 𝜃2,0 = − + − ( + )𝜃
𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧 Δ𝜉 2 Δ𝜉 2 2Δ𝑧 𝑃𝑒2 Δ𝑧 2 0,0

=1

137
middle rows,

(1 − 𝜉0 2 ) 1 2𝜃𝑖,0 (1 − 𝜉0 2 ) 1
𝑖 = 2, 𝑁 − 1: ( + 2 2 ) 𝜃𝑖−1,0 − 2 2 − ( − 2 2 ) 𝜃𝑖+1,0
2Δ𝑧 𝑃𝑒 Δ𝑧 𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧
4𝜃𝑖,1 4𝜃𝑖,0
=− 2 +
Δ𝜉 Δ𝜉 2
2 2𝜃 𝑁,0 4𝜃𝑁,1 4𝜃𝑁,0
last row, 𝑖 = 𝑁: 𝜃 − 𝑃2 Δ𝑧 =− +
𝑃𝑒2 Δ𝑧 2 𝑁−1,0 𝑒
2 Δ𝜉 2 Δ𝜉 2

𝑘+1 𝑘
You have a tridiagonal matrix A in (𝐴𝜃𝑖,0 = 𝑏𝑖,0 ) to invert

𝑗 = 1, 𝑀 − 1 (Visit the main conservation equation)

1st row,
2𝜃1,𝑗 1 − 𝜉𝑗2 1 1 − 𝜉𝑗2 1
𝑖 = 1: − 2 2−( − 2 2 ) 𝜃2,𝑗 = − ( + 2 2 ) 𝜃0,𝑗
𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧
1
1 1 2𝜃1,𝑗 1 1
(2Δ𝜉𝜉 − Δ𝜉2 ) 𝜃1,𝑗−1 + − (2Δ𝜉𝜉 + Δ𝜉2 ) 𝜃1,𝑗+1
𝑗 Δ𝜉 2 𝑗

middle rows:

𝑖 = 2, 𝑁 − 1 : These are middle grids unaffected by BCs.

last row

2 2𝜃𝑁,𝑗 1 1 2𝜃𝑁,𝑗 1 1
𝑖=𝑁 𝜃𝑁−1,𝑗 − =( − 2 ) 𝜃𝑁,𝑗−1 + −( + 2 ) 𝜃𝑁,𝑗+1
𝑃𝑒2 Δ𝑧 2 2
𝑃𝑒 Δ𝑧 2 2Δ𝜉𝜉𝑗 Δ𝜉 Δ𝜉 2 2Δ𝜉𝜉𝑗 Δ𝜉

𝐾+1 𝐾 𝑖 = 1, 𝑁
Again, 𝐴𝜃𝑖,𝑗 = 𝑏𝑖,𝑗 ;
𝑗 = 1, 𝑀 − 1

𝑗=𝑀

1st row,
2𝜃1,𝑀 (1 − 𝜉𝑀 2 ) 1 (1 − 𝜉𝑀 2 ) 1
𝑖 =1 − 2 2−( − 2 2 ) 𝜃2,𝑀 = − ( + 2 2 ) 𝜃0,𝑀
𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧 2Δ𝑧 𝑃𝑒 Δ𝑧
1
2 2𝜃1,𝑀 1 1
− Δ𝜉2 𝜃1,𝑀−1 + + 2𝑀′Δ𝜉 (2Δ𝜉𝜉 + Δ𝜉2 )
Δ𝜉 2 𝑀

middle rows, LHS (same as that of eq1), unaffected by BC =

2 2𝜃𝑖,𝑀 1 1
i = 2, N-1 − Δ𝜉2 𝜃𝑖,𝑀−1 + Δ𝜉 2
+ 2𝑀′Δ𝜉 (2Δ𝜉𝜉 + Δ𝜉2 )
𝑀

138
last row,

2𝜃𝑁−1,𝑀 2𝜃𝑁,𝑀 2 2𝜃𝑁,𝑀 1 1

𝑖=𝑁 − = − 𝜃𝑁,𝑀−1 + + 2𝑀′Δ𝜉 ( + )
𝑃𝑒2 Δ𝑧 2 𝑃𝑒2 Δ𝑧 2 Δ𝜉 2 Δ𝜉 2 2Δ𝜉𝜉𝑀 Δ𝜉 2

𝑘+1 𝑘
Again, 𝐴𝜃𝑖,𝑀 = 𝑏𝑖,𝑀

You have called the Thomas Algorithm (M+1) times to invert the tridiagonal A matrix to solve
𝜃𝑖,𝑗 , 𝑖 = 1, 𝑁
} using the guess values on the RHS of the tridiagonal matrix .
𝑗 = 0, 𝑀

You must have noted that we did not prepare the tridiagonal matrix, and instead directly
substituted the BCs in the discretized equations! As an exercise, prepare the tridiagonal matrix
by defining individuals elements of rows for every ‘j’ viz a(i, j), b(i, j) ⋯ ⋯, substitute BCs and
see if you get the same equations for different ′𝑗𝑠′ .

Now, this is the time to compare the solved 𝜃𝑖,𝑗 values with the guess values you made at the
beginning of the iterations before starting the iterations.

⇒ From the programming point of view, you should be able to choose reasonable values of all
variables including, D, Vmax , M and Pe,mass transport , or H and Pe,heat transport .

⇒ As earlier noted, solving elliptic PDE is computational extensive, requiring guess values,
iterations and convergence. Very often, one artificially inserts a transient term
𝜕𝐶 𝜕𝑇 ⃗⃗
𝜕𝑉
( 𝜕𝑡𝐴 or 𝜌𝐶𝑝 𝜕𝑡 or 𝜌 𝜕𝑡 ) and seeks SS solutions, which is the focus of the last two lectures.

139
 Lecture #26-27
𝑇𝑖𝑚𝑒 − 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 2𝐷 𝑝𝑎𝑟𝑎𝑏𝑜𝑙𝑖𝑐 𝑃𝐷𝐸: 𝐴𝐷𝐼 𝑀𝑒𝑡ℎ𝑜𝑑

We are solving, for examples:

𝜕𝐶𝐴
+ 𝑉. ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 + (−𝑟𝐴 ) ; 𝐶𝐴 (𝑡, 𝑟, 𝑥)
𝜕𝑡
or
𝜕𝑇
𝜌𝐶𝑝 ( + 𝑉. ∇𝑇) = 𝑘∇2 𝑇 + (−𝑟𝐴 )(Δ𝐻) ; 𝑇(𝑡, 𝑟, 𝑥)
𝜕𝑡
Note: SS solution must be the same as that of the converged solution of the analogous 2D
𝜕𝐶 𝜕𝑇
elliptic PDE ( 𝜕𝑡𝐴 = = 0) discussed in the previous lectures.
𝜕𝑡

Let us take a general case of the time-dependent 2D PDE:

𝜕𝜙
= 𝜙𝑋𝑋 + 𝜙𝑋 + 𝜙𝑌𝑌 + 𝜙𝑌 ; 𝜙(𝑡, 𝑥, 𝑦)
𝜕𝑡
with necessary IC and BCs.

Apply Crank-Nicholson method/scheme to discretize 𝜙𝑋𝑋 , 𝜙𝑌𝑌 , 𝜙𝑋 , 𝑎𝑛𝑑 𝜙𝑌 terms, in the

similar fashion solved the time-dependent 1D PDE on 𝜙(𝑡, 𝑥):

𝜙 𝑡+1 − 𝜙 𝑡
( )
Δ𝑡 𝑖,𝑗
1 𝜙𝑖+1,𝑗 − 2𝜙𝑖,𝑗 + 𝜙𝑖−1,𝑗 𝑡+1 𝜙𝑖+1,𝑗 − 𝜙𝑖−1,𝑗 𝑡+1
= [( ) +( )
2 Δ𝑋 2 2ΔX
𝜙𝑖,𝑗+1 − 2𝜙𝑖,𝑗 + 𝜙𝑖,𝑗−1 𝑡 𝜙𝑖,𝑗+1 − 𝜙𝑖,𝑗−1 𝑡
+( ) +( )]
Δ𝑌 2 2ΔY

- This way the X-derivatives have been discretized implicitly on time, whereas the Y-
derivatives have been discretized explicitly.
- Alternatively, one can discretize Y-derivatives implicitly on time, whereas X-
derivatives can be discretized explicitly.
- Re-arranging the terms as 𝐴∅𝑥 𝑡+1 = ∅𝑦 𝑡 𝑜𝑟 𝐴∅𝑦 𝑡+1 = ∅𝑥 𝑡 from either scheme, it
is clear that ‘A’ will be a tridiagonal matrix, and one can proceed on time-step by
solving 𝜙 on X-Y plane:

140
 1 1 𝑡+1 1 1 𝑡+1 1 1
( − ) 𝜙𝑖−1,𝑗 −( + ) 𝜙𝑖,𝑗 +( + ) 𝜙 𝑡+1
2Δ𝑋 2 4Δ𝑋 Δ𝑡 Δ𝑋 2 2Δ𝑋 2 4Δ𝑋 𝑖+1,𝑗
1 1 𝑡 1 1 𝑡 1 1 𝑡
=( − 2
) 𝜙 𝑖,𝑗−1 − ( − 2
) 𝜙𝑖,𝑗 − ( + 2
) 𝜙𝑖,𝑗+1
4Δ𝑌 2Δ𝑌 Δ𝑡 Δ𝑌 4Δ𝑌 2Δ𝑌
- This way you are ‘marching’ in ‘𝑗’ direction and ‘sweeping’ in ‘i’ direction.
By the second scheme, if you ‘march’ in ‘i’ direction and ‘sweep’ in ‘𝑗’-direction, you will get the
following equation:

1 1 𝑡+1 1 𝑡+1 1 𝑡+1 1 1

(4Δ𝑌 − 2Δ𝑌 2 ) 𝜙𝑖,𝑗−1 + (Δ𝑡 + Δ𝑌 2 ) 𝜙𝑖,𝑗 − (4Δ𝑌 + 2Δ𝑌 2 ) 𝜙𝑖,𝑗+1

1 1 𝑡 1 1 𝑡 1 1 𝑡
= ( − ) 𝜙𝑖−1,𝑗 + ( − 2
) 𝜙𝑖,𝑗 + ( 2
+ ) 𝜙𝑖+1,𝑗
2Δ𝑋 2 4Δ𝑋 Δ𝑡 Δ𝑋 2Δ𝑋 4Δ𝑋
− Both schemes will work and the Crank-Nicholson method will produce 2nd order accuracy.

- Alternate Direct Implicit (ADI) method is an improved method producing 4th order
accuracy without any extra computational cost (for more, refer the book by Ferziger):

First discretize 𝜙𝑋 and 𝜙𝑋𝑋 terms implicitly and 𝜙𝑌 and 𝜙𝑌𝑌 terms explicitly to solve 𝜙
over half step (Δ𝑡⁄2) and then discretize 𝜙𝑌 and 𝜙𝑌𝑌 terms implicitly and 𝜙𝑋 and 𝜙𝑋𝑋 terms
explicitly over the next half step:

Step1:
1
𝜙𝑡+ ⁄2 − 𝜙 𝑡 1 1
( ) = [(𝜙𝑋𝑋 + 𝜙𝑋 )𝑡+ ⁄2 + (𝜙𝑌𝑌 + 𝜙𝑌 )𝑡 ]
Δ 𝑡⁄2 2
𝑖,𝑗

𝑖𝑚𝑝𝑙𝑖𝑐𝑖𝑡 𝑒𝑥𝑝𝑙𝑖𝑐𝑖𝑡

Step2:
1⁄
𝜙 𝑡+1 − 𝜙𝑡+ 2 1 1
( ) = [(𝜙𝑋𝑋 + 𝜙𝑋 )𝑡+ ⁄2 + (𝜙𝑌𝑌 + 𝜙𝑌 )𝑡+1 ]
Δ 𝑡⁄2 2
𝑖,𝑗

𝑒𝑥𝑝𝑙𝑖𝑐𝑖𝑡 𝑖𝑚𝑝𝑙𝑖𝑐𝑖𝑡

𝑡+1⁄2
𝐴𝜙𝑥 = 𝜙𝑦𝑡
or }
𝑡+1⁄
and 𝐵𝜙𝑦𝑡+1 = 𝜙𝑥 2

The procedure can also be understood by the following illustration:

141
 𝑡 𝑡 + 1⁄2 𝑡+1
𝑖, 𝑗 + 1 𝑖, 𝑗 + 1 𝑖, 𝑗 + 1
𝑡 + 1⁄2 𝑡 + 1⁄2
𝑠𝑤𝑒𝑒𝑝 ′𝑖′ 𝑠𝑤𝑒𝑒𝑝 ′𝑗′
𝑖 − 1, 𝑗 𝑖, 𝑗 𝑖 + 1, 𝑗 𝑚𝑎𝑟𝑐ℎ ′𝑗′ 𝑖 − 1, 𝑗 𝑖, 𝑗 𝑖 + 1, 𝑗 𝑚𝑎𝑟𝑐ℎ ′𝑖′ 𝑖 − 1, 𝑗 𝑖, 𝑗 𝑖 + 1, 𝑗
𝑖, 𝑗 − 1 𝑖, 𝑗 − 1
𝑖, 𝑗 − 1

Ex: Consider 2D diffusion in a rectangular shaped solid porous carbon block. The block is
initially soaked with moisture, say at concentration 𝐶𝑜 (𝑚𝑜𝑙𝑒𝑠⁄𝑚3 ). At 𝑡 = 0+ all four sides of
the block are exposed to dry air (moisture concentration = 𝐶𝑖 ≪ 𝐶𝑜 ). We are interested in
calculating the unsteady-state concentration profiles of moisture within the block, ie. 𝐶(𝑡, 𝑥, 𝑦) =
?. Pore diffusion coefficient for moisture in solid is 𝐷𝑝𝑜𝑟𝑒 .
𝑦
Ans:
𝐶𝑖 𝑚𝑜𝑙𝑒𝑠⁄𝑚3
𝐿

C(t, x, y) =?
Δ𝑦
𝑤

Δ𝑋 X

A species balance over ′Δ𝑋Δ𝑦. 1′ 𝐶𝑉 will yield the following conservation equation:
𝜕𝐶𝐴
+ 𝑉. ∇𝐶𝐴 = 𝐷𝑝𝑜𝑟𝑒 ∇2 𝐶𝐴 + (−𝑟𝐴 )
𝜕𝑡

𝜕𝐶𝐴 𝜕2 𝐶 𝜕2 𝐶𝐴
= 𝐷𝑝𝑜𝑟𝑒 ( 𝜕𝑋 2𝐴 + )
𝜕𝑡 𝜕𝑦 2

𝑡=0 𝐶 = 𝐶𝑜 𝑓𝑜𝑟 𝐿 ≥ 𝑥 ≥ 0 ; 𝑤 ≥ 𝑦 ≥ 0

0+ 𝐶 = 𝐶𝑖∗ @ 𝑥 = 0 & 𝐿 for 𝑤 > 𝑦 > 0

}
and @ 𝑦 = 0 & 𝑤 for 𝐿 > 𝑥 > 0

(𝐶𝑖∗ is the solid phase moisture concentration at the surface of the block in equilibrium with
𝐶𝑖 in atmoshphere).

142
Step1: discretize ‘𝑖’ implicitly and ‘𝑗’ explicitly over (𝑡 & 𝑡 + 1⁄2)

𝑡+1⁄2 𝑡 1⁄
𝐶𝑖,𝑗 − 𝐶𝑖,𝑗 𝐷𝑝𝑜𝑟𝑒 𝐶𝑖+1,𝑗 − 2𝐶𝑖,𝑗 + 𝐶𝑖−1,𝑗 𝑡+ 2 𝐶𝑖,𝑗+1 − 2𝐶𝑖,𝑗 + 𝐶𝑖,𝑗−1 𝑡
= [( ) +( )]
Δ𝑡⁄ 2 Δ𝑋 2 Δ𝑦 2
2

(Δ𝑋 = L⁄𝑁 , Δ𝑦 = w⁄𝑀)

M
M−1
(i, j + 1)

0 1 𝑖 − 1, j (𝑖, j) 𝑖 + 1, j 𝑁−1 𝑁

(𝑖, j − 1)
−1
1
0
Arrange:

𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄ 𝐷𝑝𝑜𝑟𝑒 2 𝑡+1⁄2 𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄ 𝐷𝑝𝑜𝑟𝑒 𝑡 𝐷𝑝𝑜𝑟𝑒 2 𝑡

𝐶𝑖−1,𝑗2 − ( + ) 𝐶𝑖,𝑗 + 𝐶𝑖+1,𝑗2 = − 𝐶𝑖,𝑗−1 +( − ) 𝐶𝑖,𝑗 −
2Δ𝑋 2 Δ𝑋 2 Δ𝑡 2Δ𝑋 2 2Δ𝑦 2 Δ𝑦 2 Δ𝑡
𝐷𝑝𝑜𝑟𝑒 𝑡
𝐶𝑖,𝑗+1
2Δ𝑦 2

𝑖 = 1, 𝑁 − 1
𝑗 = 1, 𝑀 − 1

Step-2:

Discretize BCs: 𝐶𝑜,𝑗 = 𝐶𝑁,𝑗 = 𝐶 ∗ ; 𝑗 = 0, 𝑀

𝐶𝑖,0 = 𝐶𝑖,𝑀 = 𝐶 ∗ ; 𝑖 = 0, 𝑁

Prepare the tridiagonal matrix:

𝐷𝑝𝑜𝑟𝑒 𝐷𝑝𝑜𝑟𝑒 2 𝐷𝑝𝑜𝑟𝑒 𝐷𝑝𝑜𝑟𝑒 𝑡
𝑎(𝑖, 𝑗) = ; 𝑏(𝑖, 𝑗) = − ( Δ𝑋 2 + Δ𝑡 ) ; 𝑐(𝑖, 𝑗) = ; 𝑑(𝑖, 𝑗) = − 2Δ𝑦 2 𝐶𝑖,𝑗−1 +
2Δ𝑋 2 2Δ𝑋 2
𝐷𝑝𝑜𝑟𝑒 2 𝑡 𝐷𝑝𝑜𝑟𝑒 𝑡
( Δ𝑦 2 − Δ𝑡 ) 𝐶𝑖,𝑗 − 𝐶𝑖,𝑗+1
2Δ𝑦 2

𝑖 = 1, 𝑁 − 1
}
𝑗 = 1, 𝑀 − 1

143
Substitute BCs and the following coefficients will be modified:

𝑑(1, 𝑗) = 𝑑(1, 𝑗) − 𝑎(1, 𝑗)𝐶 ∗

} 𝑗 = 1, 𝑀 − 1
𝑑(𝑁 − 1, 𝑗) = 𝑑(𝑁 − 1, 𝑗) − 𝑐(𝑁 − 1, 𝑗)𝐶 ∗

Convince yourself that no other coefficients will change. This was actually an easy
problem, when all four boundary conditions were simple, i.e., functional values were prescribed.
Problems are complicated when you have ‘flux’/gradient or mixed boundary conditions; for
example,
𝜕𝐶
−𝐷 𝜕𝑋 = 𝑘𝑚 (𝐶 − 𝐶𝑎𝑡𝑚 )

𝑜𝑟 = Flux (known)
𝜕𝐶
and/or −𝐷 𝜕𝑦 = 𝑘𝐶 𝑜𝑟 𝑘𝑚 (𝐶 − 𝐶𝑎𝑡𝑚 )

In such cases the other coefficients may also change. Revert to the previous step.

Tridiagonal (𝑁 − 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦(𝑖, 𝑗)) 𝑖 = 1, 𝑁 − 1

⇒ You will be calling the subroutine ′𝑀 − 1′ times as you ‘march’ along ′𝑗′ direction. Now, you
𝑡+1⁄2 𝑖 = 1, 𝑁 − 1
have the values for 𝐶𝑖,𝑗 ,
𝑗 = 1, 𝑀 − 1

Step3:

Now march along ′𝑖′ direction and ‘sweep’ along ′𝑗′ direction to solve for
𝑡+1 𝑡+1⁄2
𝐶𝑖,𝑗 from 𝐶𝑖,𝑗 .

𝑡+1 𝑡+1⁄2 1
𝐶𝑖,𝑗 −𝐶𝑖,𝑗 𝐷𝑝𝑜𝑟𝑒 𝐶𝑖−1,𝑗 −2𝐶𝑖,𝑗 +𝐶𝑖+1,𝑗 𝑡+ ⁄2 𝐶𝑖,𝑗−1 −2𝐶𝑖,𝑗 +𝐶𝑖,𝑗+1 𝑡+1
Δ𝑡⁄ = [( ) +( ) ]
2 2 Δ𝑋 2 Δ𝑦 2

𝑒𝑥𝑝𝑙𝑖𝑐𝑖𝑡 𝑖𝑚𝑝𝑙𝑖𝑐𝑖𝑡

Arrange:

𝐷𝑝𝑜𝑟𝑒 𝑡+1 𝐷𝑝𝑜𝑟𝑒

𝑡+1 2 𝐷𝑝𝑜𝑟𝑒 𝑡+1 𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄ 𝐷𝑝𝑜𝑟𝑒
𝐶𝑖,𝑗−1 − ( Δ𝑦 2 + Δ𝑡 ) 𝐶𝑖,𝑗 + 𝐶𝑖,𝑗+1 = − 2Δ𝑋 2 𝐶𝑖−1,𝑗2 + ( Δ𝑋 2 −
2Δ𝑦 2 2Δ𝑦 2
2 𝑡+1⁄2 𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄
) 𝐶𝑖,𝑗 − 2Δ𝑋 2 𝐶𝑖+1,𝑗2 ;
Δ𝑡

𝑗 = 1, 𝑀 − 1
}
𝑖 = 1, 𝑁 − 1

144
Step-4:

Discretize the BCs: Same as before.

Prepare the tridiagonal matrix:

𝐷𝑝𝑜𝑟𝑒 𝐷𝑝𝑜𝑟𝑒 2 𝐷𝑝𝑜𝑟𝑒 𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄

𝑎(𝑖, 𝑗) = ; 𝑏(𝑖, 𝑗) = − ( Δ𝑦 2 + Δ𝑡 ) ; 𝑐(𝑖, 𝑗) = ; 𝑑(𝑖, 𝑗) = − 2Δ𝑋 2 𝐶𝑖−1,𝑗2 +
2Δ𝑦 2 2Δ𝑦 2
𝐷𝑝𝑜𝑟𝑒 2 𝑡+1⁄2 𝐷𝑝𝑜𝑟𝑒 𝑡+1⁄
( Δ𝑋 2 − Δ𝑡 ) 𝐶𝑖,𝑗 − 2Δ𝑋 2 𝐶𝑖+1,𝑗2 ;

𝑖 = 1, 𝑁 − 1
}
𝑗 = 1, 𝑀 − 1

Substitute the discretized BCs, and only the following coefficients will be modified:

𝑑(𝑖, 1) = 𝑑(𝑖, 1) − 𝑎(𝑖, 1)𝐶 ∗

} 𝑖 = 1, 𝑁 − 1
𝑑(𝑖, 𝑀 − 1) = 𝑑(𝑖, 𝑀 − 1) − 𝑐(𝑖, 𝑀 − 1)𝐶 ∗

Tridiagonal (𝑀 − 1, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦(𝑖, 𝑗)) 𝑗 = 1, 𝑀 − 1

⇒ You will be calling the subroutine ′𝑁 − 1′ times as you ‘march’ along ′𝑖′ direction. Thus, you
𝑡+1 𝑖 = 1, 𝑁 − 1
have solved for 𝐶𝑖,𝑗 ,
𝑗 = 1, 𝑀 − 1

Recap (solving 2nd order ODE and PDEs at a glance)

BVP: 𝐴𝜙𝑖 = 𝑏 (Direct Method)

1D parabolic: 𝐴𝜙𝑖𝑡+1 = 𝜙𝑖𝑡 (Crank-Nicholson Method)

𝑔𝑢𝑒𝑠𝑠 𝑔𝑢𝑒𝑠𝑠
2D Elliptic: 𝐴𝜙𝑖,𝑗 = 𝜙𝑗,𝑖 or 𝐴𝜙𝑗,𝑖 = 𝜙𝑖,𝑗 (Method of Lines)

2D parabolic: (ADI Method)

𝑡+1⁄2 𝑡+1⁄2
𝐴𝜙𝑖 = 𝜙𝑗𝑡 𝐴𝜙𝑗 = 𝜙𝑖𝑡
} or }
𝑡+1⁄ 𝑡+1⁄
and 𝐵𝜙𝑗𝑡+1 = 𝜙𝑖 2 and 𝐵𝜙𝑖𝑡+1 = 𝜙𝑗 2

where, A and B are the Tridiagonal matrices.

145
 Lecture #28
Example: Consider the SS flow of a liquid through a long tube. Reynolds number is 180. At
time 𝑡 = 0, a tracer is injected into the liquid at inlet to the tube. Diffusion coefficient of tracer
in the liquid is 𝐷 𝑐𝑚2 ⁄𝑠 . Determine the time profiles of the tracer concentrations (𝑟, 𝑥) in the
tube, ie. 𝐶(𝑡, 𝑟, 𝑥) =?

Soln.

𝐿 𝑅
𝐶𝑜 Δ𝑟
𝑅 C(t,r,x)=?
𝑟 𝑟

Δ𝑋 Δ𝑟
𝑂 X
𝑟2
V(r) = Umax (1 − )
𝑅2

A species balance over ′2𝜋𝑟Δ𝑟Δ𝑥 ′ 𝐶𝑉 yields the following eqn

𝜕𝐶
+ 𝑉. ∇𝐶 = 𝐷∇2 𝐶 + (−𝑟𝐴 )
𝜕𝑡

or
𝜕𝐶 𝜕𝐶 𝜕2 𝐶 1 𝜕 𝜕𝐶 𝐿>𝑋>0
+ 𝑉(𝑟) 𝜕𝑋 = 𝐷 (𝜕𝑋 2 + 𝑟 𝜕𝑟 (𝑟 𝜕𝑟 ))
𝜕𝑡 𝑅>𝑟>0

𝜕𝐶 𝜕𝐶 𝜕2 𝐶 1 𝜕𝐶 𝜕2 𝐶
or = (−𝑉(𝑟) 𝜕𝑋 + 𝐷 𝜕𝑋 2 ) + 𝐷 (𝑟 𝜕𝑟 + 𝜕𝑟 2 )
𝜕𝑡

𝑡=0 𝐶=0 𝐿 ≥ 𝑥 ≥ 0 and 𝑅 ≥ 𝑟 ≥ 0

0+ 𝐶 = 𝐶𝑜 @ 𝑥 = 0
𝜕𝐶 } for all 𝑅 > 𝑟 > 0
= 0@ 𝑋 = 𝐿
𝜕𝑋

(long tube approximation)

𝜕𝐶
=0 @𝑟 = 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐)
𝜕𝑟 } 𝑓𝑜𝑟 𝑎𝑙𝑙 𝐿 > 𝑋 > 0
=0 @𝑟 = 𝑅 (𝑛𝑜𝑛 − 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒 𝑤𝑎𝑙𝑙𝑠)

146
Note: You will encounter discontinuity at 𝑟 = 0 in the radial diffusion terms while discretizing.
Considering ∇𝐶 = 0, you will be solving the approximated equation instead.

𝜕𝐶 𝜕𝐶 𝜕2 𝐶 𝜕2 𝐶
= (−𝑉(𝑟) 𝜕𝑋 + 𝐷 𝜕𝑋 2 ) + 2𝐷 𝜕𝑟 2 𝑎𝑡 𝑟 = 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝐿 > 𝑋 > 0
𝜕𝑡

(𝑖, M)
(𝑖, M − 1)
(i, j + 1)

𝑖=0 1 𝑖 − 1, j (𝑖, j) 𝑖 + 1, j 𝑁−1 𝑁

(𝑖, j − 1)
−1
1

j= 0

Discretize the approximated equation:

𝑡+1⁄2 𝑡 𝑡+1⁄2
𝐶𝑖,𝑗 − 𝐶𝑖,𝑗 1 𝐶𝑖+1,𝑗 − 𝐶𝑖−1,𝑗 𝐶𝑖+1,𝑗 − 2𝐶𝑖,𝑗 + 𝐶𝑖−1,𝑗
= [((−𝑉𝑗 ) +𝐷 )
Δ𝑡⁄ 2 2Δ𝑋 Δ𝑋 2
2 implicit
𝑡
𝐶𝑖,𝑗+1 − 2𝐶𝑖,𝑗 + 𝐶𝑖,𝑗−1 ; 𝑗=0
+ 2𝐷 ( )] }
Δ𝑟 2 𝑖 = 1, 𝑁

explicit

Arrange;

𝑉𝑗 𝐷 𝑡+1⁄ 2 𝐷 𝑡+1⁄ 𝑉𝑗 𝐷 𝑡+1⁄

( + 2
) 𝐶𝑖−1,𝑗2 − ( + 2
) 𝐶𝑖,𝑗 2 − ( − 2
) 𝐶𝑖+1,𝑗2
4Δ𝑋 2Δ𝑋 Δ𝑡 Δ𝑋 4Δ𝑋 2Δ𝑋
𝐷 𝑡 2 2𝐷 𝑡 𝐷 𝑡 𝑗=0
= − ( 2 ) 𝐶𝑖,𝑗−1 − ( − 2 ) 𝐶𝑖,𝑗 − ( 2 ) 𝐶𝑖,𝑗+1 ; }
Δ𝑟 Δ𝑡 Δ𝑟 Δ𝑟 𝑖 = 1, 𝑁

147
 Similarly, discretize the main equation:
𝑡+1⁄2 𝑡 𝑡+1⁄2
𝐶𝑖,𝑗 − 𝐶𝑖,𝑗 1 𝐶𝑖+1,𝑗 − 𝐶𝑖−1,𝑗 𝐶𝑖+1,𝑗 − 2𝐶𝑖,𝑗 + 𝐶𝑖−1,𝑗
= [((−𝑉𝑗 ) +𝐷 )
Δ𝑡⁄ 2 2Δ𝑋 Δ𝑋 2
2 implicit
𝑡
1 𝐶𝑖,𝑗+1 − 𝐶𝑖,𝑗−1 𝐶𝑖,𝑗+1 − 2𝐶𝑖,𝑗 + 𝐶𝑖,𝑗−1 𝑗 = 1, 𝑀
+𝐷( + )] }
𝑟𝑗 2Δ𝑟 Δ𝑟 2 𝑖 = 1, 𝑁

explicit

Arrange,

𝑉𝑗 𝐷 𝑡+1⁄2 2 𝐷 𝑡+1⁄2 𝑉𝑗 𝐷 𝑡+1⁄2

( + ) 𝐶 − ( + ) 𝐶 − ( − ) 𝐶
4Δ𝑋 2Δ𝑋 2 𝑖−1,𝑗 Δ𝑡 Δ𝑋 2 𝑖,𝑗 4Δ𝑋 2Δ𝑋 2 𝑖+1,𝑗
𝐷 𝐷 𝑡 2 𝐷 𝑡
=( − 2
) 𝐶𝑖,𝑗−1 − ( − 2 ) 𝐶𝑖,𝑗
4𝑟𝑗 Δ𝑟 2Δ𝑟 Δ𝑡 Δ𝑟
𝐷 𝐷 𝑡 𝑗 = 1, 𝑀
−( + ) 𝐶𝑖,𝑗+1 ; }
4𝑟𝑗 Δ𝑟 2Δ𝑟 2 𝑖 = 1, 𝑁

𝑟𝑗 2
Note: 𝑉𝑗 = 𝑈𝑚𝑎𝑥 (1 − ) ; 𝑟𝑗 = 𝑗Δ𝑟
𝑅2

Discretize BCs & IC

𝑡 = 0, 𝐶(𝑖, 𝑗) = 0 𝑜r a small number 𝐶𝑖𝑛𝑡 ≪ 𝐶𝑜 (𝑁 ≥ 𝑖 ≥ 0 𝑎𝑛𝑑 𝑀 ≥ 𝑗 ≥ 0)

0+ 𝐶(0, 𝑗) = 𝐶𝑜
𝐶(𝑁+1,𝑗)−𝐶(𝑁−1,𝑗)
=0 } 𝑀>𝑗>0
Δ𝑋
𝑜𝑟 𝐶(𝑁 + 1, 𝑗) = 𝐶(𝑁 − 1, 𝑗)

𝐶(𝑖, −1) = 𝐶(𝑖, +1)

}𝑁 > 𝑖 > 0
𝐶(𝑖, 𝑀 + 1) = 𝐶(𝑖, 𝑀 − 1)

Prepare tridiagonal matrix

0 𝑉 𝐷 2 𝐷 0 𝑉 𝐷
𝑗 = 0: 𝑎(𝑖, 0) = (4Δ𝑋 + 2Δ𝑋 2 ) ; 𝑏(𝑖, 0) = − (Δ𝑡 + Δ𝑋 2 ) ; 𝑐(𝑖, 0) = − (4Δ𝑋 − 2Δ𝑋 2) ; 𝑑(𝑖, 0) =
𝐷 𝑡 2 2𝐷 𝑡 𝐷 𝑡
− (Δ𝑟 2 ) 𝐶𝑖,−1 − (Δ𝑡 − Δ𝑟 2 ) 𝐶𝑖,0 − (Δ𝑟 2 ) 𝐶𝑖,+1 ; 𝑖 = 1, 𝑁

Substitute BCs,

2𝐷 𝑡 2 2𝐷 𝑡
𝑑(1,0) = − ( 2
) 𝐶1,1 − ( − 2 ) 𝐶1,0 − 𝑎(1,0)𝐶𝑜
Δ𝑟 Δ𝑡 Δ𝑟
𝑎(𝑁, 0) = 𝑎(𝑁, 0) + 𝑐(𝑁, 0)

148
 2𝐷𝑡 2 𝑡 2𝐷
𝑑(𝑁, 0) = − (Δ𝑟 2 ) 𝐶𝑁,1 − (Δ𝑡 − Δ𝑟 2 ) 𝐶𝑁,0

1⁄
Tridiagonal (𝑁, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦 𝑡+ 2 (𝑖, 0)) ; 𝑖 = 1, 𝑁

Preparation of tridiagonal matrix… continue….

𝑉𝑗 𝐷 2 𝐷 𝑉𝑗 𝐷
𝑎(𝑖, 𝑗) = ( + 2
) ; 𝑏(𝑖, 𝑗) = − ( + 2
) ; 𝑐(𝑖, 𝑗) = − ( − ) ; 𝑑(𝑖, 𝑗)
4Δ𝑋 2Δ𝑋 Δ𝑡 Δ𝑋 4Δ𝑋 2Δ𝑋 2
𝐷 𝐷 𝑡 2 𝐷 𝑡
=( − 2
) 𝐶𝑖,𝑗−1 − ( − 2 ) 𝐶𝑖,𝑗
4𝑟𝑗 Δ𝑟 2Δ𝑟 Δ𝑡 Δ𝑟
𝐷 𝐷 𝑡 𝑗 = 1, 𝑀
−( + ) 𝐶𝑖,𝑗+1 }
4𝑟𝑗 Δ𝑟 2Δ𝑟 2 𝑖 = 1, 𝑁

Sbustitute BC (𝑗 = 1, 𝑀 − 1)

𝑑(1, 𝑗) = 𝑑(1, 𝑗) − 𝑎(1, 𝑗)𝐶𝑜

𝑎(𝑁, 𝑗) = 𝑎(𝑁, 𝑗) + 𝑐(𝑁, 𝑗)

1⁄ 𝑖 = 1, 𝑁
Tridiagonal (𝑁, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦 𝑡+ 2 (𝑖, 𝑗)) ; }
𝑗 = 1, 𝑀 − 1

You have called the tridiagonal subroutine ′𝑀 − 1′ times for the interior nodes.

Substitute BC @ 𝑗 = 𝑀

𝐷 𝑡 2 𝐷 𝑡
𝑑(1, 𝑀) = − ( 2 ) 𝐶𝑖,𝑀−1 − ( − 2 ) 𝐶𝑖,𝑀 − a(1, N). Co
Δ𝑟 Δ𝑡 Δ𝑟
2 𝑡 2 𝐷 𝑡
𝑑(𝑖, 𝑀) = − ( 2
) 𝐶𝑖,𝑀−1 − ( − 2
) 𝐶𝑖,𝑀
Δ𝑟 Δ𝑡 Δ𝑟

𝑎(𝑁, 𝑀) = 𝑎(𝑁, 𝑀) + 𝑐(𝑁, 𝑀)

𝐷 𝑡 2 𝑡 𝐷
𝑑(𝑁, 𝑀) = − (Δ𝑟 2 ) 𝐶𝑁,𝑀−1 − (Δ𝑡 − Δ𝑟 2) 𝐶𝑁,𝑀

1⁄
Tridiagonal (𝑁, 𝑎, 𝑏, 𝑐, 𝑑, 𝑦 𝑡+ 2 (𝑖, 𝑀)) ; 𝑖 = 1, 𝑁

(Therefore, you have called the Thomas Algorithm (𝑀 + 1) times while sweeping 𝑗 =
0, 𝑀 rows)

149
For the last time in this course, let us write down complete eqn
1⁄
𝐴𝑦 𝑡+ 2 = 𝑑𝑡 𝑓𝑜𝑟 𝑗 = 0, 𝑖 = 1, 𝑁

𝑡+1⁄2
2 𝐷 𝑉0 𝐷
− ( + 2) − ( + ) ⋯ ⋯ ⋯
Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
𝑉0 𝐷 2 𝐷 𝑉0 𝐷 𝑌𝑖
⋯ ( + ) −( + 2) − ( − ) ⋯
4Δ𝑋 2Δ𝑋 2 Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
⋮ ⋮ ⋮ ⋮ ⋮
𝐷 2 𝐷
⋯ ⋯ ⋯ ( 2) − ( + 2 )] { }
[ Δ𝑥 Δ𝑡 Δ𝑥
𝑡
𝑑(1,0)
⋮
𝑑(𝑖, 0)
= 𝑖 = 2, 𝑁 − 1
⋮
⋮
{ 𝑑(𝑁, 0) }

Similarly, you can write a set of equations for 𝑗 = 1, 𝑀 − 1 and 𝑗 = 𝑀 using the coefficient as
above:

𝑗 = 1, 𝑀 − 1

𝑡+1⁄2
2 4 𝑉𝑗 𝐷
− ( + 2) − ( + ) ⋯ ⋯ ⋯
Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
𝑉𝑗 𝐷 2 𝐷 𝑉𝑗 𝐷 𝑌𝑖
⋯ ( + ) −( + 2) − ( − ) ⋯
4Δ𝑋 2Δ𝑋 2 Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
⋮ ⋮ ⋮ ⋮ ⋮
𝐷 2 𝐷
[ ⋯ ⋯ ⋯ ( 2) − ( + 2 )] { }
Δ𝑥 Δ𝑡 Δ𝑥
𝑡
𝑑(𝑖, 𝑗)
⋮
𝑑(𝑖, 𝑗)
= 𝑖 = 2, 𝑁 − 1
⋮
⋮
{ 𝑑(𝑁, 𝑗) }

150
𝑗=𝑀

𝑡+1⁄2
2 4 𝑉𝑀 𝐷
− ( + 2) − ( − ) ⋯ ⋯ ⋯
Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
𝑉𝑀 𝐷 2 𝐷 𝑉𝑀 𝐷 𝑌𝑖
⋯ ( + ) −( + 2) − ( − ) ⋯
4Δ𝑋 2Δ𝑋 2 Δ𝑡 Δ𝑥 4Δ𝑋 2Δ𝑋 2
⋮ ⋮ ⋮ ⋮ ⋮
𝐷 2 𝐷
⋯ ⋯ ⋯ ( 2) − ( + 2 )] { }
[ Δ𝑥 Δ𝑡 Δ𝑥
𝑡
𝑑(1, 𝑀)
⋮
𝑑(𝑖, 𝑀)
= 𝑖 = 2, 𝑁 − 1
⋮
⋮
{ 𝑑(𝑁, 𝑀) }

Now, sweep in ′𝑗′ direction and march in ′𝑖′ direction for (𝑡 + 1) based on the 𝑦(𝑖, 𝑗) values you
determined at (𝑡 + 1⁄2) step (above)

1⁄ 𝑖 = 1, 𝑁
𝐴𝑦 𝑡+1 (𝑖, 𝑗) = 𝑑𝑡+ 2 }
𝑗 = 0, 𝑀

Discretize 𝑗 = 0

𝑡+1 𝑡+1⁄2
𝐶𝑖,𝑗 − 𝐶𝑖,𝑗
Δ𝑡⁄
2
𝑡+1⁄2
1 𝐶𝑖+1,𝑗 − 𝐶𝑖−1,𝑗 𝐶𝑖+1,𝑗 − 2𝐶𝑖,𝑗 + 𝐶𝑖−1,𝑗
= [((−𝑉𝑗 ) +𝐷 )
2 2Δ𝑋 Δ𝑋 2
explicit
𝐶𝑖,𝑗+1 − 2𝐶𝑖,𝑗 + 𝐶𝑖,𝑗−1 𝑡+1
+ 2𝐷 ( ) ] (𝑖 = 1, 𝑁)
Δ𝑟 2

implicit

151
𝑗 = 1, 𝑀

𝑡+1 𝑡+1⁄2
𝐶𝑖,𝑗 − 𝐶𝑖,𝑗
Δ𝑡⁄
2
𝑡+1⁄2
1 𝐶𝑖+1,𝑗 − 𝐶𝑖−1,𝑗 𝐶𝑖+1,𝑗 − 2𝐶𝑖,𝑗 + 𝐶𝑖−1,𝑗
= [((−𝑉𝑗 ) +𝐷 )
2 2Δ𝑋 Δ𝑋 2
𝑡+1
1 𝐶𝑖,𝑗+1 − 𝐶𝑖,𝑗−1 𝐶𝑖,𝑗+1 − 2𝐶𝑖,𝑗 + 𝐶𝑖,𝑗−1
+𝐷( + ) ] (𝑖 = 1, 𝑁)
𝑟𝑗 2Δ𝑟 Δ𝑟 2

In this last lecture of the course, I leave it here for you to do the remaining part (arranging the
terms and applying BCs for 𝑖 = 1, (2 ⋯ 𝑁 − 1), 𝑁 rows to invert the matrix) as you sweep in
′𝑗′ direction, as an exercise. The topic on ADI stops here.

Note: (1) The SS solution 𝑦(𝑡, 𝑥, 𝑟) 𝑎𝑠 𝑡 → ∞ of the time-dependent 2D parabolic equation

must be the same as that of the corresponding 2D elliptic PDE you have learnt how to solve in
the preceding lecture, using ‘Method of Lines’, i.e. for
𝜕𝐶
+ 𝑉. ∇𝐶 = 𝐷∇2 𝐶 + (−𝑟𝐴 ) ;
𝜕𝑡

𝐶(𝑡, 𝑥, 𝑟) 𝑎𝑠 𝑡 → ∞ must be the same as that of

𝑉. ∇𝐶 = 𝐷∇2 𝐶 + (−𝑟𝐴 ) ⇒ 𝐶(𝑥, 𝑟)

(2) A question arises. When asked to solve the elliptic (2D) PDE, should not or cannot we
𝜕𝐶
artificially insert the transient term and seek the SS solution to the corresponding time-
𝜕𝑡
dependent 2D parabolic equation? Very often, yes. Recall that, solving elliptic PDE requires
iterations and there is always a convergence issue. How many iterations? On the other hand,
the parabolic equation does not require iterations, and you march on ′𝑡′ axis solving 𝑦(𝑥, 𝑟) at
every time step without iterations. Therefore, more than often the ADI method is preferred
over ‘Method of Lines’ for solving an elliptic (2D) PDE. Insert the transient term and solve till
you have SS solution.

(3) Before closing this chapter, let us answer how we address non-linearity in the differential
𝜕𝑉
term, for example, V𝜕𝑥 of the NS equation? The answer is simple. By iterations! Guess velocity
fields (Vg). Discretize the derivative term as before. Solve for velocity fields as before. Iterate

152
till there is convergence. Alternatively, Taylor’s series can also be used to approximate velocity
fields by linearization, in which case guesses are required for the velocity gradients.

End – Semester Exam

[In a regular semester, these course materials are usually covered in 28 lectures of 1 h 15 min
duration each, or 42 lectures of 50 min duration each. For record and due acknowledgement,
most of these materials were part of my graduate level lectures from Prof. Hermann F. Fasel,
AME, which I audited way back in 1995 at the University of Arizona, Tucson, USA. At Kanpur, I
offer this course to graduate students only. To this end, you are welcome to send me any
comments, or mistakes or errors you notice in the lectures, to my email id:
vermanishith@gmail.com.]

153
Heterogeneous Chemical
Reaction Engineering

1
 Lecture 01
Chemical Reaction Engineering (heterogeneous reaction)
Recommended books:

𝐶ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝐸𝑛𝑔𝑖𝑛𝑒𝑒𝑟𝑖𝑛𝑔 𝐾𝑖𝑛𝑒𝑡𝑖𝑐𝑠 by JM Smith, McGraw Hill, 2nd ed.

{𝐶ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝑅𝑒𝑎𝑐𝑡𝑜𝑟 𝐴𝑛𝑎𝑙𝑦𝑠𝑖𝑠 𝑎𝑛𝑑 𝐷𝑒𝑠𝑖𝑔𝑛 by Froment and Bischoff, Wily, 3rd ed.
𝐶ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝑎𝑛𝑑 𝐶𝑎𝑡𝑎𝑙𝑦𝑡𝑖𝑐 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝐸𝑛𝑔𝑖𝑛𝑒𝑒𝑟𝑖𝑛𝑔 by James Carberry, Dover

The focus is on the catalytic reactions in multi-phase systems:

𝑠 + 𝑔, 𝑠 + 𝑙,
𝑠 + 𝑙 + 𝑔,
{
𝑙+𝑔
(2 − 𝑜𝑟 𝑚𝑢𝑙𝑡𝑖 − 𝑝ℎ𝑎𝑠𝑒 𝑠𝑦𝑠𝑡𝑒𝑚𝑠)
𝑃𝑡
𝑒𝑔. 𝑁2 + 3𝐻2 → 2𝑁𝐻3

(𝑛𝑜𝑛 − 𝑐𝑎𝑡𝑎𝑙𝑦𝑡𝑖𝑐 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛: 𝐶 + 𝑂2 → 𝐶𝑂2 )

Review:

a) Representation of chemical reactions

∑𝑖 𝜈𝑖 𝐴𝑖 = 0
−𝑣𝑒 𝑓𝑜𝑟 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠
𝑠𝑡𝑜𝑖𝑐ℎ𝑖𝑜𝑚𝑒𝑡𝑟𝑖𝑐 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 ∶ {
+𝑣𝑒 𝑓𝑜𝑟 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠
𝑒. 𝑔. 𝑁2 + 3𝐻2 → 2𝑁𝐻3
⟹ 2𝑁𝐻3 − 3𝐻2 − 𝑁2 = 0 (𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛)
1 3
𝑜𝑟 𝑁2 + 2 𝐻2 → 𝑁𝐻3
2
1 3
⟹ 𝑁𝐻3 − 𝑁 − 2 𝐻2 = 0 (𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑎𝑡𝑖𝑜𝑛)
2 2

Both are stoichiometrically balanced equations, but do not show the path of the reactions…..to
be discussed later)

b) Reaction rate (𝒓):

𝑎𝐴 + 𝑏𝐵 + ⋯ → 𝑞𝑄 + 𝑠𝑆 + ⋯ ⋯

2
 # 𝑜𝑓 𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝑡ℎ𝑒 ′𝑖′ 𝑠𝑝𝑒𝑐𝑖𝑒𝑠
1 1 𝑑𝑁𝑖
𝑟= 𝜈𝑖 𝑤 𝑑𝑡
𝑎𝑡 𝑎 𝑡𝑖𝑚𝑒 ′𝑡′ 𝑖𝑛 𝑎
𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒 𝑠𝑦𝑠𝑡𝑒𝑚

𝑎𝑙𝑤𝑎𝑦𝑠 + 𝑣𝑒

, 𝑤ℎ𝑒𝑟𝑒 V = volume of the catalyst (m3 )

𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦
𝑑𝑁𝑖
= −𝑣𝑒 𝑓𝑜𝑟 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 or A = surface area (m2 )
𝑑𝑡
{ M = mass (g)
= +𝑣𝑒 𝑓𝑜𝑟 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠

⟹ ′ 𝑟 ′ is defined in such a way that it is an intrinsic property of the reaction (specific to the
reaction), and not species-specific; hence divided by ′𝑤′.
1 𝑑𝑁𝐴 1 𝑑𝑁𝐵 1 𝑑𝑄 1 𝑑𝑆
𝑟=− = − (𝑏𝑤) = (𝑞𝑤) = (𝑠𝑤)
(𝑎𝑤) 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡

⟹ Therefore, the rate of the reaction can be

Note that there is
} measured by measuring any of the
no indices ′i′ for the rate of raction
reactants or products.

+𝑣𝑒 − 𝑣𝑒 − 𝑣𝑒
11 𝑑𝑁𝐻3 1 𝑑𝑁2 1 1 𝑑𝐻2
𝑜𝑟, 𝑟 = = − =−
2 𝑤 𝑑𝑡 𝑤 𝑑𝑡 3 𝑤 𝑑𝑡

c) Reactions: Elementary or non – elementary (definition)

𝐴+𝐵 →𝐶

If the above reaction represents only mass balance:- Non – elementary

If the reaction represents path of the reaction:- Elementary

𝑒𝑔. 𝑁2 + 3𝐻2 → 2𝑁𝐻3

⟹ “One mole of 𝑁2 collides with 3 moles of 𝐻2 to produce 2 moles of 𝑁𝐻3 ” : only mass
balance (or stoichiometric balance) holds good; this is not a path of the reaction ⟹ It is
a non – elementary reaction.

⟹ The actual path(s) of the reaction is/are elementary, and are called mechanistic steps.

Mechanism is nothing but the sequence of elementary reactions.

3
 𝑒𝑔. 𝐻2 + 𝐶𝑙2 → 2𝐻𝐶𝑙
Stoichiometrically (mass) balanced reaction
But, it is a non – elementary reaction. Physical chemistry tells us that this
is not the path of the reaction. Actual paths are as follows:

𝐶𝑙2 ⇌ 2𝐶𝑙 Sequence of elementary reactions,

𝐶𝑙 + 𝐻2 → 𝐻 + 𝐻𝐶𝑙 } these are mechanistic steps or actual
⋮
paths.
𝑒𝑡𝑐

Revisit the book by Levenspiel or Fogler for certain rules/guidelines for a reaction to be
elementary, which are necessary conditions, but not sufficient for a reaction to be elementary;
the mechanism must be proposed from the knowledge of physical chemistry, and validated:

⟹ Maximum 3 – body collision of the reactant or product species (molecularity cannot exceed >
3). That is, for an elementary reaction, 𝑎𝐴 + 𝑏𝐵 → 𝑐𝐶 + 𝑑𝐷

a + b ≤ 3; c + d ≤ 3; and a, b, c, d, are + ve and whole numbers.

⟹ If the reversible reaction is non – elementary, the forward reaction is also non – elementary,
or vice –versa (note that no reaction is truly irreversible).

Some trivia:

𝑘1
1. 2𝐴 ⇌ 𝐵
𝑘2

𝑟 ≡ (𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙𝑙𝑦) = 𝑘1 𝐶𝐴 2 − 𝑘2 𝐶𝐵 . Is it an elementary or non –

elementary reaction? The answer is we do not know. The mechanism of the reaction is
required. However, if it is an elementary reaction,
𝑟 = 𝑘1 𝐶𝐴 2 − 𝑘2 𝐶𝐵 , and not vice versa.

𝑘1
1
2. 𝐴 ⇌𝐵 ⟹ Non – elementary, even if r is measured experimentally as
2
𝑘2

𝑘1 𝐶𝐴 0.5 − 𝑘2 𝐶𝐵 .

3. 2𝐴 ⇌ 4𝐵 ⟹ Non – elementary, because the reversible reaction cannot be

elementary.

4
4. 2𝐴 ⇌ 0.5𝐵 ⟹ Same as above.

𝑘1
5. 𝐴 ⇌ 𝐵 ; 𝑟𝑒𝑥𝑝𝑡 = 𝑘1 − 𝑘2 𝐶𝐵 ⟹ Non – elementary. Note: A zeroth order
𝑘2

reaction is always non- elementary.

6. 𝐼𝑓 2𝐴 ⇌ 4𝐵 + 𝐶 is an elementary reaction, what will be the rate? ⟹ Wrong

question. This reaction can never be elementary.

5
 Lecture 02
d) Rate Expression

For an elementary reaction, 𝑎𝐴 + 𝑏𝐵 → 𝑐𝐶 + 𝑑𝐷,

𝑟 = 𝑟(𝐶, 𝑇): 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛

where, C is the concentration of the reactant species.

= 𝑓(𝑇) 𝜙(𝐶) or 𝑓(𝑇) ∏𝑖 𝐶𝑖 |𝜈𝑖 |

𝑡𝑤𝑜 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛𝑠

𝑒𝑔. 𝐼𝑓 2𝐴 + 𝐵 → 𝐴2 𝐵 𝑖𝑠 𝑎𝑛 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑟𝑦 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟 = 𝑘𝐶𝐴 2 𝐶𝐵

𝑜𝑟, r = (k 0 T m e−E⁄RT ) CA 2 CB (0 ≤ 𝑚 ≤ 1)

where, m = 0 ⇒ k 0 e−E⁄RT CA 2 CB ∶ Arrhenius theory

1⁄ −E⁄RT
= 1⁄2 ⇒ k0 T 2e CA 2 CB ∶ Collision theory

=1 ⇒ rate is proportional to T : Transition state theory

Arrhenius theory is the most common

𝑟 = 𝑘0 𝑒 −𝐸⁄𝑅𝑇 ∏𝑖 𝐶𝑖 𝜈𝑖 𝑘0 = 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑓𝑎𝑐𝑡𝑜𝑟

;
𝑓 𝑛 𝑜𝑓 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝐸 = 𝐴𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 (𝐽⁄𝑚𝑜𝑙𝑒)

Below is the energy diagram for the reaction path:

Energy (J/mole)

𝐸1
𝐸−1

2A, B |Δ𝐻| (ℎ𝑒𝑎𝑡 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

𝐻1 𝐴2 𝐵 𝐻2

Reaction coordinate

In the energy diagram shown above,

6
 = H2 − H1 (−ve)
ΔH {
(E, overall = |E−1 − E1 | (always + ve)
E1 ≡ energy barrier for forward reaction = exothermic reaction
E−1 = energy barrier for reversible reaction (heat content of reactants >
that of product)
, else − ve for endothermic reaction.

The kinetic rate coefficients or constants can be written as

𝑘1 = 𝐴1 𝑒 −𝐸1 ⁄𝑅𝑇 ; 𝑘−1 = 𝐴2 𝑒 −𝐸−1 ⁄𝑅𝑇

𝐸1 , 𝐸−1 are always +𝑣𝑒 and are energy barriers for the forward and reversible reactions,
respectively; As are the pre-exponential factors.

⇒ The plot is also known as energy – barrier diagram.

⇒ E⁄RT is a dimensionless quantity. Therefore, “RT” can be construed as the “characteristic”

energy. E ≠ f(T); it is dependent on the path of the reaction. See the X-coordinate on the
energy barrier diagram.

⇒ The ratio denotes the minimum amount of energy required to overcome the barrier, relative
to the energy available at the reaction temperature, T, say, provided thermally via reactor
heating. Actual energy supplied at the temperature “T” requires an energy balance across the
reactor, to be discussed much later in the course, and can be expressed as m(RT), where “m” is
some positive number.

⇒ Considering the exponential nature of the rate coefficient (k) function, high ‘E’ means that the
rate of reaction is less sensitive to change in temperature (Δ𝑇) 𝑎𝑡 ℎ𝑖𝑔ℎ 𝑇; low ‘E’ means the rate
of reaction is relatively less sensitive to change in temperature (Δ𝑇) 𝑎𝑡 𝑙𝑜𝑤 𝑇. For more, see the
table for the rate constants calculated for different cases of low and high activation energies and
temperatures, in the Levenspiel’s book (Table 2.1):

𝐸 𝑇 𝑇

𝐸 𝑇 𝑇
e) Development of rate expression for a non- elementary reaction:

(A mechanism is required to develop the rate expressions. In other words, all mechanistic steps
must be known a priori)

Example: H2 + Br2 → 2HBr ( Overall stoichiometric reaction)

7
( Note: experimental measurements will not be consistent with 𝑟 expressed as k1 (H2 )(Br2 ) −
k −1 (HBr)2

1. Mechanistic steps (Free radical mechanism):

𝑘1
𝐵𝑟2 ⇌ 2𝐵𝑟 (𝐼𝑛𝑖𝑡𝑖𝑎𝑡𝑖𝑜𝑛⁄𝑇𝑒𝑟𝑚𝑖𝑛𝑎𝑡𝑖𝑜𝑛)
𝑘−1

𝑘2
 𝐵𝑟 + 𝐻2 ⇌ 𝐻𝐵𝑟 + 𝐻
𝑘−2
𝑃𝑟𝑜𝑝𝑎𝑔𝑎𝑡𝑖𝑜𝑛
𝑘3
𝐻 + 𝐵𝑟2 ⇌ 𝐻𝐵𝑟 + 𝐵𝑟
𝑘−3 }

These are all elementary reactions; they represent the actual paths of the reactions.

2. Simplify: 𝐻 + 𝐵𝑟 − 𝐵𝑟 ≫ 𝐻𝐵𝑟 + 𝐵𝑟 𝑜𝑟 𝑘3 ≫ 𝑘−3 (𝑛𝑒𝑔𝑙𝑖𝑔𝑖𝑏𝑙𝑒)

3. Write down the rate of reaction:

𝑑(𝐵𝑟2 ) 𝑑(𝐻2 ) 1 𝑑(𝐻𝐵𝑟) moles

𝑟=− =− = ; concentration in and ′w′ = volume
𝑑𝑡 𝑑𝑡 2 𝑑𝑡 cc

4. Choose a species as per your convenience of calculation or the experimental measurements.

𝑑(𝐵𝑟2 )
= −𝑘1 (𝐵𝑟2 ) + 𝑘−1 (𝐵𝑟) 2 − 𝑘3 (𝐵𝑟2 )(𝐻 ) (See the book by Carberry)
𝑑𝑡

5. Quasi- steady state assumption: 𝐵𝑟,  𝐻 are radicals & short-lived species. Therefore,

𝑑(𝐻 ) 𝑑(𝐵𝑟 )
= ≈0
𝑑𝑡 𝑑𝑡

(At the time scale of change in the concentration of the major species, such species remain
‘stationary’ or constant (almost zero, very-very small amount)).

𝑁𝐴 Major Species

‘Radicals’

Therefore, 𝑡

8
 𝑑(𝐻 )
= 𝑘2 (𝐵𝑟 )(𝐻2 ) − 𝑘−2 (𝐻𝐵𝑟)(𝐻) − 𝑘(𝐻)(𝐵𝑟2 ) = 0
𝑑𝑡

𝑑(𝐵𝑟 )
= 2𝑘1 (𝐵𝑟2 ) − 2𝑘−1 (𝐵𝑟 )2 − 𝑘2 (𝐵𝑟 )(𝐻2 ) + 𝑘−2 (𝐻𝐵𝑟)(𝐻) +
𝑑𝑡
𝑘3 (𝐻)(𝐵𝑟2 ) = 0

1 𝑑(𝐵𝑟.) 1 𝑑(𝐵𝑟2 )
(𝑁𝑜𝑡𝑒: 𝐵𝑟2 → 2𝐵𝑟 ; 𝑟= =− )
2 𝑑𝑡 1 𝑑𝑡

6. Substituting (remove all intermediate terms/species and retain major species)

𝑘1
𝑘3 𝑘2 √ (𝐵𝑟2 )3⁄2 (𝐻2 )
𝑘−1
𝑟=
𝑘−2 (𝐻𝐵𝑟) + 𝑘3 (𝐵𝑟2 )

(Note that the temperature dependence- and concentration dependence terms cannot be
separated)

Some Special Cases:

𝑘
1. 𝑟 = 𝑘2 √𝑘 1 (𝐵𝑟2 )1⁄2 (𝐻2 ) when (𝐻𝐵𝑟) ≪ (𝐵𝑟2 ) 𝑎t the begining of the reaction
−1

𝑘
𝑘3 𝑘2 √ 1 (𝐵𝑟2 )3⁄2 (𝐻2 )
𝑘−1
2. 𝑟 = when (𝐵𝑟2 ) ≪ (𝐻𝐵𝑟) towards the end of the reaction.
𝑘−2 (𝐻𝐵𝑟)

In either special case, 𝑟 = 𝑓(𝑇)𝜙(𝐶)

But, Note that it must not imply that the reaction is elementary!

Also, note that the order of a non-elementary reaction can be fraction; −ve; > 3

Go back to the first case:

𝑘
𝑟 = 𝑘, 𝑎𝑝𝑝 𝑓(𝐶𝑖 ) ⇒ 𝑘, 𝑎𝑝𝑝 = 𝑘2 √𝑘 1
−1

𝑒 −𝐸1⁄𝑅𝑇
Therefore, e−Eapp⁄RT = 𝑒 −𝐸2 ⁄𝑅𝑇 √
𝑒 −𝐸−1⁄𝑅𝑇

9
 E (apparent) does not have
physical meaning; can be + ve, −ve
𝐸1 −𝐸−1
𝐸𝑎𝑝𝑝 = (𝐸2 + ) no energy barrier or diagram.
2
However, E2 , E1 , E−1 are always + ve
{
  
𝑙𝑛(𝑘, 𝑎𝑝𝑝)  𝐸𝑎𝑝𝑝
p  𝐸𝑎𝑝𝑝 𝐸𝑎𝑝𝑝

1⁄𝑇
Conclusion: To obtain the rate expression for a non-elementary reaction, rate mechanism must
to be known a priori.

10
 # 𝑜𝑓 𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝑡ℎ𝑒 ′𝑖′ 𝑠𝑝𝑒𝑐𝑖𝑒𝑠
1 1 𝑑𝑁𝑖
𝑟= 𝜈𝑖 𝑤 𝑑𝑡
𝑎𝑡 𝑎 𝑡𝑖𝑚𝑒 ′𝑡′ 𝑖𝑛 𝑎
𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒 𝑠𝑦𝑠𝑡𝑒𝑚

𝑎𝑙𝑤𝑎𝑦𝑠 + 𝑣𝑒

, 𝑤ℎ𝑒𝑟𝑒 V = volume of the catalyst (m3 )

𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦
𝑑𝑁𝑖
= −𝑣𝑒 𝑓𝑜𝑟 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 or A = surface area (m2 )
𝑑𝑡
{ M = mass (g)
= +𝑣𝑒 𝑓𝑜𝑟 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠

⟹ ′ 𝑟 ′ is defined in such a way that it is an intrinsic property of the reaction (specific to the
reaction), and not species-specific; hence divided by ′𝑤′.
1 𝑑𝑁𝐴 1 𝑑𝑁𝐵 1 𝑑𝑄 1 𝑑𝑆
𝑟=− = − (𝑏𝑤) = (𝑞𝑤) = (𝑠𝑤)
(𝑎𝑤) 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡

⟹ Therefore, the rate of the reaction can be

Note that there is
} measured by measuring any of the
no indices ′i′ for the rate of raction
reactants or products.

+𝑣𝑒 − 𝑣𝑒 − 𝑣𝑒
11 𝑑𝑁𝐻3 1 𝑑𝑁2 1 1 𝑑𝐻2
𝑜𝑟, 𝑟 = = − =−
2 𝑤 𝑑𝑡 𝑤 𝑑𝑡 3 𝑤 𝑑𝑡

c) Reactions: Elementary or non – elementary (definition)

𝐴+𝐵 →𝐶

If the above reaction represents only mass balance:- Non – elementary

If the reaction represents path of the reaction:- Elementary

𝑒𝑔. 𝑁2 + 3𝐻2 → 2𝑁𝐻3

⟹ “One mole of 𝑁2 collides with 3 moles of 𝐻2 to produce 2 moles of 𝑁𝐻3 ” : only mass
balance (or stoichiometric balance) holds good; this is not a path of the reaction ⟹ It is
a non – elementary reaction.

⟹ The actual path(s) of the reaction is/are elementary, and are called mechanistic steps.

Mechanism is nothing but the sequence of elementary reactions.

3
 𝑒𝑔. 𝐻2 + 𝐶𝑙2 → 2𝐻𝐶𝑙
Stoichiometrically (mass) balanced reaction
But, it is a non – elementary reaction. Physical chemistry tells us that this
is not the path of the reaction. Actual paths are as follows:

𝐶𝑙2 ⇌ 2𝐶𝑙 Sequence of elementary reactions,

𝐶𝑙 + 𝐻2 → 𝐻 + 𝐻𝐶𝑙 } these are mechanistic steps or actual
⋮
paths.
𝑒𝑡𝑐

Revisit the book by Levenspiel or Fogler for certain rules/guidelines for a reaction to be
elementary, which are necessary conditions, but not sufficient for a reaction to be elementary;
the mechanism must be proposed from the knowledge of physical chemistry, and validated:

⟹ Maximum 3 – body collision of the reactant or product species (molecularity cannot exceed >
3). That is, for an elementary reaction, 𝑎𝐴 + 𝑏𝐵 → 𝑐𝐶 + 𝑑𝐷

a + b ≤ 3; c + d ≤ 3; and a, b, c, d, are + ve and whole numbers.

⟹ If the reversible reaction is non – elementary, the forward reaction is also non – elementary,
or vice –versa (note that no reaction is truly irreversible).

Some trivia:

𝑘1
1. 2𝐴 ⇌ 𝐵
𝑘2

𝑟 ≡ (𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙𝑙𝑦) = 𝑘1 𝐶𝐴 2 − 𝑘2 𝐶𝐵 . Is it an elementary or non –

elementary reaction? The answer is we do not know. The mechanism of the reaction is
required. However, if it is an elementary reaction,
𝑟 = 𝑘1 𝐶𝐴 2 − 𝑘2 𝐶𝐵 , and not vice versa.

𝑘1
1
2. 𝐴 ⇌𝐵 ⟹ Non – elementary, even if r is measured experimentally as
2
𝑘2

𝑘1 𝐶𝐴 0.5 − 𝑘2 𝐶𝐵 .

3. 2𝐴 ⇌ 4𝐵 ⟹ Non – elementary, because the reversible reaction cannot be

elementary.

4
4. 2𝐴 ⇌ 0.5𝐵 ⟹ Same as above.

𝑘1
5. 𝐴 ⇌ 𝐵 ; 𝑟𝑒𝑥𝑝𝑡 = 𝑘1 − 𝑘2 𝐶𝐵 ⟹ Non – elementary. Note: A zeroth order
𝑘2

reaction is always non- elementary.

6. 𝐼𝑓 2𝐴 ⇌ 4𝐵 + 𝐶 is an elementary reaction, what will be the rate? ⟹ Wrong

question. This reaction can never be elementary.

5
 Lecture 02
d) Rate Expression

For an elementary reaction, 𝑎𝐴 + 𝑏𝐵 → 𝑐𝐶 + 𝑑𝐷,

𝑟 = 𝑟(𝐶, 𝑇): 𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡𝑎𝑙 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛

where, C is the concentration of the reactant species.

= 𝑓(𝑇) 𝜙(𝐶) or 𝑓(𝑇) ∏𝑖 𝐶𝑖 |𝜈𝑖 |

𝑡𝑤𝑜 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑒 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛𝑠

𝑒𝑔. 𝐼𝑓 2𝐴 + 𝐵 → 𝐴2 𝐵 𝑖𝑠 𝑎𝑛 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑟𝑦 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟 = 𝑘𝐶𝐴 2 𝐶𝐵

𝑜𝑟, r = (k 0 T m e−E⁄RT ) CA 2 CB (0 ≤ 𝑚 ≤ 1)

where, m = 0 ⇒ k 0 e−E⁄RT CA 2 CB ∶ Arrhenius theory

1⁄ −E⁄RT
= 1⁄2 ⇒ k0 T 2e CA 2 CB ∶ Collision theory

=1 ⇒ rate is proportional to T : Transition state theory

Arrhenius theory is the most common

𝑟 = 𝑘0 𝑒 −𝐸⁄𝑅𝑇 ∏𝑖 𝐶𝑖 𝜈𝑖 𝑘0 = 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑓𝑎𝑐𝑡𝑜𝑟

;
𝑓 𝑛 𝑜𝑓 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝐸 = 𝐴𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 (𝐽⁄𝑚𝑜𝑙𝑒)

Below is the energy diagram for the reaction path:

Energy (J/mole)

𝐸1
𝐸−1

2A, B |Δ𝐻| (ℎ𝑒𝑎𝑡 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

𝐻1 𝐴2 𝐵 𝐻2

Reaction coordinate

In the energy diagram shown above,

6
 Lecture 04
Review: Ideal reactors

𝐵𝑎𝑡𝑐ℎ 𝑃𝑙𝑢𝑔 𝑓𝑙𝑜𝑤 𝑀𝑖𝑥𝑒𝑑 𝑡𝑎𝑛𝑘 𝑟𝑒𝑎𝑐𝑡𝑜𝑟 (𝐶𝑆𝑇𝑅)

𝑣𝑜
𝑋𝐴𝑖
𝑣𝑜
𝑋𝐴𝑖 𝑋𝐴𝑓 𝑋𝐴𝑓

𝑝𝑒𝑟𝑓𝑒𝑐𝑡𝑙𝑦 𝑚𝑖𝑥𝑒𝑑; 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑚𝑖𝑥𝑖𝑛𝑔 𝑖𝑛 𝑟 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑠𝑎𝑚𝑒 𝑎𝑠 𝑡ℎ𝑒 𝑏𝑎𝑡𝑐ℎ

𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝑖𝑠 𝑧𝑒𝑟𝑜 𝑚𝑖𝑥𝑖𝑛𝑔 𝑖𝑛 𝑎𝑥𝑖𝑎𝑙 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑜𝑟 } 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠 𝑒𝑥𝑐𝑒𝑝𝑡 𝑡ℎ𝑎𝑡
𝑠𝑝𝑎𝑡𝑖𝑎𝑙𝑙𝑦 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑧 ≠ 𝑣𝑧 (𝑟) 𝑡ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑓𝑙𝑜𝑤.
(𝑇, 𝜌, 𝑐, 𝐶𝑝 )

Design equation of ideal reactors:

𝜈𝐴 𝐴 + 𝜈𝐵 𝐵 + ⋯ → 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑠 ( 𝑟 ≡ 𝑟𝑎𝑡𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

Batch:
𝑋𝐴𝑓 𝑉 V = volume of the reaction
𝑜 𝑑𝑋𝐴
𝑡 = 𝐶𝐴𝑂 ∫𝑋 mixture (t)
𝐴𝑖 𝑉 (−𝜈𝐴 𝑟)
XA = conversion(t)
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒
𝑁𝐴𝑂 − 𝑁𝐴 Vo = volume of the mixture
𝑓𝑜𝑟 𝑡ℎ𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑡𝑜
𝑋𝐴 (𝑡) = at reference time
𝑟𝑒𝑎𝑐ℎ 𝑁𝐴𝑂
NAO = #of moles at reference time
NA = #of moles at time ′t ′

Plug flow reactor:

𝑉 𝑋𝐴𝑓 𝑑𝑋
𝜏= = 𝐶𝐴𝑂 ∫𝑋 𝐴
; vo = volumetric flow rate of
𝑣𝑜 𝐴𝑖 (−𝜈𝐴 𝑟) feed at reference point
(𝑠𝑝𝑎𝑐𝑒 − 𝑡𝑖𝑚𝑒)

𝐹𝐴𝑂 −𝐹𝐴
If flow: 𝑋𝐴 = ( )
𝐹𝐴𝑂
(# of moles⁄time ) ≡ molar flow rate

(Note: space time ≠ residense time because of volume-change)

15
 = H2 − H1 (−ve)
ΔH {
(E, overall = |E−1 − E1 | (always + ve)
E1 ≡ energy barrier for forward reaction = exothermic reaction
E−1 = energy barrier for reversible reaction (heat content of reactants >
that of product)
, else − ve for endothermic reaction.

The kinetic rate coefficients or constants can be written as

𝑘1 = 𝐴1 𝑒 −𝐸1 ⁄𝑅𝑇 ; 𝑘−1 = 𝐴2 𝑒 −𝐸−1 ⁄𝑅𝑇

𝐸1 , 𝐸−1 are always +𝑣𝑒 and are energy barriers for the forward and reversible reactions,
respectively; As are the pre-exponential factors.

⇒ The plot is also known as energy – barrier diagram.

⇒ E⁄RT is a dimensionless quantity. Therefore, “RT” can be construed as the “characteristic”

energy. E ≠ f(T); it is dependent on the path of the reaction. See the X-coordinate on the
energy barrier diagram.

⇒ The ratio denotes the minimum amount of energy required to overcome the barrier, relative
to the energy available at the reaction temperature, T, say, provided thermally via reactor
heating. Actual energy supplied at the temperature “T” requires an energy balance across the
reactor, to be discussed much later in the course, and can be expressed as m(RT), where “m” is
some positive number.

⇒ Considering the exponential nature of the rate coefficient (k) function, high ‘E’ means that the
rate of reaction is less sensitive to change in temperature (Δ𝑇) 𝑎𝑡 ℎ𝑖𝑔ℎ 𝑇; low ‘E’ means the rate
of reaction is relatively less sensitive to change in temperature (Δ𝑇) 𝑎𝑡 𝑙𝑜𝑤 𝑇. For more, see the
table for the rate constants calculated for different cases of low and high activation energies and
temperatures, in the Levenspiel’s book (Table 2.1):

𝐸 𝑇 𝑇

𝐸 𝑇 𝑇
e) Development of rate expression for a non- elementary reaction:

(A mechanism is required to develop the rate expressions. In other words, all mechanistic steps
must be known a priori)

Example: H2 + Br2 → 2HBr ( Overall stoichiometric reaction)

7
( Note: experimental measurements will not be consistent with 𝑟 expressed as k1 (H2 )(Br2 ) −
k −1 (HBr)2

1. Mechanistic steps (Free radical mechanism):

𝑘1
𝐵𝑟2 ⇌ 2𝐵𝑟 (𝐼𝑛𝑖𝑡𝑖𝑎𝑡𝑖𝑜𝑛⁄𝑇𝑒𝑟𝑚𝑖𝑛𝑎𝑡𝑖𝑜𝑛)
𝑘−1

𝑘2
 𝐵𝑟 + 𝐻2 ⇌ 𝐻𝐵𝑟 + 𝐻
𝑘−2
𝑃𝑟𝑜𝑝𝑎𝑔𝑎𝑡𝑖𝑜𝑛
𝑘3
𝐻 + 𝐵𝑟2 ⇌ 𝐻𝐵𝑟 + 𝐵𝑟
𝑘−3 }

These are all elementary reactions; they represent the actual paths of the reactions.

2. Simplify: 𝐻 + 𝐵𝑟 − 𝐵𝑟 ≫ 𝐻𝐵𝑟 + 𝐵𝑟 𝑜𝑟 𝑘3 ≫ 𝑘−3 (𝑛𝑒𝑔𝑙𝑖𝑔𝑖𝑏𝑙𝑒)

3. Write down the rate of reaction:

𝑑(𝐵𝑟2 ) 𝑑(𝐻2 ) 1 𝑑(𝐻𝐵𝑟) moles

𝑟=− =− = ; concentration in and ′w′ = volume
𝑑𝑡 𝑑𝑡 2 𝑑𝑡 cc

4. Choose a species as per your convenience of calculation or the experimental measurements.

𝑑(𝐵𝑟2 )
= −𝑘1 (𝐵𝑟2 ) + 𝑘−1 (𝐵𝑟) 2 − 𝑘3 (𝐵𝑟2 )(𝐻 ) (See the book by Carberry)
𝑑𝑡

5. Quasi- steady state assumption: 𝐵𝑟,  𝐻 are radicals & short-lived species. Therefore,

𝑑(𝐻 ) 𝑑(𝐵𝑟 )
= ≈0
𝑑𝑡 𝑑𝑡

(At the time scale of change in the concentration of the major species, such species remain
‘stationary’ or constant (almost zero, very-very small amount)).

𝑁𝐴 Major Species

‘Radicals’

Therefore, 𝑡

8
 𝑑(𝐻 )
= 𝑘2 (𝐵𝑟 )(𝐻2 ) − 𝑘−2 (𝐻𝐵𝑟)(𝐻) − 𝑘(𝐻)(𝐵𝑟2 ) = 0
𝑑𝑡

𝑑(𝐵𝑟 )
= 2𝑘1 (𝐵𝑟2 ) − 2𝑘−1 (𝐵𝑟 )2 − 𝑘2 (𝐵𝑟 )(𝐻2 ) + 𝑘−2 (𝐻𝐵𝑟)(𝐻) +
𝑑𝑡
𝑘3 (𝐻)(𝐵𝑟2 ) = 0

1 𝑑(𝐵𝑟.) 1 𝑑(𝐵𝑟2 )
(𝑁𝑜𝑡𝑒: 𝐵𝑟2 → 2𝐵𝑟 ; 𝑟= =− )
2 𝑑𝑡 1 𝑑𝑡

6. Substituting (remove all intermediate terms/species and retain major species)

𝑘1
𝑘3 𝑘2 √ (𝐵𝑟2 )3⁄2 (𝐻2 )
𝑘−1
𝑟=
𝑘−2 (𝐻𝐵𝑟) + 𝑘3 (𝐵𝑟2 )

(Note that the temperature dependence- and concentration dependence terms cannot be
separated)

Some Special Cases:

𝑘
1. 𝑟 = 𝑘2 √𝑘 1 (𝐵𝑟2 )1⁄2 (𝐻2 ) when (𝐻𝐵𝑟) ≪ (𝐵𝑟2 ) 𝑎t the begining of the reaction
−1

𝑘
𝑘3 𝑘2 √ 1 (𝐵𝑟2 )3⁄2 (𝐻2 )
𝑘−1
2. 𝑟 = when (𝐵𝑟2 ) ≪ (𝐻𝐵𝑟) towards the end of the reaction.
𝑘−2 (𝐻𝐵𝑟)

In either special case, 𝑟 = 𝑓(𝑇)𝜙(𝐶)

But, Note that it must not imply that the reaction is elementary!

Also, note that the order of a non-elementary reaction can be fraction; −ve; > 3

Go back to the first case:

𝑘
𝑟 = 𝑘, 𝑎𝑝𝑝 𝑓(𝐶𝑖 ) ⇒ 𝑘, 𝑎𝑝𝑝 = 𝑘2 √𝑘 1
−1

𝑒 −𝐸1⁄𝑅𝑇
Therefore, e−Eapp⁄RT = 𝑒 −𝐸2 ⁄𝑅𝑇 √
𝑒 −𝐸−1⁄𝑅𝑇

9
 E (apparent) does not have
physical meaning; can be + ve, −ve
𝐸1 −𝐸−1
𝐸𝑎𝑝𝑝 = (𝐸2 + ) no energy barrier or diagram.
2
However, E2 , E1 , E−1 are always + ve
{
  
𝑙𝑛(𝑘, 𝑎𝑝𝑝)  𝐸𝑎𝑝𝑝
p  𝐸𝑎𝑝𝑝 𝐸𝑎𝑝𝑝

1⁄𝑇
Conclusion: To obtain the rate expression for a non-elementary reaction, rate mechanism must
to be known a priori.

10
 Lecture 05
- Continue…….. ( non-ideal reactors)

Concept of “pockets” and criteria for macro-mixing:

1) Contents of individual pockets are well mixed with themselves, but do not mix with
second pockets.
2) Pockets should represent chemical reaction conditions. For 𝐴 + 𝐵 → 𝐶, one cannot,
therefore, reduce the size of a pocket to one molecule; then the reaction will not occur.
Pockets should have a finite volume.

3) Pocket-contents should have same residence time, but different pockets may have

different residence times.

⇒ Macro- mixing → macro – fluids (segregated fluids) → macro - reactor

Ideal PFR is a good example of macro – reactors

𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑝𝑜𝑐𝑘𝑒𝑡𝑠
(𝑠𝑎𝑡𝑖𝑠𝑓𝑦 𝑐𝑟𝑖𝑡𝑒𝑟𝑖𝑎)

 Fast fluidized bed reactor and liquid-liquid emulsion (LLE) membrane-based reactive
continuous separation processes are examples of industrial or practical reactors. In the
former rector, particles do not mix with each other and a spatially constant concentration
may be assumed within the micron-sized particles. In the second case, small globules (or
droplets) are dispersed and immiscible in a continuous phase without mixing.

⇒ Micro – mixing → micro – fluids (non – segregated fluids) → micro – reactor

1. Ideal CSTR is a good example of micro – reactors.

2. Ideal CSTR cannot be a macro – fluid reactor because residence times are
different for different fluid elements. However, for 𝐴 → 𝐵 (1st order
reaction), an ideal CSTR can be treated as a macro –reactor. In other words,
it makes no difference for first-order kinetics whether one assumes non-
segregated or segregated flow, or micro- or macromixing in an ideal CSTR.
Time of reaction alone and not degree of micromixing determines
conversion. This is expected because a first-order process depends upon
time of reaction of the reacting species and not on the interaction of
molecules (See Carberry).

20
 Design of a non – ideal flow reactor (∈= 0)

1. Segregated flow model (a real/non – ideal reactor is modeled as a segregated or macro

reactor) V
2. CA0 𝑋𝐴 = ?

 Each pocket is a batch reactor (well-mixed) within themselves. 𝑋 = 𝑋(𝑡) is known a

priori for all types of reaction.
 Each pocket has a residense time (some pockets can exit early and some can exit late).

RTD
(We should be able to determine RTD of each pocket or volume – element of the reactor)

Definition (there are two types of RTDs):

volume size x (Δt) = fraction of fluid

J’(t) having residense time between
t & (t + Δt)

0
𝑡 𝑡 + Δ𝑡 time

Δ𝑡

𝐽′(𝑡)
defined in such a way that
J′ (t)δt ≡ fraction of fluid
having RT
t < RT < t + δt
(area under the curve = 1)

δ𝑡 ∞
∫0 𝐽′ (𝑡)𝑑𝑡 = 1
0 𝑡 𝑡 + δ𝑡 time
21
or,
𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 − 𝑏𝑎𝑠𝑖𝑠

1.0
J(t)
𝐼𝑡 𝑖𝑠 𝑐𝑙𝑒𝑎𝑟 𝑡ℎ𝑎𝑡
𝑡

𝐽(𝑡) = ∑ 𝐽′ (𝑡)Δ𝑡
Δ𝐽
𝑜𝑟 𝐽′ (𝑡) =
0 Δ𝑡 }
𝑡1 𝑡𝑖𝑚𝑒, 𝑡

1.0

𝑡
𝐽(𝑡) 𝐽(𝑡) = ∫ 𝐽′ (𝑡)𝑑𝑡
(0 < 𝑡 < 𝑡1 ) 0
𝑑𝐽(𝑡)
𝑜𝑟 𝐽′ (𝑡) =
𝑑𝑡 }

0 𝑡1 time, t

Thus, for a non – ideal reactor: 𝑙𝑖𝑡𝑡𝑙𝑒 𝑝𝑜𝑐𝑘𝑒𝑡𝑠 (𝑏𝑎𝑡𝑐ℎ) 𝑟𝑒𝑎𝑐𝑡𝑜𝑟𝑠

V

v, CAO 𝑋 =?

We require the following: (𝑁𝑜𝑡𝑒: 𝑡 ≡ 𝑅𝑇𝐷 𝑡𝑖𝑚𝑒

≠ 𝑟𝑒𝑎𝑙 𝑡𝑖𝑚𝑒)
1) 𝑋(𝑡) for a batch reactor
2) 𝑅𝑇𝐷 (𝐽 𝑜𝑟 𝐽′ ) of the reactor under consideration

∞
𝑋 = ∫0 𝑋(𝑡)𝐽′ (𝑡)𝑑𝑡 𝑜𝑟 ∑ 𝑋(𝑡)𝐽′ (𝑡)Δ𝑡 : Performance/design equation of a
segregated/macro reactor
(averaged over 𝑅𝑇)

22
 Similarly, any intrinsic property, 𝑟 𝑎𝑡 𝑜𝑢𝑡𝑙𝑒𝑡
∞
= ∫0 𝑟(𝑡)𝐽′ (𝑡)𝑑𝑡

𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑
𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 ℎ𝑎𝑣𝑖𝑛𝑔 𝑅𝑇 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡 & 𝑡 + Δ𝑡
𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦

The average residence time always holds good as follows:

∞
(𝜏) 𝑜𝑟 𝑡 = 𝑅𝑇𝐷 = ∫ 𝑡 𝐽′(𝑡) 𝑑𝑡
0

Determining RTD experimentally (two methods)

(𝑡𝑟𝑎𝑐𝑒𝑟 𝑎𝑛𝑎𝑙𝑦𝑠𝑖𝑠: no reaction)
𝑀 (𝑑𝑜𝑠𝑒/𝑑𝑒𝑙 𝑓𝑛 )
output

𝒐𝒓, 𝐶𝑖𝑛 𝑠𝑡𝑒𝑝 − 𝑖𝑛𝑝𝑢𝑡 𝐴𝑟𝑒𝑎 ≡ 𝑀(𝑑𝑜𝑠𝑒)

time

output

𝑜𝑟 𝐶𝑖𝑛

time

23
 Lecture 03
Another example of developing rate expression for a non-elementary reaction:

Ex. Free radical addition polymerization kinetics

⇒ In general, such reactions include

1) Monomer
2) Initiator
3) Catalyst

, and the mechanistic steps are initiation, propagation, and termination.

𝑘𝑖
a) Initiation: 𝑎𝑀1 + 𝑏𝐼 → 𝑃1
𝑎𝑐𝑡𝑖𝑣𝑒 𝑝𝑜𝑙𝑦𝑚𝑒𝑟 𝑟𝑎𝑑𝑖𝑐𝑎𝑙
𝑚𝑜𝑛𝑜𝑚𝑒𝑟 𝑖𝑛𝑖𝑡𝑖𝑎𝑡𝑜𝑟

(𝑝𝑒𝑟𝑜𝑥𝑖𝑑𝑒, 𝑠𝑢𝑛𝑙𝑖𝑔ℎ𝑡, 𝑈𝑉 𝑠𝑜𝑢𝑟𝑐𝑒, 𝑒𝑡𝑐)

(𝑒𝑔. 𝑒𝑡ℎ𝑦𝑙𝑒𝑛𝑒, 𝑠𝑡𝑦𝑟𝑒𝑛𝑒)

𝑘𝑝𝑟
b) Propagation: 𝑃1 + 𝑀1 → 𝑃2
⋮
𝑘𝑝𝑟
𝑃𝑛−1 + 𝑀1 → 𝑃𝑛
𝑘𝑡
c) Termination: 𝑃𝑛 + 𝑃𝑚 → 𝑀𝑛+𝑚 𝑜𝑟 𝑀𝑛 + 𝑀𝑚 ,
(𝑖𝑛𝑎𝑐𝑡𝑖𝑣𝑒⁄𝑠𝑡𝑎𝑏𝑙𝑒 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠)

The rates of the reactions are developed as

𝑑𝑀1
⇒ = −𝑎𝑟𝑖 − 𝑘𝑝𝑟 𝑀1 ∑ 𝑃𝑛 (𝑟𝑖 =
𝑑𝑡
𝑟𝑎𝑡𝑒 𝑜𝑓 𝑖𝑛𝑖𝑡𝑖𝑎𝑡𝑖𝑜𝑛 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑓𝑜𝑟 𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑟𝑎𝑑𝑖𝑐𝑎𝑙𝑠)
𝑑𝑃1
⇒ = 𝑟𝑖 − 𝑘𝑝𝑟 𝑀1 𝑃1 − 𝑘𝑡 𝑃1 ∑ 𝑃𝑛
𝑑𝑡

⋮ 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑖𝑡𝑠𝑒𝑙𝑓(𝑃1 )

𝑑𝑃𝑛
⇒ = 𝑘𝑝𝑟 𝑀1 𝑃𝑛−1 − 𝑘𝑝𝑟 𝑀1 𝑃𝑛 − 𝑘𝑡 𝑃𝑛 ∑ 𝑃𝑛 (𝒏 ≥ 𝟐)
𝑑𝑡
𝑓𝑟𝑜𝑚 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑡𝑜 𝑓𝑜𝑟𝑚
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑃𝑛+1
11
Make use of PSSA (pseudo steady state assumption) or QSSA:

𝑑𝑃𝑛
≈ 0 (true for all reactions involving the short-lived reactive radical species)
𝑑𝑡

Add ∑𝑛 ∶

0 = 𝑟𝑖 − 𝑘𝑡 (𝑃𝑛 )2
(Thus, under the PSSA, the initiation and termination rates are equal)

𝑑𝑀1 𝑟
Therefore, = −𝑎𝑟𝑖 − 𝑘𝑝𝑟 𝑀1 √( 𝑖 ) (substitute 𝑃𝑛 𝑖𝑛 𝑡ℎ𝑒 1𝑠𝑡 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)
𝑑𝑡 𝑘 𝑡

( usually, the amounts of

𝑟𝑖
= −𝑘𝑝𝑟 𝑀1 √( ) monomer is small
𝑘 𝑡
a ≈ 0)

𝑑𝑀1 𝑘 ( assuming the rate of initiation

= −𝑘𝑝𝑟 √( 𝑖 ) 𝐼 1⁄2 𝑀1 is 1st order, i. e. k i I )
𝑑𝑡 𝑘

rate at which monomer is consumed

or the rate of overall polymerization reaction

See the book by Froment and Bischoff on examples of polyethylene synthesis or thermal cracking
of the long-chain hydrocarbons in sunlight.

Homework (Froment & Bischoff, 2nd ed): 1.11, 1.12, 1.14 (development of rate
expression for the non- elementary reactions)

One more example of developing rate expression for a non-elementary reaction

(All catalytic reactions are non-elementary!)

𝑇=450𝑜 𝐶
𝑒𝑔. 𝐶𝑂 + 𝐻2 𝑂 → 𝐶𝑂2 + 𝐻2 (𝑤𝑎𝑡𝑒𝑟 − 𝑔𝑎𝑠 𝑠ℎ𝑖𝑓𝑡 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)
(𝐹𝑒 − 𝑜𝑥𝑖𝑑𝑒𝑠)

- It is a non-elementary reaction (not exact path of the reaction)

Some introduction to catalyst:

′𝑋′ 𝑎𝑐𝑡𝑖𝑣𝑒 𝑠𝑖𝑡𝑒𝑠 (′𝑋 ′ )

(𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑛𝑖𝑐𝑎𝑙𝑙𝑦 𝑎𝑐𝑡𝑖𝑣𝑒)
12
Mechanistic steps (or elementary steps) are as follows:
𝑘1
𝐻2 𝑂 + 𝑋 → 𝐻2 + 𝑋𝑂
𝑘2 } 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑟𝑦 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛𝑠/𝑠𝑡𝑒𝑝𝑠
𝑋𝑂 + 𝐶𝑂 → 𝐶𝑂2 + 𝑋

𝑑(𝐶𝑂) 𝑑(𝐻2 𝑂) 𝑑(𝐶𝑂2 ) 𝑑(𝐻2 )

1) 𝑟 = − =− = =
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑(𝐶𝑂)
2) = −𝑘2 [𝑋𝑂][𝐶𝑂]
𝑑𝑡
𝑑(𝑋𝑂)
3) = 𝑘1 [𝑋][𝐻2 𝑂] − 𝑘2 [𝑋𝑂][𝐶𝑂]
𝑑𝑡

𝑑(𝑋)
= −𝑘1 [𝑋][𝐻2 𝑂] + 𝑘2 [𝑋𝑂][𝐶𝑂]
𝑑𝑡

𝑑(𝑋𝑂) 𝑑(𝑋)
4) PSSA: = ≈0
𝑑𝑡 𝑑𝑡
𝑘1 [𝑋][𝐻2 𝑂]
[𝑋𝑂] =
𝑘2 [𝐶𝑂]
𝑘2 [𝑋𝑂][𝐶𝑂]
} identical expressions
[𝑋] =
𝑘1 [𝐻2 𝑂]

⇒ All it means is that we need one more independent equation.

Site conservation/balance:

[𝑋] + [𝑋𝑂] = [𝑋𝑜 ]

𝑡𝑜𝑡𝑎𝑙 𝑠𝑖𝑡𝑒𝑠 (𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝑜𝑓
𝑓𝑟𝑒𝑒 𝑠𝑖𝑡𝑒𝑠 𝑜𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑠𝑖𝑡𝑒𝑠 𝑡ℎ𝑒 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙)

(At any time, site is either ′𝑋′ or ′𝑋𝑂′)

[𝑋𝑜 ]𝑘1 𝑘2 [𝐶𝑂][𝐻2 𝑂] 𝑓(𝑇)𝜙(𝐶)

𝑟= ≠
𝑘1 [𝐻2 𝑂]+𝑘2 [𝐶𝑂]
(which is the feature of
an elementary reaction)

Note:

⇒ order of the reaction cannot be not defined here

(𝑒𝑔. 𝑟 ≠ 𝑘1 (𝐶1 )𝛼 (𝐶2 )𝛽 ⋯ )

13
 Lecture 04
Review: Ideal reactors

𝐵𝑎𝑡𝑐ℎ 𝑃𝑙𝑢𝑔 𝑓𝑙𝑜𝑤 𝑀𝑖𝑥𝑒𝑑 𝑡𝑎𝑛𝑘 𝑟𝑒𝑎𝑐𝑡𝑜𝑟 (𝐶𝑆𝑇𝑅)

𝑣𝑜
𝑋𝐴𝑖
𝑣𝑜
𝑋𝐴𝑖 𝑋𝐴𝑓 𝑋𝐴𝑓

𝑝𝑒𝑟𝑓𝑒𝑐𝑡𝑙𝑦 𝑚𝑖𝑥𝑒𝑑; 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑚𝑖𝑥𝑖𝑛𝑔 𝑖𝑛 𝑟 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑠𝑎𝑚𝑒 𝑎𝑠 𝑡ℎ𝑒 𝑏𝑎𝑡𝑐ℎ

𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝑖𝑠 𝑧𝑒𝑟𝑜 𝑚𝑖𝑥𝑖𝑛𝑔 𝑖𝑛 𝑎𝑥𝑖𝑎𝑙 − 𝑑𝑖𝑟𝑒𝑐𝑡𝑜𝑟 } 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠 𝑒𝑥𝑐𝑒𝑝𝑡 𝑡ℎ𝑎𝑡
𝑠𝑝𝑎𝑡𝑖𝑎𝑙𝑙𝑦 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑧 ≠ 𝑣𝑧 (𝑟) 𝑡ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑓𝑙𝑜𝑤.
(𝑇, 𝜌, 𝑐, 𝐶𝑝 )

Design equation of ideal reactors:

𝜈𝐴 𝐴 + 𝜈𝐵 𝐵 + ⋯ → 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑠 ( 𝑟 ≡ 𝑟𝑎𝑡𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

Batch:
𝑋𝐴𝑓 𝑉 V = volume of the reaction
𝑜 𝑑𝑋𝐴
𝑡 = 𝐶𝐴𝑂 ∫𝑋 mixture (t)
𝐴𝑖 𝑉 (−𝜈𝐴 𝑟)
XA = conversion(t)
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒
𝑁𝐴𝑂 − 𝑁𝐴 Vo = volume of the mixture
𝑓𝑜𝑟 𝑡ℎ𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑡𝑜
𝑋𝐴 (𝑡) = at reference time
𝑟𝑒𝑎𝑐ℎ 𝑁𝐴𝑂
NAO = #of moles at reference time
NA = #of moles at time ′t ′

Plug flow reactor:

𝑉 𝑋𝐴𝑓 𝑑𝑋
𝜏= = 𝐶𝐴𝑂 ∫𝑋 𝐴
; vo = volumetric flow rate of
𝑣𝑜 𝐴𝑖 (−𝜈𝐴 𝑟) feed at reference point
(𝑠𝑝𝑎𝑐𝑒 − 𝑡𝑖𝑚𝑒)

𝐹𝐴𝑂 −𝐹𝐴
If flow: 𝑋𝐴 = ( )
𝐹𝐴𝑂
(# of moles⁄time ) ≡ molar flow rate

(Note: space time ≠ residense time because of volume-change)

15
 𝑉 𝐶𝐴𝑂 (𝑋𝐴𝑓 −𝑋𝐴𝑖 )
CSTR: 𝜏= =
𝑣𝑜 (−𝜈𝐴 𝑟)𝑓

𝐝𝐞𝐬𝐢𝐠𝐧: 𝛕(𝐗 𝐀 )
}?
𝐨𝐫 𝐗 𝐀 (𝛕)

If there is no volume – change (i.e., no expansion)

𝑋𝐴𝑓 𝑑𝑋
𝐴
𝑡 = 𝐶𝐴𝑂 ∫𝑋 ∶ 𝑏𝑎𝑡𝑐ℎ 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝐴𝑖 (−𝜈𝐴 𝑟)

𝑋 𝑑𝑋𝐴
𝜏 = 𝐶𝐴𝑂 ∫𝑋 𝐴 ∶ 𝑝𝑙𝑢𝑔 𝑓𝑙𝑜𝑤 reactor (PFR)
𝐴𝑖 (−𝜈𝐴 𝑟)

𝑡=𝜏!
𝑏𝑎𝑡𝑐ℎ 𝑡𝑖𝑚𝑒 = 𝑠𝑝𝑎𝑐𝑒 𝑡𝑖𝑚𝑒.

Notes: (1) If there is no expansion, differential volume in a PFR moves without mixing with the
neighboring elements, i.e., each volume element is like a small ‘batch’ reactor with one
concentration. Therefore, batch time is the same as space time of an ideal PFR.

(2) V (plug flow) << V (CSTR) for the same conversion, if order > +𝒗𝒆, and vice-versa.

𝑔𝑟𝑎𝑑𝑢𝑎𝑙 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒 𝑖𝑛 𝑎𝑏𝑟𝑢𝑝𝑡 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑒

𝑡ℎ𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛

Therefore, 𝐶𝑎𝑣𝑔 ≫ 𝐶𝑒𝑥𝑖𝑡 ,

𝑟𝑃𝐹𝑅 ≫ 𝑟𝐶𝑆𝑇𝑅 ,

𝑉𝑃𝐹𝑅 ≪ 𝑉𝐶𝑆𝑇𝑅 ,

(3)
𝑁𝐴𝑂 −𝑁𝐴 𝐶𝐴𝑂 −𝐶𝐴
without expansion: 𝑋𝐴 = = (𝑉 = 𝑐𝑜𝑛𝑠𝑡. )
𝑁𝐴𝑂 𝐶𝐴𝑂
𝐶𝐴 = 𝐶𝐴𝑂 (1 − 𝑋𝐴 )
𝑁𝐴𝑂 −𝑁𝐴
with expansion: 𝑋𝐴 = : 𝑉 = 𝑉𝑜 (1 + 𝜀𝑋 𝑋𝐴 )
𝑁𝐴𝑂 𝑒𝑥𝑎𝑝𝑎𝑛𝑠𝑖𝑜𝑛
1−𝑋𝐴
𝑒𝑓𝑓𝑒𝑐𝑡
𝐶𝐴 = 𝐶𝐴𝑂 ( )
1+𝜀𝑋𝐴

16
Multiple reactors:
𝑉𝑖
PFR in series: 𝑋
1)𝐴𝑖 𝑋𝐴𝑓
1 2 3
𝑇𝑤𝑜 𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛𝑠
𝑤𝑖𝑙𝑙 𝑔𝑖𝑣𝑒 𝑡ℎ𝑒
𝑉 = ∑ 𝑉𝑖 𝑠𝑎𝑚𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛

𝑉
𝑋𝐴𝑖 𝑋𝐴𝑓

CSTR in series

1 2 3
𝑋𝐴𝑓 (𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛 1)
𝑋𝐴𝑖 𝑋𝐴𝑓 > 𝑋𝐴𝑓 (𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛 2)
𝑓𝑜𝑟 + 𝑣𝑒 𝑜𝑟𝑑𝑒𝑟
𝑓𝑜𝑟 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑣𝑜𝑙𝑢𝑚𝑒.
𝑋𝐴𝑖
𝑉 = ∑ 𝑉𝑖
𝑋𝐴𝑓
𝑉 > ∑ 𝑉𝑖 + 𝑣𝑒 𝑜𝑟𝑑𝑒𝑟

𝑉 < ∑ 𝑉𝑖 − 𝑣𝑒 𝑜𝑟𝑑𝑒𝑟
Parallel configurations: 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛
1) PFR:

𝑉 = ∑ 𝑉𝑖 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛

𝑉𝑖
𝑉 𝑉𝑖
𝑣𝑜 (𝜏 = = )
𝑣𝑜 𝑣𝑜𝑖
𝑎𝑑𝑗𝑢𝑠𝑡 ′𝜏𝑖 ′ 𝑖𝑛 𝑒𝑎𝑐ℎ 𝑟𝑒𝑎𝑐𝑡𝑜𝑟.

2)

𝑉 = ∑ 𝑉𝑖 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛

𝑉𝑖 𝑉 𝑉𝑖
𝑣𝑜 (𝜏 = = )
𝑣𝑜 𝑣𝑜𝑖
𝑎𝑑𝑗𝑢𝑠𝑡 ′𝜏𝑖 ′ 𝑖𝑛 𝑒𝑎𝑐ℎ 𝑟𝑒𝑎𝑐𝑡𝑜𝑟.

17
 Non-ideal reactors

Ideal PFR and Ideal CSTR : two extreme situations of ideality, or two model reactors

𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑚𝑖𝑥𝑖𝑛𝑔
𝑧𝑒𝑟𝑜(𝑎𝑥𝑖𝑎𝑙)𝑚𝑖𝑥𝑖𝑛𝑔

𝑓𝑙𝑜𝑤 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛

Real reactors:

1.

non − uniform concentration

(stagnation zones at corners)
(non − ideality arises from a bad
𝑣 = 𝑣(𝑟) design of mixer or stirrer system)
𝑐 = 𝑐(𝑟)

2. Gas flow over packed bed:

dispersion (mixing) in voids

3. Temperature gradient induces convective flow

(𝑛𝑜𝑛 − 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙) 𝑚𝑖𝑥𝑖𝑛𝑔

4. 𝐴→𝑅 A R it causes
{ dispersion⁄diffusion
in axial (flow) direction
− 𝑙ocal mixing

zero mixing in an ideal PFR

and early⁄late mixing in a non − ideal⁄real reactor

𝑖𝑛ℎ𝑒𝑟𝑒𝑛𝑡𝑙𝑦 𝑝𝑟𝑒𝑠𝑒𝑛𝑡 𝑖𝑛 𝑎 𝑡𝑢𝑏𝑢𝑙𝑎𝑟 𝑟𝑒𝑎𝑐𝑡𝑜𝑟.

18
Micro – mixing ⇒ Mixing of fluid (elements) on molecular level.

𝐴 there is a finite probability of

A & B mixing somewhere in the reactor.
B

Macro - mixing ⇒
mixing of pocket − zones
(little volume)

Homework (Froment & Bischoff, 2nd ed): 12.5 (only 1st order)

19
 Lecture 05
- Continue…….. ( non-ideal reactors)

Concept of “pockets” and criteria for macro-mixing:

1) Contents of individual pockets are well mixed with themselves, but do not mix with
second pockets.
2) Pockets should represent chemical reaction conditions. For 𝐴 + 𝐵 → 𝐶, one cannot,
therefore, reduce the size of a pocket to one molecule; then the reaction will not occur.
Pockets should have a finite volume.

3) Pocket-contents should have same residence time, but different pockets may have

different residence times.

⇒ Macro- mixing → macro – fluids (segregated fluids) → macro - reactor

Ideal PFR is a good example of macro – reactors

𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑝𝑜𝑐𝑘𝑒𝑡𝑠
(𝑠𝑎𝑡𝑖𝑠𝑓𝑦 𝑐𝑟𝑖𝑡𝑒𝑟𝑖𝑎)

 Fast fluidized bed reactor and liquid-liquid emulsion (LLE) membrane-based reactive
continuous separation processes are examples of industrial or practical reactors. In the
former rector, particles do not mix with each other and a spatially constant concentration
may be assumed within the micron-sized particles. In the second case, small globules (or
droplets) are dispersed and immiscible in a continuous phase without mixing.

⇒ Micro – mixing → micro – fluids (non – segregated fluids) → micro – reactor

1. Ideal CSTR is a good example of micro – reactors.

2. Ideal CSTR cannot be a macro – fluid reactor because residence times are
different for different fluid elements. However, for 𝐴 → 𝐵 (1st order
reaction), an ideal CSTR can be treated as a macro –reactor. In other words,
it makes no difference for first-order kinetics whether one assumes non-
segregated or segregated flow, or micro- or macromixing in an ideal CSTR.
Time of reaction alone and not degree of micromixing determines
conversion. This is expected because a first-order process depends upon
time of reaction of the reacting species and not on the interaction of
molecules (See Carberry).

20
 Design of a non – ideal flow reactor (∈= 0)

1. Segregated flow model (a real/non – ideal reactor is modeled as a segregated or macro

reactor) V
2. CA0 𝑋𝐴 = ?

 Each pocket is a batch reactor (well-mixed) within themselves. 𝑋 = 𝑋(𝑡) is known a

priori for all types of reaction.
 Each pocket has a residense time (some pockets can exit early and some can exit late).

RTD
(We should be able to determine RTD of each pocket or volume – element of the reactor)

Definition (there are two types of RTDs):

volume size x (Δt) = fraction of fluid

J’(t) having residense time between
t & (t + Δt)

0
𝑡 𝑡 + Δ𝑡 time

Δ𝑡

𝐽′(𝑡)
defined in such a way that
J′ (t)δt ≡ fraction of fluid
having RT
t < RT < t + δt
(area under the curve = 1)

δ𝑡 ∞
∫0 𝐽′ (𝑡)𝑑𝑡 = 1
0 𝑡 𝑡 + δ𝑡 time
21
or,
𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 − 𝑏𝑎𝑠𝑖𝑠

1.0
J(t)
𝐼𝑡 𝑖𝑠 𝑐𝑙𝑒𝑎𝑟 𝑡ℎ𝑎𝑡
𝑡

𝐽(𝑡) = ∑ 𝐽′ (𝑡)Δ𝑡
Δ𝐽
𝑜𝑟 𝐽′ (𝑡) =
0 Δ𝑡 }
𝑡1 𝑡𝑖𝑚𝑒, 𝑡

1.0

𝑡
𝐽(𝑡) 𝐽(𝑡) = ∫ 𝐽′ (𝑡)𝑑𝑡
(0 < 𝑡 < 𝑡1 ) 0
𝑑𝐽(𝑡)
𝑜𝑟 𝐽′ (𝑡) =
𝑑𝑡 }

0 𝑡1 time, t

Thus, for a non – ideal reactor: 𝑙𝑖𝑡𝑡𝑙𝑒 𝑝𝑜𝑐𝑘𝑒𝑡𝑠 (𝑏𝑎𝑡𝑐ℎ) 𝑟𝑒𝑎𝑐𝑡𝑜𝑟𝑠

V

v, CAO 𝑋 =?

We require the following: (𝑁𝑜𝑡𝑒: 𝑡 ≡ 𝑅𝑇𝐷 𝑡𝑖𝑚𝑒

≠ 𝑟𝑒𝑎𝑙 𝑡𝑖𝑚𝑒)
1) 𝑋(𝑡) for a batch reactor
2) 𝑅𝑇𝐷 (𝐽 𝑜𝑟 𝐽′ ) of the reactor under consideration

∞
𝑋 = ∫0 𝑋(𝑡)𝐽′ (𝑡)𝑑𝑡 𝑜𝑟 ∑ 𝑋(𝑡)𝐽′ (𝑡)Δ𝑡 : Performance/design equation of a
segregated/macro reactor
(averaged over 𝑅𝑇)

22
 Similarly, any intrinsic property, 𝑟 𝑎𝑡 𝑜𝑢𝑡𝑙𝑒𝑡
∞
= ∫0 𝑟(𝑡)𝐽′ (𝑡)𝑑𝑡

𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑
𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 ℎ𝑎𝑣𝑖𝑛𝑔 𝑅𝑇 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡 & 𝑡 + Δ𝑡
𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦

The average residence time always holds good as follows:

∞
(𝜏) 𝑜𝑟 𝑡 = 𝑅𝑇𝐷 = ∫ 𝑡 𝐽′(𝑡) 𝑑𝑡
0

Determining RTD experimentally (two methods)

(𝑡𝑟𝑎𝑐𝑒𝑟 𝑎𝑛𝑎𝑙𝑦𝑠𝑖𝑠: no reaction)
𝑀 (𝑑𝑜𝑠𝑒/𝑑𝑒𝑙 𝑓𝑛 )
output

𝒐𝒓, 𝐶𝑖𝑛 𝑠𝑡𝑒𝑝 − 𝑖𝑛𝑝𝑢𝑡 𝐴𝑟𝑒𝑎 ≡ 𝑀(𝑑𝑜𝑠𝑒)

time

output

𝑜𝑟 𝐶𝑖𝑛

time

23
 Either of the two methods of injecting a tracer will work. It is a question of convenience in the
experiment. ′𝑄′, flow rate must be the same as that used in the real reactor; however, without
reaction (i.e., reaction is switched off !)

Method 1: Step Tracer Injection

Input Output
𝑣 (𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒)
𝐶𝑖𝑛
𝐶𝑖𝑛
V
𝐶𝑜𝑢𝑡 (𝑡)
𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒
0 𝐶𝑜𝑢𝑡 (𝑡)
𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑
ℎ𝑎𝑣𝑖𝑛𝑔 𝑅𝑇 ( 0, 𝑡)
𝑪𝒂𝒍𝒄𝒖𝒍𝒂𝒕𝒊𝒐𝒏𝒔:
0 𝑡 𝑡𝑖𝑚𝑒
By the very definition of 𝐽(𝑡):

𝐶𝑜𝑢𝑡 (𝑡)
𝐽(𝑡) = ⇒ J’(t) = d(J(t))/dt
𝐶𝑖𝑛

Now, turn on the reaction (no tracer!)

∞ ∞
𝑋 = ∫0 𝑋(𝑡)𝐽′ (𝑡)𝑑𝑡 = ∫0 𝑋(𝑡)𝑑(𝐽(𝑡))

𝐵𝑎𝑡𝑐ℎ 𝐵𝑎𝑡𝑐ℎ

Method 2: Pulse Injection

Input

𝑣 (𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒) 𝐶𝑒𝑥𝑖𝑡 (𝑡)

𝑀 = 𝑡𝑜𝑡𝑎𝑙 𝑎𝑚𝑜𝑢𝑛𝑡 𝑜𝑓 V
𝑡𝑟𝑎𝑐𝑒𝑟 (𝑑𝑜𝑠𝑒) 𝑣
′𝑑𝑒𝑙𝑡𝑎′
𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛

0 𝑡𝑖𝑚𝑒

24
 𝐶𝑒𝑥𝑖𝑡 (𝑡)

Output

t (t + 𝜟t)

Species balance (use definition of 𝐽′(𝑡))

𝑀𝐽′ (𝑡)𝑑𝑡 = 𝐶𝑒𝑥𝑖𝑡 (𝑡)𝑣𝑑𝑡

𝐶𝑒𝑥𝑖𝑡 𝑣 𝐶𝑒𝑥𝑖𝑡 (𝑡) 𝐶𝑒𝑥𝑖𝑡 (𝑡) 𝐶𝑒𝑥𝑖𝑡 (𝑡)
𝐽′ (𝑡) = = 𝑀⁄ = ∞ = ∞ Type equation here.
𝑀 𝑣 ∫0 𝑣𝐶𝑒𝑥𝑖𝑡 (𝑡)𝑑𝑡 ⁄𝑣 ∫0 𝐶𝑒𝑥𝑖𝑡 𝑑𝑡

∞
J(t) = ∫0 𝐽′(𝑡)𝑑𝑡

Again, turn on the reaction (no tracer)

∞ ∞
𝑋 = ∫0 𝑋(𝑡)𝐽′ (𝑡)𝑑𝑡 = ∫0 𝑋(𝑡)𝑑(𝐽(𝑡))

Either of the two methods will work: by determining J(t), J’(t) can be determined, or vice-versa.
∞
Also, be careful in evaluating or calculating such integrals: ∫0 𝑋(𝑡)𝐽′(𝑡)𝑑𝑡, which contain two
different functions that may span over different time-domains, thus having different limits of
integrations. See the examples and questions in the examinations that follow in the next lectures.

25
 15

𝐽′ (𝑡)

3 5

(𝑁𝑡⁄𝜏)

(Also note that J′(t) = d(J(t))/dt ≡ same as analytically derived)

Model: Non-ideal (mixed) reactor of volume V ≡ ′ 𝑁 ′ ideal CSTR of volume 𝑉 ⁄𝑁
Once ‘N’ is known, determine ‘X’ for ‘N’ ideal CSTR reactors (which have been discussed at the
UG level).
𝐶𝑜
𝑣, 𝐴 → Products
1 2 3 𝑛 𝑁

Species balance (SS): (𝐶𝑛−1 − 𝐶𝑛 )𝑣 = 𝑉 ⁄𝑁 (−𝑣𝐴 𝑟)

Reaction rate
𝑛 = 0, 𝐶𝐴 = 𝐶𝑜

1st order reaction,

𝜏
𝐶𝑛−1 − 𝐶𝑛 = 𝑘 𝐶𝑛
𝑁

𝑘𝜏
[(1 + ) 𝐸 − 1] 𝐶𝑛 = 0 (E is the shift operator)
𝑁

𝑛
1
𝐶𝑛 = 𝐶𝑜 ( 𝑘𝜏 )
1+
𝑁

𝐶𝑜 𝑘𝜏 𝑛
= (1 + ) ; 𝐶𝑁 = 𝐶𝑜 (1 − 𝑋𝐴 )
𝐶𝑛 𝑁

For the 2nd or other order of reaction → use numerical technique.

HW: (SMITH, 3rd ed) ≡ 6.3, 6.7, 6.15

38
 ∞ ∞
(c) 𝑋 = ∫0 𝑋(𝜃)𝐽′ (𝜃)𝑑𝜃 = ∫0 (1 − 𝑒−𝑘𝜃 )𝐽′ (𝜃)𝑑𝜃
(𝑠𝑒𝑔𝑟𝑒𝑔𝑎𝑡𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟) 1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟/𝑏𝑎𝑡𝑐ℎ 𝑟𝑒𝑎𝑐𝑡𝑜𝑟

∞
= ∫0 (1 − 𝑒−𝑘𝜃 )(0.92𝑒−𝜃 + 0.04𝑒−0.5𝜃 )𝑑𝜃 = 0.677

This must also be the same as calculated below:

As per the problem statement, the real reactor is a combination of two ideal CSTRs in parallel:

𝑄1 Therefore;
0.85V
𝑋1 for a 1st order rate–reactor in an ideal CSTR
𝑄
𝑄 𝑘 𝜏1
= (𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛)
𝑋 =? 1+𝑘 𝜏1
𝑄2
0.15V
2×1 2
= =
1+2×1 3
𝑘 𝜏2 2×2 4
𝑋2 = = =
1+𝑘 𝜏2 1+2×2 5

𝑋1 𝑄1 +𝑋2 𝑄2 2⁄ ×0.85𝑉+4⁄ ×0.075𝑉

3 5
Therefore, 𝑋 = =
𝑄1 +𝑄2 0.85𝑉+0.075𝑉

0.567+0.06
= = 0.677
0.925

(Thus, RTD calculation/data produce the same result as that from design equation)

Ex 6: Coal particles are devolatized in a fast fluidized/moving bed reactor at 450𝑜 𝐶, using 𝑁2 gas.
The average residense time of the well-mixed particles, 𝜏 = 10 min. Initial volatile contents are
46% (𝑤⁄𝑤 ). The devolatization rate follows 1st order rate kinetics:
𝑑𝑋
= −𝑘(𝑋 − 0.2); 𝑋 ≡ 𝑚𝑎𝑠𝑠 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑣𝑜𝑙𝑎𝑡𝑖𝑙𝑒𝑠.
𝑑𝑡

Determine 𝑋 (the average volatile content of the coal particles) at the reactor –exit. Assume k
= 1 min-1.

(Note: 20% of the volatiles never comes out).

33
Ans: In such case (g/l + S reaction)/reactors, each particle can be considered to be segregated.
There is no concentration distribution within a particle and particles do not mix with each other.
The reactor can be considered to be macro or segregated reactor. Therefore,
∞
𝑋 = ∫0 𝑋(𝑡)𝐽′ (𝑡)𝑑𝑡

(𝑠𝑒𝑔𝑟𝑒𝑔𝑎𝑡𝑒𝑑 𝑚𝑜𝑑𝑒𝑙)

1
Ans: 𝐽′ (𝜃) = 𝑒 −𝑡⁄10
10
(ideal/well -mixed reactor)

∞ 1 𝑑𝑋
𝑋 = ∫0 (0.2 + 0.26𝑒 −𝑘𝑡 ) 𝑥 𝑒 −𝑡⁄10 𝑑𝑡 2) = −𝑘 (𝑋 − 0.2): 𝑏𝑎𝑡𝑐ℎ 𝑘𝑖𝑛𝑒𝑡𝑖𝑐𝑠
10 𝑑𝑡
1 ∞
= 10 ∫0 (0.2𝑒 −𝑡⁄10 + 0.26𝑒 −1.1𝑡
)𝑑𝑡 𝑙𝑛 (𝑋 − 0.2)|𝑋𝑋𝑜=0.46 = −𝑘𝑡|𝑡0
1 ∞ 0.26 ∞ 𝑋−0.2
= [0.2 × 10𝑒 −𝑡⁄10 |0 − × 𝑒 −1.1𝑡 | ] 𝑙𝑛 (0.46−0.2) = −𝑘𝑡
10 1.1 0
1 0.26
= [2 + ] = 0.223 (𝑜𝑟 22.3%) 𝑋 = 0.20 + 0.26𝑒 −𝑘𝑡
10 1.1
(batch reactor)

𝑁𝑜𝑡𝑒: 𝑡 = 0 , 𝑋 = 0.46
( )
𝑡 → ∞ , 𝑋 → 0.20

HW: (SMITH, 3rd ed) ≡ 6.1, 6.2

34
 Use Laplace transformation ʆ𝐶𝑛 (𝑡) = 𝐶 (𝑠)
𝑉
𝑣(𝐶𝑛−1 − 𝐶𝑛 ) = 𝑠𝐶𝑛 (at the initial condition functional value is zero)
𝑁

𝜏 𝑉
𝑜𝑟 (1 + 𝑠 𝑁) 𝐶𝑛 − 𝐶𝑛−1 = 0 (𝜏 = 𝑣 )

𝜏
𝐶1 = 𝐶0 ⁄(1 + 𝑠 )
𝑁

𝜏 𝜏 2
𝐶2 = 𝐶1 ⁄(1 + 𝑠 ) = 𝐶0 ⁄(1 + 𝑠 )
𝑁 𝑁

𝜏 𝑛
𝐶𝑛 = 𝐶0 ⁄(1 + 𝑠 )
𝑁

𝑁 𝑛
𝜏 𝑛 𝐶0 ( )
𝐶𝑛 (𝑠) = 𝐶0 ⁄(1 + 𝑠 ) ⇒ 𝐶𝑛 = ʆ−1 [ 𝜏
𝑁 𝑛
]
𝑁 𝑠( 𝑠+ )
𝜏

𝑁 𝑛
𝐶0 ( )
𝐶𝑛 = ʆ−1 [ 𝜏
𝑁 𝑛
]
𝑠( 𝑠+ )
𝜏

𝐶𝑛 −
𝑁𝑡
𝑁𝑡 1 𝑁𝑡 2 1 𝑁𝑡 𝑛−1
𝐽(𝑡) ≡ =1−𝑒 𝜏 [1 + + ( 𝜏 ) + ⋯ + 𝑛−1! ( 𝜏 ) ]
𝐶0 𝜏 2!

See the textbooks and draw qualitatively but accurately:

Input:

𝐶𝑛 x ≡ expt tracer data

𝐽(𝑡) = for the real reactor
𝐶0
model lines =
different values for 𝑛.

(𝑁𝑡⁄𝜏)

37
 𝜕𝐶 𝜕𝐶
𝐶𝐿− = 𝐶𝐿+ ⇒ | = | = 0 (gradient is flat)
𝜕𝑧 𝐿− 𝜕𝑧 𝐿+

On non-dimensionalization,
1 𝜕 2𝜙 𝜕𝜙 (−𝑣𝐴 𝑟)𝜏
− = 𝑅𝑎 (𝑅𝑎 = )
𝑃𝑒 𝜕𝜂 2 𝜕𝜂 𝐶𝑜

𝜕𝜙 1
BCs 𝜂 = 0,
𝜕𝜂 𝑃𝑒
=𝜙−1

𝜕𝜙
= 1, =0
𝜕𝜂

For 1st order reaction,

𝑃
𝐶𝑓 4𝑎 𝑒𝑥𝑝 ( 𝑒⁄2)
=
𝐶𝑜 (1 + 𝑎)2 𝑒𝑥𝑝 (𝑎𝑃𝑒 ) − (1 − 𝑎)2 𝑒𝑥𝑝 (− 𝑎𝑃𝑒 )
2 2

4𝑘𝜏
where, 𝑎 = √1 +
𝑃𝑒

For the other reaction orders, numerical technique is required to solve the equation.

𝐈𝐧 𝐬𝐮𝐦𝐦𝐚𝐫𝐲:
non − ideal reactor (1) Find J(t) for DL or Pe
C
(2)Determine X (1 − ) from the performance equation.
Co
Note: 1. Numerous empirical equations are available for packed bed reactors to determine 𝐷𝐿 =
𝐷𝐿 (𝑅𝑒 , 𝑆𝑐), or under turbulent flow conditions, 𝑃𝑒,𝐿 = 𝑃𝑒,𝐿 (𝑅𝑒 , 𝑆𝑐 ).

2. In a laminar flow there is no axial dispersion/”eddy”.

2𝑛𝑑 3𝑟𝑑 4𝑡ℎ

𝐷𝐿 1𝑠𝑡
turbulent
𝐷𝑚
?? DL ∝ R e
𝐷𝐿 > 𝐷𝑚
(eddies)
(correlations are available)
2100
𝑢𝑑
< 10 𝑅𝑒 =
ν 42
Explain the 2nd phase in the above-figure: why does axial dispersion coefficient exceed molecular
diffusion coefficient even if Re < 2100?

u 2 d2 (not because of eddy but because of

DL = Dm +
192Dm laminar flow profile)

Aris − Taylor diffusion

coefficient
𝑟2
𝑢𝑧 = 𝑢𝑜 (1 − 2 )
𝑅

Species balance:

𝜕𝐶 𝑟2 𝜕𝐶 𝜕 2𝐶 𝐷𝑚 𝜕 𝜕𝐶
+ 2𝑢 (1 − ) = 𝐷𝑚 + (𝑟 𝜕𝑟 ); 𝑅𝑒 < 2100
𝜕𝑡 𝑅 2 𝜕𝑧 𝜕𝑧 2 𝑟 𝜕𝑟

𝑡=0 𝑐=0

0+ 𝐶 = 𝐶𝑖𝑛 @ 𝑧 = 0
𝜕𝐶
=0 @ 𝑧=𝐿
𝜕𝑧

𝐶(𝑡, 𝑟, 𝑧) → C(t, z)
averaged over ′r′

From 2D to 1D model:

It can be shown that

𝜕𝐶 𝜕𝐶 𝜕 2𝐶
+𝑢 = 𝐷𝐿 ; 𝐶 (𝑡, 𝑧)
𝜕𝑡 𝜕𝑧 𝜕𝑧 2 𝑢
+ ′𝐷𝐿 ′

𝑢2 𝑑 2
where, 𝐷𝐿 = 𝐷𝑚 +
192𝐷𝑚

extra term arising out of 1D approximation

(See Dean’s book on Transport Phenomena)

43
RTD Study at a glance: It is an approach or a model to characterize or understand extent of mixing
in a non-ideal continuous flow reactor, non-ideality arising out of either less than fully
segregation or completely non-segregation. Thus, there are two model ideal reactors as far as
the effect of mixing on reaction conversion is concerned: an ideal CSTR or an ideal PFR; all real
reactors stand in between. Therefore, “ideal reactors” imply that different fluid elements or
pockets or zones are either completely mixed, non-segregated, dispersed, micro-mixed, or
earliest mixed like in a well-mixed CSTR, or completely un-mixed, segregated, zero-dispersed,
macro-mixed or late mixed like in a PFR without diffusion and dispersion. RTD measurements
(either step input or pulse input) can be used to characterize or describe the extent or state of
mixing in a real (or non-ideal) reactor, with an objective of predicting or calculating reaction
conversion at the exit of the reactor.

If you have enough evidence (experimental or intuition or experience) to judge that a real reactor
under consideration is segregated like in an ideal PFR or laminar tubular flow reactor, or fast
fluidized gas/liquid-solid reactors or (immiscible) liquid-liquid membrane or extraction systems,
the RTD data will suffice to calculate the conversion, and the performance or design or model
equation of such non-ideal reactors is based on the RTD data (one parameter only) and the batch
kinetics. However, if there is some element or extent of mixing, the RTD measurements alone
will not suffice to describe the extent of mixing in such non-ideal reactors, and additional
information is required: “where do they (different pockets) mix and at what rate do they mix”.
Thus, there are two popular models: “tank-in-series-model” with the number of ideal CSTRs as
the model parameter and “Dispersion model” with peclet or dispersion number as the model
parameter. Once the RTD and the model parameter are known, reaction conversion can be
calculated for such non-ideal (non-segregated) reactors from the first principle (species balance
equation). First-order reaction is an exception, and the RTD data (with batch kinetics) alone are
enough to calculate or predict the conversion in non-segregated reactors.

H.W. (F&B, 2nd ed) = 12.5

44
 Lecture 10
Catalysts and Catalytic reactions:

 What are catalysts? These are physical entities; show chemical/physical activities of
deactivation, degradation, change in morphology, and have life-time.
 What do they do? Enhance (+ve)/retard (-ve) rate of reaction by participating at the
molecular or electronic levels, yet preserving/restoring their initial activities.
 How do they do? Direct alternate path (route/step) of reaction that has a lower activation
energy barrier:

𝑤⁄𝑜 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
1. The barrier has
𝐴 𝐵 come down using
catalyst
𝑎𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑒 𝑝𝑎𝑡ℎ𝑤𝑎𝑦 2. There are two
barriers: forward
𝐸𝐸
11
𝑤𝑖𝑡ℎ 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 & reverse reactions
𝐸−1 (E1 & E−1 )
𝐴 3. H1 & H2 or ΔH
Δ𝐻 is the same
𝐵
with/without
𝐻1 𝐻2 catalyst
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝐶𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑒

⇒ Catalyst enhances reaction rate or kinetics (k) of an elementary reaction without affecting
thermodynamic properties including heat of reaction & equilibrium constant (K).

⇒ No reaction is truly irreversible.

⇒ Δ𝐻 can be +𝑣𝑒 𝑜𝑟 − 𝑣𝑒 (endothermic or exothermic)

𝑤⁄𝑜 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
𝑤𝑖𝑡ℎ 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
𝐸−1

𝐸1
𝐵 (𝐻2 − 𝐻1 = Δ𝐻 > 0)
𝐴
𝐻1 𝐻2 (+𝑣𝑒)
𝑒𝑛𝑑𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐

45
 𝜕𝐶
𝑡 = 0+ , 𝑧 = 0, 𝑢𝐶𝑜 = (𝑢𝐶 − 𝐷𝐿 )|
𝜕𝑧 𝑧=0+

𝜕𝐶
= 𝐿, = 0 (long tube − approximation)
𝜕𝑧

(Note: DL ≡ dispersion coefficient of tracer)

Non-dimensionalize:

𝜙 = 𝐶 ⁄𝐶𝑜 , 𝜂 = 𝑧⁄𝐿 , 𝑡 = 𝑡⁄𝜏 , 𝑜𝑟 𝑡⁄𝐿⁄𝑢

𝐷𝐿 𝜕 2 𝜙 𝜕𝜙 𝜕𝜙
− =
𝑢𝐿 𝜕𝜂 2 𝜕𝜂 𝜕𝑡

i.e., 𝑡 = 0, 𝜙=0
𝜕𝜙 uL
𝐵𝐶 1 𝜂=0 1⁄𝑃𝑒 =𝜙−1 (Pe = )
𝜕𝜂 DL
𝜕𝜙 DL
2 =1 =0 ( is called dispersion #)
𝜕𝜂 uL
uL convection
Recall: 𝑃𝑒 (𝑎𝑥𝑖𝑎𝑙 𝑝𝑒𝑐𝑙𝑒𝑡 #) = ( )
DL dispersion

large value implies small dispersion small value implies large dispersion
(𝐶𝑆𝑇𝑅)
(𝑃𝐹𝑅)

(In general, 𝑃𝑒 > 10 ∶ 𝑃𝐹𝑅 ; < 1 ∶ CSTR)

𝐒𝐨𝐥𝐯𝐞 𝐭𝐨 𝐠𝐞𝐭 𝑪⁄𝑪𝒐 𝒐𝒓 𝝓(𝒕) 𝒐𝒓 𝑱(𝒕) 𝒏𝒖𝒎𝒆𝒓𝒊𝒄𝒂𝒍𝒍𝒚 :

𝐑𝐓𝐃

1.0 5
3

𝐽(𝑡)
𝑃𝑒 = 10
3 5
𝑃𝑒 → ∞ (𝑃𝐹𝑅)

40
 (7) Reforming: Butane→ 𝐼𝑠𝑜𝑏𝑢𝑡𝑎𝑛𝑒 (𝑝𝑜𝑙𝑦𝑚𝑒𝑟, 𝑝𝑙𝑎𝑠𝑡𝑖𝑐𝑠) (1950)
𝑀𝑜2 𝑂3 /𝐴𝑙2 𝑂3

𝑃𝑡/𝐴𝑙2 𝑂3

(8) Hydrocracking (1960) ⇒ Shell

(catalyst mixed with 𝐻2 to prevent formation or deposition on catalysts)

(9) Catalytic convertors (60 − 70𝑠 ) in automobile exhaust

- Very complicated design (unsteady-state conditions of speed, temperature, species, no
control)
1 𝐶𝑂, 𝐶𝑋 𝐻𝑦 , 𝑁𝑂 (𝑝𝑜𝑙𝑙𝑢𝑡𝑎𝑛𝑡𝑠)
- 𝐶𝑂 + 𝑂2 → 𝐶𝑂2
2
1
𝐶𝑋 𝐻𝑦 + 𝑂2 → 𝐶𝑂2 + 𝐻2 𝑂 𝑃𝑡/𝑅ℎ 𝑜𝑛 𝐴𝑙2 𝑂3 /𝑆𝑖𝑂2
2
2𝑁𝑂 + 𝑂2 → 2𝑁𝑂2
𝐵𝑎𝑢𝑥𝑖𝑡𝑒
(𝑁𝑂 + 𝐶𝑂 → 𝐶𝑂2 + 𝑁2 )
𝑂2 𝑑𝑖𝑠𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛𝑎𝑙 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
- Modern vehicles have three-ways functional catalysts

Mechanism (pathway):

Electronic (sites are electronically active); geometrical theory

all of these involve (molecular dimensions/configuration)
several electron Chemical theory (surface complex formation)
transfer steps, complex ↓
bonds formation underlying solid surface should accomodate
{ (A → A∗ )

- All these different theories & mechanism to help us drive a rate expression for the
viewpoint of designing the reactor:
𝐫 = 𝐟(𝐓, 𝐂)
𝐍𝐨𝐭𝐞𝐬:
1 1 𝑑𝑁𝑖 𝑆 = 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦
1. 𝑅𝑒𝑣𝑖𝑠𝑖𝑡: 𝑟 = 𝜈
𝑖 𝑆 𝑑𝑡 (𝐴, 𝑊, 𝑉) 𝐴 (𝑔𝑎𝑠)
2. Collision on molecular level.
(It is obvious that transport is not a right step) 𝑠𝑜𝑙𝑖𝑑
3. Unfortunately, there are transport steps from gas to solid for reactant before reaction, and from
solid to gas for products after reaction:
𝑃

(Intitutively, the transport of A or P retards the overall rate of reaction)

47
 4. Reaction may be isothermal or temp gradient may exist; can be exothermic or endothermic.

𝐵 𝑠𝑜𝑙𝑖𝑑 (𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡)
𝐴
𝑔𝑎𝑠

𝑝𝑜𝑟𝑒𝑠
𝐵
(𝐴 → Product)

a. Bulk transport of ′ A′ towards solids (convection, diffusion)

(𝑉. 𝛻𝐶 𝐷∇2 𝐶)

𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡
b. Interphase transport (film transport)
𝑠𝑡𝑒𝑝𝑠
𝑘𝑚
(diffusion)
𝐵 (𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙 𝑠𝑢𝑟𝑓𝑎𝑐𝑒)

(𝑨 encounters diffusion resistances) k m 𝛥C (−Dm 𝛻𝐶)

c. Intraphase transport (within pores, pore diffusion )

{ (reaction may take place parallel) Dpore ∇2 C

𝐴
𝑠𝑢𝑟𝑓𝑎𝑐𝑒

𝑑. Surface reaction (P is produced)

𝑒. 𝑐
𝑓. } ≡ 𝑏} for product P (transport steps)
𝑔. 𝑎

surface reaction (𝑝𝑟𝑒𝑐𝑒𝑑𝑒𝑑 𝑎𝑛𝑑 𝑓𝑜𝑙𝑙𝑜𝑤𝑒𝑑) 𝑏𝑦 𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛/𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛

𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛

′𝑋′
𝑎𝑐𝑡𝑖𝑣𝑒 𝑠𝑖𝑡𝑒 𝑙𝑎𝑠𝑡 𝑠𝑡𝑒𝑝
surface intermediate/
𝑓𝑖𝑟𝑠𝑡 𝑠𝑡𝑒𝑝
complex formation
(several possibilities)

48
Note: Determine 𝑃𝑒 by fitting the RTD data with the model predictions. For small 𝐷𝐿 , or large 𝑃𝑒
(not very large), analytical solutions are available.

1 1−𝑡
𝐽(𝑡) = [1 − 𝑒𝑟𝑓 ( ) √𝑃𝑒 ]
2 2√𝑡

2 𝑥 2
(erf(𝑥) = ∫ 𝑒 −𝑥 𝑑𝑥)
𝜋 0
√

1 𝑃𝑒 𝑃𝑒 2
𝐽′ (𝑡) = √ exp(− (1 − 𝑡) )
2 𝜋 4

If you do a dose-experiment:

𝑃𝑒 =10

𝐽′(𝑡)

3 5

Note that in both cases (J or J’), the responses will be the same as before, i.e., in mixed tanks in series
model.

Once 𝑃𝑒 or 𝐷𝐿 is known, determine 𝑋

𝜕𝐶 𝜕𝐶 𝜕 2𝐶
+𝑢 = 𝐷𝐿 + (−𝑣𝐴 𝑟)
𝜕𝑡 𝜕𝑧 𝜕𝑧 2
𝑆𝑆 𝜕 2𝐶 𝜕𝐶
𝐷𝐿 −𝑢 + (−𝑣𝐴 𝑟) = 0
𝜕𝑧 2 𝜕𝑧

𝜕𝐶
BCs 𝑧=0 𝑢𝐶𝑜 = (𝑢𝐶 − 𝐷𝐿 )|
𝜕𝑧 𝑧=0+

(𝐶 = 𝐶𝑜 if no gradient (mixing) on 𝑧 = 0+ side)

∂C ∂C
(A general interfacial boundary condition: (uC − DL ∂z )| = (uC − DL ∂z )|
L− L+

At the exit, continuity in concentration:

41
 𝑁𝑜 𝑖𝑛𝑡𝑒𝑟𝑎𝑐𝑡𝑖𝑜𝑛𝑠 between adjacent molecules:

𝐸, Δ𝐻 ≠ 𝑓(𝜃)

𝑝 1⁄
⁄𝑉 𝑉𝑚
1⁄
K𝑉𝑚

(2) Brunauer, Emmett and Teller (1938)

BET isotherm

Assumptions - Multilayer adsorption

- Energy of 1st layer ≠ that of 2nd layer; 2nd and the other layers have the
same energy level.
- We do not assume that 1st layer is saturated & then 2nd layer is formed
- Finite probability for sitting of one molecule on the vacant or occupied
sites.
𝑉 𝑐𝑃
Isotherm: 𝜃 = = 𝑝
𝑉𝑚 (𝑝𝑜 −𝑝)[1+(𝑐−1) ]
𝑝𝑜

𝑉 = 𝑉𝑜𝑙𝑢𝑚𝑒 𝑎𝑑𝑠𝑜𝑟𝑏𝑒𝑑; 𝑉𝑚 = 𝑉𝑜𝑙𝑢𝑚𝑒 𝑎𝑑𝑠𝑜𝑟𝑏𝑒𝑑 𝑓𝑜𝑟 𝑜𝑛𝑒 𝑙𝑎𝑦𝑒𝑟

𝑐 = 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟 𝑜𝑓 𝑚𝑜𝑑𝑒𝑙
(energy of the first layer)
= = 1 if the layers have the same energy.
(energy of the other layers)

If c ≫ 1 (very energetic surface)

T = constant and p ≪ po
𝑇𝑦𝑝𝑒 − c(p⁄po )
𝜃 θ=
1 + (p⁄po )
1𝑠𝑡 Kp
= (identical to Langmuir model)
𝑙𝑎𝑦𝑒𝑟 1 + Kp

𝑝 𝑝𝑜
Linearize: (𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛) (𝑐−1)
𝑝 𝑆= 𝑐𝑉𝑚
𝑝 1 (𝑐−1)𝑝 𝑉(𝑝𝑜 −𝑝)
= +
𝑉(𝑝𝑜 −𝑝) 𝑉𝑚 𝐶 𝑐𝑉𝑚 𝑝𝑜 1
=𝐼 𝑝
𝑐𝑉𝑚 𝑝𝑜 50
 Data:

𝑉
𝑝

(3) Temkin Model:

𝑟𝑎 = 𝑘𝑎 𝑝 ∶ 𝑟𝑑 = 𝑘𝑑 𝜃 = (𝑘𝑑𝑜 𝑒 −𝐸𝑑 ⁄𝑅𝑇 )𝜃
{ (𝐸𝑑 = 𝐸𝑑𝑜 − 𝑏𝜃 ; 𝜃 ↑ 𝐸𝑑 ↓)
= 𝐴 exp(𝑏 ′ 𝜃)𝜃

At equilibrium, 𝑘𝑎 𝑝 = 𝐴 exp(𝑏 ′ 𝜃)𝜃

⇒ 𝜃 ∝ ln(𝑝)

Similarly, Δ𝐻(𝐸𝑑 − 𝐸𝑎 ) ∝ 𝑓(𝜃) = −Δ𝐻𝑚 𝑙𝑛𝜃 (1 > 𝜃 > 0)

(Heat of adsorption is dependent on surface coverage)
𝜃 ↑ ΔH ↓

(4) Freundlich (empirical)

𝜃 = 𝑎𝑝𝑎 𝑚
(For multi-component adsorption ⇒ use Toth, Sips isotherms ⇒ see the book by DO
,1998)
Different types of isotherms:

Physical adsorption (Physisorption) Chemical adsorption (Chemisorption)

- No bond formation - Bond formation

- Non-specific - Specific

- Forces between adsorbed & - Forces are strong

adsorbate molecules are weak (Covalent, valence forces)

(Vander Waals forces)

- Heat of adsorption is small -Heat of adsorption is large

51
 (0.5 - 5 kcal/mole) (>10 kcal/mole)

- Activation energy is low - Can be low or high

- T ↑ amount of adsorbed molecules ↓ - T ↑ amount ↓ 𝑜𝑟 ↑

- Coverage is multilayer - monolayer

- Characterizes the surface - Basic component of catalytic effect

Fundamental mechanism of heterogeneous reaction:

Rate equations:- Langmuir – Hinshelwood model/Eley-Rideal model:

(most common)
𝐴+𝐵 ⇌𝐶 +𝐷

(Reaction only in the presence of catalyst; non-elementary reaction)

: 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 𝑠𝑢𝑟𝑓𝑎𝑐𝑒

𝑋 (𝑠𝑖𝑡𝑒𝑠)
Elementary steps: 𝐴 𝐵 𝐶 𝐷
↓↑ ↓↑ ↓↑ ↓↑
(1) Adsorption
(2) Surface-reaction 𝑋 𝑋 𝑋 𝑋
(3) Desorption
[𝐴𝑋] [𝐵𝑋] [𝐶𝑋] [𝐷𝑋]

L-H model assumes that the step (2) or surface reaction is the slowest or rate-determining step. Other
possibilities also exist. See the book by Froment - Bischoff for details.

𝑘1
𝐴 + 𝑋 ⇌ 𝐴𝑋
𝑘−1
1) 𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛
𝑘2
𝐵 + 𝑋 ⇌ 𝐵𝑋
𝑘−2 }

𝑘3
𝐴𝑋 + 𝐵𝑋 ⇌ 𝐶𝑋 + 𝐷𝑋 ∶ Surface reaction/re − arrangement
𝑘−3 (slowest; other steps are at
equilibrium; they are fast
𝑘4 enough to reach equilibrium and r = 0)
𝐶𝑋 ⇌ 𝐶 + 𝑋
𝑘−4
𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛
𝑘5
𝐷𝑋 ⇌ 𝐷 + 𝑋
𝑘−5 }

52
 𝜕𝐶 𝜕𝐶
𝐶𝐿− = 𝐶𝐿+ ⇒ | = | = 0 (gradient is flat)
𝜕𝑧 𝐿− 𝜕𝑧 𝐿+

On non-dimensionalization,
1 𝜕 2𝜙 𝜕𝜙 (−𝑣𝐴 𝑟)𝜏
− = 𝑅𝑎 (𝑅𝑎 = )
𝑃𝑒 𝜕𝜂 2 𝜕𝜂 𝐶𝑜

𝜕𝜙 1
BCs 𝜂 = 0,
𝜕𝜂 𝑃𝑒
=𝜙−1

𝜕𝜙
= 1, =0
𝜕𝜂

For 1st order reaction,

𝑃
𝐶𝑓 4𝑎 𝑒𝑥𝑝 ( 𝑒⁄2)
=
𝐶𝑜 (1 + 𝑎)2 𝑒𝑥𝑝 (𝑎𝑃𝑒 ) − (1 − 𝑎)2 𝑒𝑥𝑝 (− 𝑎𝑃𝑒 )
2 2

4𝑘𝜏
where, 𝑎 = √1 +
𝑃𝑒

For the other reaction orders, numerical technique is required to solve the equation.

𝐈𝐧 𝐬𝐮𝐦𝐦𝐚𝐫𝐲:
non − ideal reactor (1) Find J(t) for DL or Pe
C
(2)Determine X (1 − ) from the performance equation.
Co
Note: 1. Numerous empirical equations are available for packed bed reactors to determine 𝐷𝐿 =
𝐷𝐿 (𝑅𝑒 , 𝑆𝑐), or under turbulent flow conditions, 𝑃𝑒,𝐿 = 𝑃𝑒,𝐿 (𝑅𝑒 , 𝑆𝑐 ).

2. In a laminar flow there is no axial dispersion/”eddy”.

2𝑛𝑑 3𝑟𝑑 4𝑡ℎ

𝐷𝐿 1𝑠𝑡
turbulent
𝐷𝑚
?? DL ∝ R e
𝐷𝐿 > 𝐷𝑚
(eddies)
(correlations are available)
2100
𝑢𝑑
< 10 𝑅𝑒 =
ν 42
Explain the 2nd phase in the above-figure: why does axial dispersion coefficient exceed molecular
diffusion coefficient even if Re < 2100?

u 2 d2 (not because of eddy but because of

DL = Dm +
192Dm laminar flow profile)

Aris − Taylor diffusion

coefficient
𝑟2
𝑢𝑧 = 𝑢𝑜 (1 − 2 )
𝑅

Species balance:

𝜕𝐶 𝑟2 𝜕𝐶 𝜕 2𝐶 𝐷𝑚 𝜕 𝜕𝐶
+ 2𝑢 (1 − ) = 𝐷𝑚 + (𝑟 𝜕𝑟 ); 𝑅𝑒 < 2100
𝜕𝑡 𝑅 2 𝜕𝑧 𝜕𝑧 2 𝑟 𝜕𝑟

𝑡=0 𝑐=0

0+ 𝐶 = 𝐶𝑖𝑛 @ 𝑧 = 0
𝜕𝐶
=0 @ 𝑧=𝐿
𝜕𝑧

𝐶(𝑡, 𝑟, 𝑧) → C(t, z)
averaged over ′r′

From 2D to 1D model:

It can be shown that

𝜕𝐶 𝜕𝐶 𝜕 2𝐶
+𝑢 = 𝐷𝐿 ; 𝐶 (𝑡, 𝑧)
𝜕𝑡 𝜕𝑧 𝜕𝑧 2 𝑢
+ ′𝐷𝐿 ′

𝑢2 𝑑 2
where, 𝐷𝐿 = 𝐷𝑚 +
192𝐷𝑚

extra term arising out of 1D approximation

(See Dean’s book on Transport Phenomena)

43
 Lecture 12
𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
Ex. 1. 2𝐶𝑂 + 𝑂2 → 2𝐶𝑂2 (non-elementary! All catalytic reactions are
non-elementary.)

Propose a L-H_type reaction mechanism

Elementary steps:

(1) 𝐶𝑂 + 𝑋 ⇌ 𝐶𝑂𝑋 (molecules just sit on the surface; physical adsorption)

𝑂2 + 2𝑋 ⇌ 2𝑂𝑋 (chemisorption; 𝑂2 dissociates occupying two sites)

(𝑂2 + 𝑋 ⇌ 𝑂2 . 𝑋) ← it can also happen, which is physisorption)

𝑂𝑋 + 𝐶𝑂𝑋 ⇌ 𝐶𝑂2 𝑋 + 𝑋 (surface re-arrangement but ‘𝑋’ is conserved)

𝐶𝑂2 𝑋 ⇌ 𝐶𝑂2 + 𝑋 (desorption)

1 ′𝑋′
𝐶𝑂 + 𝑂2 → 𝐶𝑂2 (overall reaction)
2

How? by multiplying the 2nd step by ′ 1⁄2 ′ and adding to eliminate all
1
intermediates; stoichiometry/mass balance must hold good : (𝑂2 + 2𝑋 ⇌ 2𝑂𝑋)
2

(2) Assume that surface-re-arrangement/reaction controls, implying that the

other steps are at equilibrium (𝑟 = 0)

𝑟 = 𝑘3 (𝐶𝑂𝑋)(𝑂𝑋) − 𝑘−3 (𝐶𝑂2 𝑋)(𝑋)

≈ 𝑘3 (𝐶𝑂𝑋)(𝑂𝑋) (The reaction under consideration clearly indicates an

approximately irreversible reaction/situation)

(𝐶𝑂𝑋) (𝑂𝑋)2 (𝐶𝑂 𝑋)

(3) 𝐾1 = (𝐶𝑂)(𝑋) ; 𝐾2 = (𝑂 )(𝑋)2
; 𝐾4 = (𝐶𝑂 2)(𝑋)
2 2
𝑘3
(Note 𝐾3 𝑖𝑠 also defined as , because the step describes an elementary
𝑘−3
reaction, even if the step is assumed to be irreversible with a non-zero rate.)

55
 (4) (𝑋)𝑜 = (𝑋) + (𝐶𝑂𝑋) + (𝑂𝑋) + (𝐶𝑂2 𝑋) (sites are either free or occupied)
1
𝑋𝑜 2 𝑘3 𝐾1 √𝐾2 (𝑂2 )2 (𝐶𝑂)
𝑟= 1 2
[1+√𝐾2 (𝑂2 )2 +𝐾1 (𝐶𝑂)+𝐾4 (𝐶𝑂2 )]

Special Cases:
(1) 𝑂2 is in excess
(2) weak 𝐶𝑂2 adsorption (𝐾4 is a small quantity; [𝐶𝑂2 𝑋] is small)

𝛼𝐶𝑂
𝑟 = (𝛽
+ 𝐶𝑂)2
𝐶𝑂 1 1⁄√𝛼
√ = √ (𝛽 + 𝐶𝑂) √
𝐶𝑂
𝑟 𝛼
𝑟
𝛽⁄√𝛼
Data: 𝐶𝑂
𝑂2 𝐶𝑂 𝐶𝑂2 𝑟
   
   
Ex. 2. The following reaction between NO and CO is important in controlling the
emission of these air pollutants from IC engines:
𝑃𝑡⁄𝑅ℎ
2𝑁𝑂 + 2𝐶𝑂 → 2𝐶𝑂2 + 𝑁2

- The reaction takes place on the Pt⁄Rh supported catalyst and follows the L-H
mechanism. NO, CO and CO2 adsorb on the catalyst non-dissociatively; N2 adsorbs
dissociatively.

- Derive the rate expression for the forward reaction

- What is the rate expression for the reverse reaction?
2𝐶𝑂2 + 𝑁2 → 2𝑁𝑂 + 2𝐶𝑂

56
Ans: As per the L-H model, the surface re-arrangement or reaction rate is
controlling:

k1
NO + X ⇌ NOX
k −1
k2
CO + X ⇌ COX
k −2 [NOX] = K1 [NO][X]
ks [COX] = K 2 [CO][X]
NOX + COX ⇌ CO2 X + NX
k −s [CO2 X] = K 3 [CO2 ][X]
k3 [NX]2 = K 4 [N2 ][X]2
CO2 X ⇌ CO2 + X
k −3
k4
2NX ⇌ N2 + 2X
k −4 }

Site conservation equation:

1⁄ 1
[𝑋𝑜 ] = [𝑋] + 𝐾1 [𝑁𝑂][𝑋] + 𝐾2 [𝐶𝑂][𝑋] + 𝐾3 [𝐶𝑂2 ][𝑋] + 𝐾4 2 [𝑁 ] ⁄2 [𝑋]
2

[𝑋𝑜 ]
→ [𝑋] = 1 1
1+𝐾1 [𝑁𝑂]+𝐾2 [𝐶𝑂]+𝐾3 [𝐶𝑂2 ]+𝐾4 ⁄2 [𝑁2 ] ⁄2

Re-read the question “Determine the rate of the forward reaction: 2𝑁𝑂 + 2𝐶𝑂
→ 𝐶𝑂2 + 𝑁2 . The overall reaction is controlled by the surface arrangement step;
the other steps are at equilibrium and do not contribute to the rate of reaction.
Therefore,

𝑟𝑓𝑜𝑟𝑤𝑎𝑟𝑑 = 𝑘𝑠 [𝑁𝑂𝑋][𝐶𝑂𝑋] = 𝑘𝑠 𝐾2 𝐾1 [𝑁𝑂][𝐶𝑂][𝑋]2

𝐾1 𝐾2 𝑘𝑠 [𝑁𝑂][𝐶𝑂][𝑋𝑜 ]2
= 1 1 2
[1+𝐾1 [𝑁𝑂]+𝐾2 [𝐶𝑂]+𝐾3 [𝐶𝑂2 ]+𝐾4 ⁄2 [𝑁2 ] ⁄2 ]

1⁄ 1
𝑟𝑟𝑒𝑣𝑒𝑟𝑠𝑒 = 𝑘−𝑠 [𝐶𝑂2 𝑋][𝑁𝑋] = 𝑘−𝑠 𝐾3 𝐾4 2 [𝑁 ] ⁄2 [𝐶𝑂 ][𝑋]2
2 2
1 1
𝑘−𝑠 𝐾3 𝐾4 ⁄2 [𝑁2 ] ⁄2 [𝐶𝑂2 ][𝑋𝑜 ]2
= 1 1 2 (Note that the denominator has not
[1+𝐾1 [𝑁𝑂]+𝐾2 [𝐶𝑂]+𝐾3 [𝐶𝑂2 ]+𝐾4 ⁄2 [𝑁2 ] ⁄2 ]

changed)

57
What are the approximations you can make?

⇒ 𝐶𝑂2 adsorbs weakly: 𝐾3 ≪ 𝐾1 , 𝐾2 , 𝐾4

K1 K2 ks [NO][CO][Xo ]2
rforward = 1 1 2 (Similarly for rreverse )
[1+K1 [NO]+K2 [CO]+K4 ⁄2 [N2 ] ⁄2 ]

⇒ 𝑁2 adsorbs weakly? No. This may not be a good approximation.

⇒ 𝑁2 adsorbs physically without dissociation. This is a possibility. Therefore,

𝑘4
[𝑁 𝑋]
𝑁2 𝑋 ⇌ 𝑁2 + 𝑋 𝐾4 = [𝑁 2][𝑋]
2
𝑘−4

In this case, re-do the calculation for [𝑋𝑜 ] = ⋯ ⋯

[𝑋𝑜 ]
𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑒 𝑡𝑜 𝑔𝑒𝑡 [𝑋] =
1+⋯⋯

Thus, obtain a new rate expression for the reaction

⇒ Initial rates? These are often useful. Product concentrations, (𝑁2 ), (𝐶𝑂2 ) ≪
(𝐶𝑂), (𝑂2 ). In such case, 𝑟𝑓𝑜𝑟𝑤𝑎𝑟𝑑 𝑜𝑟 𝑟𝑒𝑣𝑒𝑟𝑠𝑒 can be approximated by neglecting the
product terms in the denominator.

H.W. (F-B 2nd ed.): 2.1, 2.2(a), 2.3(a), 2.4, 2.7(a-b)

58
 Lecture 13
Characterization of a catalyst

(a) BET area, 𝒎𝟐 /𝒈

⇒ Measurement using a gas (𝑁2 ) ⇒ It gives the non – selective surface area.
𝑒. 𝑔. Adsorbate on adsorbent
𝑁2 carbon, zeolites, Pt, Si/Al2O3

First determine the isotherm:

(1) Static method: The material is brought in contact with N2 at different pressures but at
a constant temperature. Determine the amounts of 𝑁2 (moles or volume at STP)
adsorbed at equilibrium at different pressures. 𝑇
𝑇 = 𝑓𝑖𝑥𝑒𝑑
𝑉
𝑃: 𝑃𝑒1 → 𝑃𝑒2 → ⋯ ⋯
𝑉: 𝑉1 → 𝑉2 → ⋯ ⋯
𝑃
(2) Dynamic Method: (Shell development)
- A mixture of 𝑁2 (adsorbing) and 𝐻𝑒 (non-adsorbing) gases of known composition
is used. P and T are maintained constant. The gaseous mixture is allowed to flow
or pass over/through the solid and reach an equilibrium. Difference between the
thermal conductivity (K) of the exit (equilibrated) stream (mixture) and that of the
inlet stream is measured. The compositions of the mixture are changed, keeping
the total 𝑃 & 𝑇 constant. The experiment is repeated at the new partial pressures
of 𝑁2 . Thermal conductivity of the mixture (𝑁2 + 𝐻𝑒) is calibrated wrt. its
composition.

Δ𝐾 TCD 𝐵𝑎𝑡ℎ
B 𝑈𝑝
t = 0; move the bath up. T
A Δ𝐾
T ⇒ solid is equilibrated V
Inlet (N2 + He) 𝛥𝐾 𝑖𝑠 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒𝑑
mixture O
𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛
T
𝑏𝑎𝑡ℎ t
adsorbing (𝑢𝑝/𝑑𝑜𝑤𝑛)
non − adsorbing
𝑃𝑁2 𝐿𝑜𝑤𝑒𝑟
O
T (Isotherm)
Porous solid
ΔK = change in electrical data are t
𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛
conductivity independent of }
𝑉
initial conditions
59
p
 BET model (linearize):
𝑐−1 1 1
𝑆= ; 𝐼= ⇒ 𝑉𝑚 = (𝑆𝑇𝑃)
𝑐𝑉𝑚 𝑐𝑉𝑚 𝐼+𝑆

𝑉𝑚 (𝑐𝑐,𝑆𝑇𝑃/𝑔𝑚)
𝑆𝑔 ⤏ = # 𝑜𝑓 𝑚𝑜𝑙𝑒𝑠 𝑜𝑓 𝑁2 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑 𝑡𝑜 𝑠𝑎𝑡𝑢𝑟𝑎𝑡𝑒
22400(𝑐𝑐,𝑆𝑇𝑃/𝑚𝑜𝑙𝑒)

𝑉𝑚 ×𝑁 𝑉𝑚 ×𝑁
⤏ ⇒( ) × 𝛼 ( 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑎𝑟𝑒𝑎 𝑜𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑝𝑒𝑟 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒)
22400 22400

2⁄ 4
𝑀 3 𝑉 = 𝜋𝑟 3
𝑤ℎ𝑒𝑟𝑒, 𝛼 = 1.09 ( ) 3 }
𝑁𝜌 2
𝑆 = 𝜋𝑟
𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑒𝑑 (𝑙𝑖𝑞𝑢𝑖𝑑 𝑁2 ) phase

For 𝑁2 at – 195.8oC, M = 28, 𝜌 = 0.808 g/cc, 𝛼 = 16.2 A2, N = 6.023x1023

𝒄𝒎𝟐 ⁄𝒈 : 𝑆𝑔 (𝑁2 ) = 4.35 × 104 𝑉𝑚 (𝑐𝑐/𝑔𝑚 𝑎𝑡 𝑆𝑇𝑃)

Some examples (𝒎𝟐 ⁄𝒈):

Activated clay: 150 - 220; SiO2/Al2O3 : 200 - 500; Silica gel: 200 - 600;

Activated Carbon: 500 - 1500; molecular sieves > 1000

Note: There are porous as well as non-porous silica gels.

(b) Porosity, density

(void fraction)
skeletal density, ρs
𝜌𝑔 = 𝜌𝑠 (1 −∈𝑔 ) ← {grain/particle density, ρ
g

𝑔𝑟𝑎𝑖𝑛 (𝑚𝑖𝑐𝑟𝑜) 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 ρs > ρg > ρP > ρb

(because of voids)
𝑝𝑜𝑤𝑑𝑒𝑟/𝑝𝑒𝑙𝑙𝑒𝑡 𝑑𝑒𝑛𝑠𝑖𝑡𝑦, 𝜌𝑃
𝜌𝑃 = 𝜌𝑔 (1 −∈𝑚 ) ← {

𝑝𝑜𝑤𝑑𝑒𝑟/𝑝𝑒𝑙𝑙𝑒𝑡/ 𝑚𝑎𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦

𝑝𝑎𝑐𝑘𝑒𝑑 𝑏𝑒𝑑, 𝜌𝑏 𝑑𝑒𝑛𝑠𝑖𝑡𝑦

𝜌𝑏 = 𝜌𝑃 (1 −∈𝑏 ) ← {

𝑝𝑎𝑐𝑘𝑒𝑑 𝑏𝑒𝑑 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦

(𝑁𝑜𝑡𝑒: 𝑔𝑟𝑎𝑖𝑛 𝑖𝑠 "one" 𝑝𝑜𝑤𝑑𝑒𝑟; 𝑝𝑜𝑤𝑑𝑒𝑟 𝑖𝑠 𝑝𝑙𝑢𝑟𝑎𝑙!)

60
 𝑉𝑜𝑙𝑢𝑚𝑒 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦
𝑉𝑠 𝜌𝑠 ∈𝑔 𝑜𝑟 ∈µ
𝑚𝑖𝑐𝑟𝑜 𝑉𝑔 𝜌𝑔
∈𝑚
𝑉µ 𝑉𝑃 𝜌𝑃
𝑉𝑚 𝑉𝑏 𝜌𝑏 ∈𝑏
𝑚𝑎𝑐𝑟𝑜
Terminologies:-

- Non-porous (no porosity)

- Intra-particle

- Inter – particle

- Macro pores

- Micro pores

- Bed, powder

- Grain, Sub-grain

- Pores

Note: A pellet/powder may have both micro and macro porosities.

2. There is a hierarchical structure in the materials and packed bed reactors.

𝑔𝑟𝑎𝑖𝑛 𝑔𝑟𝑎𝑖𝑛

𝐵𝑖𝑛𝑑𝑒𝑟 𝑝ℎ𝑎𝑠𝑒 𝑜𝑟
𝑣𝑜𝑖𝑑 𝐵𝑖𝑛𝑑𝑒𝑟 𝑝ℎ𝑎𝑠𝑒 𝑜𝑟 𝑣𝑜𝑖𝑑
𝑐𝑦𝑙𝑖𝑛𝑑𝑟𝑖𝑐𝑎𝑙 𝑝𝑒𝑙𝑙𝑒𝑡𝑠

𝑠𝑝ℎ𝑒𝑟𝑖𝑐𝑎𝑙 𝑝𝑒𝑙𝑙𝑒𝑡𝑠

𝜌𝑠 𝜌𝑔 𝜌𝑝 𝜌𝑏
# 𝑜𝑓 𝑝𝑒𝑙𝑙𝑒𝑡 𝑉𝑏 𝑉𝑝 (𝑝𝑒𝑙𝑙𝑒𝑡
𝜖𝑔 𝜖𝑝 𝜖𝑏 }
𝑣𝑜𝑙𝑢𝑚𝑒)
(𝜖𝜇 𝜖𝑚 )
𝑉𝑏 − 𝑛𝑉𝑝
(𝜖𝑏 = )
𝑉𝑏 𝜖𝑏 𝑝𝑎𝑐𝑘𝑒𝑑 𝑏𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 − 𝑝𝑒𝑙𝑙𝑒𝑡𝑠
61
Measurements:

Displacement method: Mercury – Helium displacement

𝑤𝑖𝑙𝑙 𝑛𝑜𝑡 𝑝𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑒 𝑤𝑖𝑙𝑙 𝑝𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑒 𝑎𝑙𝑙

𝑚𝑖𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑒𝑠 𝑝𝑜𝑟𝑒𝑠 𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑎𝑑𝑠𝑜𝑟𝑏𝑖𝑛𝑔

𝑇𝑤𝑜 𝑒𝑥𝑡𝑟𝑒𝑚𝑒𝑟 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠

measure displacement

′𝑜𝑛𝑒 ′ 𝑔𝑟𝑎𝑖𝑛

𝑉𝐻𝑔 − 𝑉𝐻𝑒 = 𝑝𝑜𝑟𝑒 𝑣𝑜𝑙𝑢𝑚𝑒

𝑉𝐻𝑔 − 𝑉𝐻𝑒
𝜖𝑔 = 𝑉𝐻𝑔

(c) Pore size distribution (PSD)

1. Mercury-penetration method (Mercury porosimetry)

Non-wetting liquid

- used for large pores (generally micron-size > 0.1 - 1 𝜇m or 100 – 1000 nm)
𝑃 𝑃 𝑃 × 𝜋𝑎2 = −(2𝜋𝑎)𝛤𝑐𝑜𝑠𝜃
𝑎𝑝𝑝𝑙𝑦 𝑐𝑜𝑛𝑡𝑎𝑐𝑡
𝑡𝑜 𝑝𝑢𝑠ℎ 𝐻𝑔 2𝛤𝑐𝑜𝑠𝜃 𝑎𝑛𝑔𝑙𝑒
𝑖𝑛𝑡𝑜 𝑝𝑜𝑟𝑒𝑠 𝑝𝑜𝑟𝑒 − 𝑠𝑖𝑧𝑒 (𝑎) = −
𝜎 𝐻𝑔 𝑃
𝑃
8.75×105
𝑎= 𝑓𝑜𝑟 𝐻𝑔 (𝜃 = 140𝑜 𝑓𝑜𝑟 𝐻𝑔)
𝑃(𝑝𝑠𝑖)
𝐻𝑔

𝑂𝑛𝑒 ′𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒′ → otherwise Hg will penetrate the pores or voids between the grains
(pores in powder), rather than ‘grain’ pores.

62
 𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒
𝑉
𝑃 𝑉 𝑎
0 0 −
  − 𝑉𝑜
  −

𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 𝑎1 𝑎 𝑎2
𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒𝑠
𝑣𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑝𝑜𝑟𝑒𝑠 (𝑃1 > 𝑃2 )
𝑜𝑟 𝑎 ↓ 𝑃 ↑
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙Type equation here. 𝑝𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑒𝑑 𝑎𝑡 𝑃 𝑎𝑙𝑙 𝑝𝑜𝑟𝑒𝑠 > 𝑎1 𝑜𝑟 𝑎 ↑ 𝑃 ↓
ℎ𝑎𝑣𝑒 𝑏𝑒𝑒𝑒𝑛 𝑝𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑒𝑑

𝑑𝑉
𝑑𝑎
1 𝑑𝑉
⇒ 𝜖(𝑎) = 𝑉 (𝑉𝑜𝑖𝑑 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛)
𝑜 𝑑𝑎

𝑂 (𝑎) (𝑎 + 𝑑𝑎) ∞
𝑐𝑚3
( )
𝑐𝑚3 − ′𝑐𝑚′
∞
𝜖𝑡𝑜𝑡𝑎𝑙 = ∫0 𝜖(𝑎)𝑑𝑎
𝑚𝑜𝑠𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑜𝑟𝑒𝑠 (𝑎)
𝑎𝑟𝑒 𝑙𝑜𝑐𝑎𝑡𝑒𝑑 ℎ𝑒𝑟𝑒

1 𝑑𝑉
𝐼𝑓 𝑦𝑜𝑢 𝑐ℎ𝑜𝑜𝑠𝑒 𝜖(𝑎) =
𝑉𝑜 𝑑(𝑙𝑜𝑔𝑎)
𝑝𝑜𝑤𝑑𝑒𝑟 𝑜𝑟 𝑝𝑒𝑙𝑙𝑒𝑡
𝑑𝑉
⇒ 𝑑(𝑙𝑜𝑔𝑎)

𝑎µ 𝑎𝑚
micro - pore 𝑚𝑎𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑒 log(𝑎)
size 𝑠𝑖𝑧𝑒
(very often log 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛)

Some calculations:

𝑎 = 100𝐴𝑜 𝑜𝑟 10 𝑛𝑚 → 𝑃 = 8750 𝑃𝑠𝑖/ (~ 500 𝑎𝑡𝑚)

= 1000𝐴𝑜 𝑜𝑟 100 𝑛𝑚 → 𝑃 = 875 𝑃𝑠𝑖/ (~50 𝑎𝑡𝑚)
= 1 𝜇𝑚 → 𝑃 = 87.5 𝑃𝑠𝑖/ (~5 𝑎𝑡𝑚)

(The material of construction is a concern in such high pressure systems)

63
 There is an IUPAC classification: micro meso macro

50 𝐴𝑜 450 𝐴𝑜
(5 𝑛𝑚) (45 𝑛𝑚)

nanopores : < 2-3 nm

2. Capillary condensation (very small pores)

(𝑚𝑖𝑐𝑟𝑜 − 𝑚𝑒𝑠𝑜 𝑝𝑜𝑟𝑒𝑠)

Kelvin Effects:
𝑣𝑎𝑝𝑜𝑢𝑟 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒(𝑃𝑜 )
𝑓𝑙𝑎𝑡 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑟 → ∞: 𝑃𝑜 (𝑇)
𝑃𝑜 (𝑇) 𝑤𝑒𝑡𝑡𝑖𝑛𝑔 𝑙𝑖𝑞.
𝜃 > 180𝑜
𝜃
𝑜 𝑜 𝑃𝑜′ < 𝑃𝑜
180 𝜃 < 180
𝑛𝑜𝑛 − 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑤𝑎𝑡𝑒𝑟 − 𝑑𝑟𝑜𝑝𝑙𝑒𝑡𝑠
(𝑛𝑜 − 𝑤𝑎𝑙𝑙 𝑒𝑓𝑓𝑒𝑐𝑡𝑠) 𝜃 𝑤𝑒𝑡𝑡𝑖𝑛𝑔 𝑙𝑖𝑞. 𝑡𝑜 𝑒𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑒 𝑖𝑠 ℎ𝑖𝑔ℎ𝑒𝑟 𝑡ℎ𝑎𝑛
𝑡ℎ𝑒 𝑓𝑙𝑎𝑡 𝑤𝑎𝑡𝑒𝑟 𝑠𝑢𝑟𝑓𝑎𝑐𝑒
𝑛𝑜𝑛 − 𝑤𝑒𝑡𝑡𝑖𝑛𝑔 𝑃𝑜′ > 𝑃𝑜
𝑙𝑖𝑞.
𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑠𝑢𝑐ℎ 𝑑𝑟𝑜𝑝𝑠 𝑡𝑜
𝑑𝑖𝑠𝑎𝑝𝑝𝑒𝑎𝑟 𝑖𝑠 𝑙𝑒𝑠𝑠

Capillary has the same effects: Capillary vapor pressures

𝜎𝑉 𝑐𝑜𝑠𝜃
𝑙 𝜃
𝑃𝑎𝑑𝑠 = 𝑃𝑜 𝑒𝑥𝑝 (− 𝑅𝑇(𝑎−𝛿) ) ∶ 𝐾𝑒𝑙𝑣𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛(1) 𝜃

Both
𝑟→∞ 𝑓𝑖𝑙𝑚 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 situations are possible

→𝑎←
𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑡𝑒𝑛𝑠𝑖𝑜𝑛 𝑚𝑜𝑙𝑎𝑟 𝑣𝑜𝑙𝑢𝑚𝑒

2𝜎 𝑉𝑙 𝑐𝑜𝑠𝜃
𝑃𝑑𝑒𝑠 = 𝑃𝑜 𝑒𝑥𝑝 (− ) ∶ 𝐾𝑒𝑙𝑣𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 (2) −1⁄𝑛
𝑅𝑇(𝑎−𝛿) 𝑃𝑜
𝛿(𝐴𝑜 ) = 9.52 (𝑙𝑜𝑔 ( ))
𝑃
𝑑𝑒𝑠:
𝑛 = 1 𝑓𝑜𝑟 𝑁2
𝑎𝑑𝑠: 𝐶𝑜𝑛𝑑𝑒𝑛𝑠𝑎𝑡𝑖𝑜𝑛 𝑎𝑥𝑖𝑎𝑙
(𝑝𝑜𝑟𝑒 − 𝑓𝑖𝑙𝑙𝑛𝑔 𝑖𝑠 𝑜𝑐𝑐𝑢𝑟𝑠 𝑜𝑛 𝑤𝑎𝑙𝑙𝑠
𝑟𝑎𝑑𝑖𝑎𝑙) (𝑠𝑙𝑜𝑤𝑙𝑦; 𝑖𝑡 𝑓𝑖𝑙𝑙𝑠 𝑢𝑝 𝑃𝑜𝑟𝑒𝑠 𝑎𝑙𝑟𝑒𝑎𝑑𝑦 𝑓𝑖𝑙𝑙𝑒𝑑 (𝑤𝑒𝑡);
𝑡ℎ𝑒 𝑝𝑜𝑟𝑒𝑠) 𝑒𝑣𝑎𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑜𝑐𝑐𝑢𝑟𝑠 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑡𝑜𝑝
→𝛿← 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 (𝑑𝑒𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 𝑖𝑠 𝑎𝑥𝑖𝑎𝑙)

64
 Therefore, routes are different for filling & emptying, and there is a hysteresis

1. 𝑃𝑑𝑒𝑠 < 𝑃𝑎𝑑𝑠 for the same 𝑉𝑎𝑑𝑠/𝑑𝑒𝑠 .

2. 𝐴𝑙𝑠𝑜, 2 ways of doing experiment to determine
𝑎𝑑𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛 𝑉(𝑃): 𝑎𝑑𝑠/𝑑𝑒𝑠.
𝑉
lower is ‘a’, lower is 𝑃𝑎𝑑𝑠 ; first small sized pores will be
𝑃𝑎𝑑𝑠 filled in the low pressure to high pressure experiment.
𝑃𝑑𝑒𝑠 On the other hand, big size pores will desorb first in the
𝑃
high pressure to low pressure experiment.
(desorption carves are preferred)

Expt:
𝑃
𝑉 𝑝 𝑎
𝑔𝑎𝑠 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒
(𝑁2 )
  
𝑔𝑎𝑢𝑔𝑒
  
  
𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡

(𝒅) 𝑨𝒗𝒈 𝒑𝒐𝒓𝒆 − 𝒔𝒊𝒛𝒆 (𝒂)

2𝑉𝜖𝑔 𝑆 = 2𝜋𝑟𝐿
𝑎= 𝑔𝑟𝑎𝑖𝑛 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 (𝑎𝑠𝑠𝑢𝑚𝑖𝑛𝑔
𝑆 𝑉 = 𝜋𝑟 2 𝐿
𝑐𝑦𝑙𝑖𝑛𝑑𝑟𝑖𝑐𝑎𝑙
𝑠𝑎𝑚𝑝𝑙𝑒/𝑔𝑟𝑎𝑖𝑛 𝑣𝑜𝑙𝑢𝑚𝑒 𝑟 = 2𝑉⁄𝑆
𝑝𝑜𝑟𝑒𝑠)
𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑎𝑟𝑒𝑎

65
 Lecture 14 𝑝𝑒𝑙𝑙𝑒𝑡 (𝑝𝑜𝑤𝑑𝑒𝑟
/𝑐ℎ𝑢𝑛𝑘)
Material Characterization (……continued)

Example: Al2O3 pellets: Mass = 3.15 g (per pellet)

𝑡 = 1⁄4 "
Data: macro − pore volume = 0.205 cc/g These data have
}
𝑚𝑖𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑒 𝑣𝑜𝑙𝑢𝑚𝑒 = 0.4 𝑐𝑐/𝑔 come from PSD analysis.
𝑣𝑜𝑖𝑑𝑠
( ) (𝑖𝑛 𝑎 𝑔𝑟𝑎𝑖𝑛)
𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑔𝑟𝑎𝑖𝑛 𝑑𝑝 = 1"
a) Density of the pellet = ? 𝑔𝑟𝑎𝑖𝑛
𝜌𝑃 = 𝑀⁄𝑉 = 3.15⁄ 𝜋𝑑 2 = 0.978 𝑔/𝑐𝑐 (𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒)
( 4𝑝 × 𝑡)
𝑚𝑎𝑐𝑟𝑜 − 𝑣𝑜𝑙𝑢𝑚𝑒
3.22

0.205 𝑐𝑐/𝑔
b) Determine ∈𝑚 (𝑚𝑎𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦) = 1 = 0.205 × 0.978 = 0.2005
⁄𝜌𝑝 (𝑐𝑐/𝑔)

𝑡𝑜𝑡𝑎𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑝𝑒𝑙𝑙𝑒𝑡

0.4 𝑐𝑐/𝑔
∈𝜇 (𝑚𝑖𝑐𝑟𝑜 − 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦) = 1 = 0.4 × 0.978 = 0.3912
⁄𝜌𝑝 (𝑐𝑐/𝑔)

c) Determine skeletal volume fraction = (Total volume – all pore volumes)/Total volume
(𝑠𝑜𝑙𝑖𝑑𝑠 𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛 ∈𝑠 )

1⁄ − 0.205 − 0.4
𝜌𝑝 0.205 0.4
= 1⁄ = (1 − 1⁄ − 1⁄ )
𝜌𝑝 0.978 0.978
∈𝑚 ∈𝜇

It is the same as = (1 − ∈𝑚 − ∈𝜇 ) = (1 − 0.2005 − 0.3912) = 0.4083

(Note: Assume that there is no binder, and pellet is prepared by press-fitting)

d) Particle/grain density 𝑝𝑒𝑙𝑙𝑒𝑡

ρp (g/cc) (weight of pellet is
Basis: 1 cc of pellet:
grain volume fraction(cc/cc of pellet) because of grains only)

𝑖𝑛𝑐𝑙𝑢𝑠𝑖𝑣𝑒 𝑜𝑓
𝑔𝑟𝑎𝑖𝑛 (𝜇) 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 𝑜𝑟 𝑣𝑜𝑖𝑑
0.978
= (1−0.2005) (see the denominator: 1 - ∈𝑚 )

= 1.23 𝑔/𝑐𝑐

66
 3.15
Or, it can also be calculated: ( 3.22 − 0.2005 × 3.22 )
( )

0.978
e) skeleton (solid) density = (1−∈ = 2.39 𝑔/𝑐𝑐
𝑚 −∈𝜇 )

3.15 mass of pellet

𝑜𝑟 ( ) or (volume of pellet)×∈s
3.22−3.22x(∈ m + ∈μ )
0.978
=
0.4083

(c) Diffusion coefficient in pores

Mechanism of diffusion:

(1) Bulk-phase diffusion (resistance is because of collision between the molecules)

(molecules will experience intermolecular forces)

𝐷𝐴𝐵 = 1⁄3 𝑉 𝜆
(𝑐𝑚2 /𝑠)

𝑚𝑒𝑎𝑛 − 𝑓𝑟𝑒𝑒 𝑝𝑎𝑡ℎ

𝑙𝑖𝑛𝑒𝑎𝑟 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑜𝑓 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑒𝑠

3⁄
𝛼𝑇 2
= 1⁄ ∶ 𝐶ℎ𝑎𝑝𝑚𝑎𝑛 − 𝐸𝑛𝑠𝑘𝑜𝑔 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
𝑃(𝑀𝑎𝑣𝑔) 2

1 1
(1⁄𝑀𝑎𝑣𝑔 = + )
𝑀𝐴 𝑀𝐵

(2) Knudsen diffusion (porous media)

(Resistance is only from walls)

2
𝐷𝐾 = 3 𝑎 𝑉 (𝑎 = 𝑝𝑜𝑟𝑒 𝑠𝑖𝑧𝑒)

𝑇 (𝐾)
𝑐𝑚2 /𝑠 = 9700 𝑎√
𝑀𝑎
(𝑐𝑚)

67
 If pore size << mean free path
or a < < λ particles will K n (Knudsen #)
collide with walls
before they colloid with themselves = λ⁄a
DK ≪ DAB }

𝐼𝑓 𝑎 ≫ 𝜆 ⇒ 𝐷𝐴𝐵 ≪ 𝐷𝐾 (intermolecular collision is more prominent)

For gases at room temperature (?) and 1 atm

0.11 𝑡𝑜 1 𝑐𝑚2 /𝑠 (𝑔𝑎𝑠)

𝜆 = 1000 𝐴𝑜 𝑜𝑟 100 𝑛𝑚 and 𝐷𝐴𝐵 = { 10−5 𝑖𝑛 𝑙𝑖𝑞𝑢𝑖𝑑
10−8 𝑖𝑛 𝑠𝑜𝑙𝑖𝑑

(3) Surface diffusion (~10−10 𝑐𝑚2 /𝑠)

𝑐𝑟𝑒𝑒𝑝𝑖𝑛𝑔/ℎ𝑜𝑝𝑝𝑖𝑛𝑔 𝑚𝑒𝑐ℎ𝑎𝑛𝑖𝑠𝑚
𝐷𝑆 = 𝐴𝑒 −𝐸𝑠 ⁄𝑅𝑇

(4) Intra-lattice or intraege diffusion (solid may not have pores but cracks)

grain boundary
(in − between the lattices)

eg. 𝐻2 purification by 𝑃𝑑 membrane.

- very slow and highly activated process

≈ 10−10 𝑡𝑜 10−15 𝑐𝑚2 /𝑠

(5) Combination of bulk and Knudsen diffusion

1 1 1
2𝑎 𝑎 ≈ 𝜆 (𝐷 = 𝐷 +𝐷 )
𝑐𝑜𝑚𝑏 𝐾 𝐴𝐵

(𝑏𝑜𝑡ℎ 𝑎𝑟𝑒 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑡)

(a) Micro-dispersed pores: Use 𝐷(𝑒𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒) 𝑜𝑟 𝐷𝑒

𝐽𝐴 = −𝐷𝐴𝐵 ∇𝐶 𝑎𝑛𝑎𝑙𝑜𝑔𝑜𝑢𝑠 ⇒ 𝐽 = −𝐷𝑒 ∇𝐶
𝐷𝑒 = 𝜖 𝐷𝐶𝑜𝑚𝑏
: 𝑝𝑎𝑟𝑎𝑙𝑙𝑒𝑙 𝑝𝑜𝑟𝑒𝑠
𝑓𝑟𝑎𝑐𝑡𝑖𝑜𝑛
(𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦)

Pores may be tortuous

𝜖𝐷𝐶𝑜𝑚𝑏
𝐷𝑒 =
𝜏
𝑡𝑜𝑟𝑡𝑢𝑜𝑠𝑖𝑡𝑦 𝑓𝑎𝑐𝑡𝑜𝑟
68
 𝜏 ≈ 1⁄𝜖 (𝑒𝑥𝑝𝑡 𝑑𝑎𝑡𝑎)

𝐷𝑒 = 𝜖 2 𝐷𝐶𝑜𝑚𝑏

(b) Bimodel pore size

𝜖𝜇 2 (1+3𝜖𝑚 )
𝐷𝑒 = 𝐷𝑚 𝜖𝑚 2 + 𝐷𝜇
1−𝜖𝑚
1 1 1
= +
𝐷𝑚 𝐷𝐴𝐵 (𝐷𝐾 )𝑚
1 1 1
= + 𝜖𝜇 𝜖𝑚
𝐷𝜇 𝐷𝐴𝐵 (𝐷𝐾 )𝜇

(c) Random pore model:

-Peterson, AIChE (Vol 3.4(1957))
- Gerales, AIChE, 27(4), 1980.

Experimental:

𝑦𝐴2
𝑔𝑎𝑠 𝐴
𝑣𝐴
Δ𝑧 porous solid
steady − state: (no reaction)
𝑦𝐴1 detector
𝑁2
𝑣𝑁2 (𝑐𝑎𝑟𝑟𝑖𝑒𝑟 𝑔𝑎𝑠) (𝑚𝑒𝑎𝑠𝑢𝑟𝑒 𝐴 in 𝑁2 )

𝑀𝑜𝑑𝑒𝑙 𝑝𝑜𝑟𝑒 − 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 𝑖𝑠 𝑠𝑖𝑚𝑖𝑙𝑎𝑟 𝑡𝑜 𝑡ℎ𝑎𝑡 𝑜𝑓 𝑏𝑢𝑙𝑘 − 𝑝ℎ𝑎𝑠𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛:

𝐽 = −𝐷𝐴𝐵 ∇𝐶 = −𝐷𝑒 (𝐶1Δ𝑧

−𝐶2
)

𝑑𝐶 𝑃 𝑦𝐴2 −𝑦𝐴1
𝑁𝐴 = −𝐷𝑒 = −𝐷𝑒 ( ) (𝑓𝑙𝑢𝑥)
dz RT Δ𝑧

𝑃
= 𝑣𝑁2 𝐶𝐴 = 𝑣𝑁2 𝑦 (flux)
RT 𝐴2

69
 Determine De from the above-measurement/calculation

Pulse: 𝑔𝑎𝑠

𝑛𝑜 𝑝𝑜𝑟𝑒𝑠
𝑢𝑛𝑠𝑡𝑒𝑎𝑑𝑦 − 𝑠𝑡𝑎𝑡𝑒
𝑒𝑥𝑝𝑒𝑟𝑖𝑚𝑒𝑛𝑡
Δ𝑧
𝐶𝐴 𝐼𝑓 𝑝𝑜𝑟𝑜𝑢𝑠

𝑑𝑒𝑡𝑒𝑐𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛
𝑁2
𝑡
∞
(Δ𝑧)2 ∫0 𝐶𝐴 𝑑𝑡
𝑫𝑒 = 𝜖 ∞ (cm2/s)
6 ∫0 𝐶𝐴 𝑡 𝑑𝑡
𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 𝑖𝑛
𝑠𝑜𝑙𝑖𝑑𝑠

d. Effective thermal conductivity:

𝑞 = −𝜆∇𝑇 ; 𝑞𝑒 = −𝜆𝑒 ∇𝑇

𝜆𝑓 𝜖
𝜆𝑒 = 𝜆 𝑠 ( )
𝜆𝑠

(10−4 − 10−5 𝑐𝑎𝑙 ⁄𝑐𝑚 − 𝑠 𝑜 𝑐)

Ref: Masamune & Smith

Ind Chem Eng Date 8, 54 (1963)

HW (Smith, 3rd ed): 8.1, 8.6, 8.7, 8.8

HW (F&B, 2nd ed) : 3.2, 3.4

70
 Lecture 15
Example (1):

P(k𝑃𝑎 ) V (adsorbed amount,

𝑙𝑖𝑞. 𝑁2 : cc @ STP/g of sample)
1.6 61
6.6 127
37.4 170

(a) Find Sg (m2/g) using Langmuir model

(b) Calculate 𝐷𝑒𝑓𝑓 of 𝑁2 in this sample at 100 oC.

Assume uniform pore size & 𝜖 = 0.5. 𝐷𝑘 is dominant; 𝜌𝑝 = 1 g/cc

Plot:

𝑉 𝐾𝑃
𝑃⁄ 𝜃= =
𝑉 𝑉𝑚 1+𝐾𝑃
1⁄ 3
𝑃 1 𝑃
𝑉𝑚 𝑉𝑚 = 185 𝑐𝑚 ⁄𝑔𝑚
1⁄
= +
𝐾𝑉𝑚
𝑉 𝐾𝑉𝑚 𝑉𝑚
𝑃

2
(a) Use 𝑆𝑔 = (4.35 × 104 )𝑉𝑚 = 8050000 𝑐𝑚 ⁄𝑔 𝑜𝑟 805 𝑚2 /𝑔
1 1 1
(b) = + ⇒ 𝐷𝑐𝑜𝑚𝑏 ≈ 𝐷𝑘 (Dk dominates)
𝐷𝑐𝑜𝑚𝑏 𝐷𝐴𝐵 𝐷𝑘
2 𝑇 2𝑉𝑝𝑜𝑟𝑒 2 𝜖𝑔
𝐷𝑘 (𝑐𝑚 ⁄𝑠) = 9700 𝑎 (𝑐𝑚)√ ; 𝑎= = ×
𝑚𝐴 𝑆𝑔 𝑆𝑔 𝜌𝑔

(assuming cylindrical pores)

′𝜖𝑔′ 𝑜𝑟 ′𝜖𝑝′
= 12.4 𝑜 𝐴 𝑜𝑟 1.24 nm 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 or
grain
373
= 9700 × 1.24 × 10−7 √
28

71
 2
= 4.39 × 10−3 𝑐𝑚 ⁄𝑠
𝜖𝐷𝑐𝑜𝑚𝑏 2
𝐷𝑒𝑓𝑓 = = 𝜖 2 𝐷𝑐𝑜𝑚𝑏 = 𝜖 2 𝐷𝑘 = 0.52 × 4.39 × 10−3 = 1.1 × 10−3 𝑐𝑚 ⁄𝑠
𝜏

Example 2
𝑃(𝑘𝑃 ) 0.8 3.3 18.7 30.7 38.0 42.7 57.3 67.3
𝑉(𝑐𝑐 𝑎𝑡 𝑆𝑇𝑃 6.1 12.7 17.0 19.7 21.5 23.0 27.7 33.5
/𝑔𝑚)

(a) Plot BET isotherm (2) Find 𝑆𝑔 using BET equation.

𝑃
𝑆(𝑠𝑙𝑜𝑝𝑒)
𝑉(𝑃𝑜 −𝑃) 𝐵𝐸𝑇 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛:
𝐶−1
=
1 𝐶𝑉𝑚 𝑃 1 𝐶−1 𝑃
𝐼= = +( ) 𝑃𝑜
𝑉𝑚 𝐶 𝑉(𝑃𝑜 −𝑃) 𝑉𝑚 𝐶 𝐶𝑉𝑚

𝑃⁄ 𝑜
𝑃
1 3
𝑉𝑚 = = 13.69 𝑐𝑚 ⁄𝑔
𝐼+𝑆

2
𝑆𝑔 = (4.35 × 104 ) 𝑉𝑚 = 600000 𝑐𝑚 ⁄𝑔 or 60 m2/g

(3) CO adsorption on Pt surface at 0oC: How plots will change if you raise the
temperature to 50 oC?
50𝑜 𝐶
𝐼𝑠𝑜𝑡ℎ𝑒𝑟𝑚
0𝑜 𝐶
0𝑜 𝐶
𝑉 𝑜
50 𝐶 𝑃⁄
𝑉

𝑃 𝑃

72
 Non-porous catalyst (single particle analysis)
(Effect of interphase transport on reaction)
(𝑖𝑛𝑡𝑟𝑎𝑝ℎ𝑎𝑠𝑒 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑖𝑠 𝑓𝑎𝑠𝑡)

So far, 1. We developed the intrinsic rate expression, r = f(C,T) based on the (L-H) mechanism

2. Characterized the catalyst/material: dp, ε: (εμ, εm ), BET area, 𝑎, ρ: (ρg, ρp, ρb), DK

(once again, be careful with the different nomenclatures or subscripts used in different books
and manuscripts)

Next let us discuss (interphase transport + kinetics) under isothermal condition (one temperature in
the reaction-system):

𝐶𝑏
𝐶𝑏
𝑟𝑝 𝑜𝑟 No
𝐶𝑆
reaction 𝐶𝑆 concentration
′𝑘𝑚′ takes place at gradient inside
the external surface the particle?

Step 1: 𝐒𝐭𝐞𝐩 𝟐: Diffusion flux from

𝑏𝑢𝑙𝑘 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 bulk to the external
surface
isothermal ⃗ . ∇𝐶 J = k m (Cb − Cs )
(𝑉
A→R dC
(moles/m3 ) + 𝐷𝑑𝑖𝑠𝑝 ∇2 𝐶) = −D |
dr r = rp
(definition for 𝑘𝑚 )
(non − dimensionalize to define
Sherwood No)

𝐶𝑅,𝑠
𝐶𝑏
(𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓
𝐴 𝑛𝑒𝑎𝑟 𝑜𝑟 𝑎𝑡 𝐶𝑠 . 𝑋 (𝑎𝑑𝑠𝑜𝑟𝑏𝑒𝑑
𝐶𝑆 𝑡ℎ𝑒 𝑠𝑢𝑟𝑓𝑎𝑐𝑒, ≠
𝐶𝑅,𝑏 𝑠𝑝𝑒𝑐𝑖𝑒𝑠, 𝑚𝑜𝑙𝑒𝑠/𝑚2 )
𝑚𝑜𝑙𝑒𝑠/𝑚3 )

73
 (𝑛𝑜 𝑙𝑜𝑐𝑎𝑙
R (heterogeneous rate, 𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚2 ) = 𝑘𝑚 (𝐶𝑏 − 𝐶𝑆 ) 𝑎𝑐𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑜𝑓
𝑚/𝑠 𝑚𝑜𝑙𝑒𝑠/𝑚3 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡 𝑖𝑛 𝑡ℎ𝑒
𝑒𝑥𝑡 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑓𝑖𝑙𝑚)

= 𝑅(𝐶𝑆 )

= 𝑘𝐶𝑆 (if 1st order)

𝑘𝑚 = 𝑓(𝑅𝑒,𝑝 , 𝑆𝑐 )

⇒ 𝑆ℎ = 𝑓(𝑅𝑒,𝑝 , 𝑆𝑐 )

𝑘𝑚 𝑑𝑝 𝑣𝑜 𝑑𝑝 𝜌𝑓 𝜈
= 𝑓( , )
𝐷𝑚 𝜇𝑓 𝐷𝑚

Correlations for calculating km are available for various scenarios (types of fluids, 𝑅𝑒 , Sc, etc)

𝑘𝑚 𝜌 2⁄ 𝑓
𝐽𝐷 = (𝑆𝑐) 3 = ⁄2 (𝐶𝑜𝑙𝑏𝑢𝑟𝑛′ 𝑠 𝑎𝑛𝑎𝑙𝑜𝑔𝑦)
𝐺 𝑔𝑟𝑎𝑝ℎ𝑠 𝑎𝑟𝑒
𝑎𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒
Re,p ε 𝐽𝐷
𝐽𝐷
−0.51
< 190 0.37 1.66(𝑅𝑒 , 𝑝)
> 190 0.983(𝑅𝑒 , 𝑝)−0.4

𝑅𝑒
In general, 𝐽𝐷 (𝑅𝑒 , 𝑑𝑝 , 𝑆𝑐, 𝑡𝑦𝑝𝑒𝑠 𝑜𝑓 𝑓𝑙𝑢𝑖𝑑𝑠, 𝑇)

Non-isothermal:

𝑅𝑒 , 𝑝 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚2 : 𝐽𝑚 = 𝑘𝑚 (𝐶𝑏 − 𝐶𝑠 ) = 𝑅(𝐶𝑠 , 𝑇𝑠 )

𝐶𝑏 𝑐𝑎𝑙 ⁄𝑠 − 𝑚2 : 𝐽ℎ = ℎ(𝑇𝑏 − 𝑇𝑠 ) = 𝑅(𝐶𝑠 , 𝑇𝑠 ) × Δ𝐻(𝑇𝑠 )
[𝑘(𝑇𝑠 )]
𝐶𝑠
𝑇𝑏
𝑇𝑠 (𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐)

𝑇𝑠 (𝑒𝑛𝑑𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐)
𝑇𝑏

References are available to determine ℎ 𝑎𝑛𝑑 𝑘𝑚 .

74
 Isothermal nth order reaction with interphase transport

𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑟𝑎𝑡𝑒, 𝑟 ≡ 𝑘𝐶 𝑛 (𝒎𝒐𝒍𝒆𝒔⁄𝒔 − 𝒎𝟑 𝒐𝒇 𝒄𝒂𝒕) 𝑁𝑜𝑡𝑒:

𝑟 = 𝑘𝑠 𝑎𝐶 𝑛
𝐴 → 𝑅: 𝑜𝑟 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚2 𝑜𝑓 𝑐𝑎𝑡 }
6
𝑜𝑟 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑔 𝑜𝑓 𝑐𝑎𝑡 𝑒𝑓𝑓𝑒𝑐𝑡 𝑜𝑓 ′𝑎′ (𝑑𝑝)
𝑜𝑟 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑠𝑖𝑧𝑒

= 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) (1)

𝑚2 ⁄𝑚3 𝑜𝑓 catalyst volume

𝐶𝑠 is unknown & must be substituted to determine rate of reaction. Reaction takes place at or
near the surface-concentration level.

Define:

a) 𝑅𝑏 = 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑟𝑎𝑡𝑒 𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑⁄𝑎𝑝𝑝𝑎𝑟𝑒𝑛𝑡

= 𝑘𝐶𝑏 𝑛 (𝑁𝑜 Δ𝐶 𝑎𝑐𝑟𝑜𝑠𝑠 𝑡ℎ𝑒 𝑓𝑖𝑙𝑚) /𝑔𝑙𝑜𝑏𝑎𝑙/𝑡𝑟𝑢𝑒
𝑅
b) 𝜂 = 𝑒𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 𝑓𝑎𝑐𝑡𝑜𝑟 = 𝑅
𝑏

𝑜𝑟 𝑟𝑎𝑡𝑒 𝑎𝑡 𝑏𝑢𝑙𝑘 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠

𝑖. 𝑒, 𝑟𝑎𝑡𝑒 𝑤⁄𝑜 𝑒𝑥𝑡𝑒𝑟𝑛𝑎𝑙 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑟𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒

𝑘𝐶𝑠 𝑛 𝐶𝑠 𝑛 (𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙)
= 𝑛 =( )
𝑘𝐶𝑏 𝐶 𝑏 𝑇𝑏 = 𝑇𝑠
𝐷𝑎 = 𝐷𝑎𝑚𝑘𝑜ℎ𝑙𝑒𝑟 𝑛𝑢𝑚𝑏𝑒𝑟 𝑘𝐶𝑏 𝑛−1
c) =( )
𝑘𝑚 𝑎
(Definition of 𝐷𝑎 originates from the non-dimensionalization of equation (1) on
substitution of 𝐶𝑠 (intermediate variable) using definition of 𝜂 in (b) above)

Let us revert to eq (1)

1⁄
𝑘𝑚 𝑎 (𝐶𝑏 − 𝐶𝑏 𝜂 𝑛) = (𝑘𝐶𝑏 𝑛 )𝜂

Non-dimensionalize/re-arrange:

1⁄ 𝑘𝐶𝑏 𝑛−1
1−𝜂 𝑛 =( )𝜂
𝑘𝑚 𝑎

1⁄ 𝑘𝐶𝑏 𝑛−1
1−𝜂 𝑛 = 𝐷𝑎 𝜂 where, 𝐷𝑎 = 𝑘𝑚 𝑎

75
Note: 𝐷𝑎 is large, >>1 ; transport or diffusion step controls (slow) and kinetics is fast.

𝐷𝑎 ≪ 1 ⇒ kinetic controlled (slow) or diffusion rate is fast.

1⁄
Or 𝐷𝑎 𝜂 + 𝜂 𝑛 − 1 = 0; 0 ≤ 𝐷𝑎 ≤ ∞

Plot: 𝑛 < 0 (diffusion helps the reaction)

𝑛 = 0 (no − dependence on
1.0 concentration)
𝜂
𝑛 > 0 (diffusion hurts the reaction)
kinetic controlled

𝐷𝑎
One can also write: 𝑛<0
1.0
𝜂 = (1 − 𝜂𝐷𝑎 )𝑛 𝑛=0 (𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑜𝑛𝑙𝑦 𝑜𝑛𝑒 𝑝𝑜𝑖𝑛𝑡)
𝜂
45𝑜 (𝐶ℎ𝑒𝑐𝑘) 𝑛>0

Concentration profiles: 𝜂 𝐷𝑎

(𝑟 = 𝑘𝐶 𝑛 ; 𝐴 → 𝑅)

𝑅𝑒,𝑝
𝐶𝑏

𝐶𝑠

′ 𝑘𝑚 ′

𝑁𝑜𝑡𝑒: 𝐼𝑓 𝑅𝑒 ↑ 𝑘𝑚 𝑎 ↑ Δ𝐶 ↓ 𝐶𝑠 → 𝐶𝑏 𝑟 ↑ 𝑎𝑛𝑑 𝑟 → 𝑘𝐶𝑏 𝑛

}
𝐷𝑎 ↓ 𝜂 → 1 𝐶𝑠 → 𝐶𝑏 𝑜𝑟 𝑟 → 𝑘𝐶𝑏 𝑛

76
 Lecture 16
Special Cases (continued…)

1st order ′𝑅′ = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑘𝐶𝑠 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 )

= 𝐾𝑝 𝐶𝑏 (Definition of overall mass transfer coefficient)

1 1 1
Therefore, = +
𝐾𝑝 𝑘𝑚 𝑎 𝑘

𝐸
𝑎𝑘𝑘𝑚 𝑎𝑘𝑚 𝐴𝑒 − ⁄𝑅𝑇 𝐸′⁄
𝐾𝑝 = 𝑎𝑘 = −𝐸⁄𝑅𝑇
= 𝐴′𝑒 − 𝑅𝑇
𝑚 +𝑘 𝑎𝑘𝑚 + 𝐴𝑒

(𝑁𝑜𝑡𝑒: 𝑘𝑚 ℎ𝑎𝑠 𝑎 𝑤𝑒𝑎𝑘 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 𝑜𝑛 𝑇)

If 𝑘𝑚 𝑎 > > 𝑘 𝐾𝑝 = 𝑘 ; 𝑅 ≡ 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 (𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑖𝑠 𝑓𝑎𝑠𝑡) = 𝑘𝐶𝑏

In other words, 𝐶𝑏 − 𝐶𝑠 ≈ 0 (No ′Δ𝐶′ or concentration 𝑑𝑟𝑜𝑝)

𝐶𝑠 → 𝐶𝑏

𝐶𝑏
𝐶𝑠

If 𝑘𝑚 𝑎 < < 𝑘 𝐾𝑝 = 𝑘𝑚 𝑎 ; 𝑅 ≡ 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 (𝑘𝑖𝑛𝑒𝑡𝑖𝑐

𝑖𝑠 𝑓𝑎𝑠𝑡) = 𝑘𝑚 𝑎𝐶𝑏

𝐶𝑠 → 0 𝑜𝑟 𝐶𝑏 − 𝐶𝑠 ≈ 𝐶𝑏

𝐶𝑏

𝐶𝑠 --> 0

Thus, the effects (mass transfer vis a vis kinetics) earlier explained through Da have now been
explained using or defining Kp.

77
Temp - dependence:

~𝑡𝑟𝑢𝑒 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛
𝐻𝑖𝑔ℎ 𝑇
𝑒𝑛𝑒𝑟𝑔𝑦
ln(𝐾𝑝 ) (𝑑𝑖𝑓𝑓. 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 𝐿𝑜𝑤 𝑇
(𝑒𝑥𝑝𝑡) 𝐸𝑎𝑝𝑝 ~ 0 (𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
(𝑤𝑒𝑎𝑘 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒)

1⁄
𝑇

Non-isothermal + interphase transport

(𝛿𝑚 , 𝛿ℎ ) ≡ 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝛿𝑚 = 𝛿𝑚 (𝑅𝑒 , 𝑆𝑐)

( )
𝛿ℎ = 𝛿ℎ (𝑅𝑒 , 𝑃𝑟)

Mass – transfer always hurts the rate

𝐶𝑏 𝐶𝑠 < 𝐶𝑏
𝑇𝑠
Heat – transfer may or may not hurt the rate
𝐶𝑠
𝑇𝑠 > 𝑇𝑏 (𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐): ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 ℎ𝑒𝑙𝑝𝑠 𝑟𝑎𝑡𝑒
𝑇𝑏
𝑇𝑠

𝑇𝑠 < 𝑇𝑏 (𝑒𝑛𝑑𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐): ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 ℎ𝑢𝑟𝑡𝑠 𝑟𝑎𝑡𝑒

nth order reaction rate

𝑅 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑘𝑠 𝐶𝑠 𝑛 (𝑚𝑜𝑙𝑒𝑠⁄ )
𝑠 − 𝑚3 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
𝑅𝑜𝑏𝑠 ⁄𝑅𝑔𝑙𝑜𝑏𝑎𝑙 𝑘𝑠 𝐶𝑠 𝑛 𝑘 𝐶 𝑛
𝜂 (𝑑𝑒𝑓𝑖𝑛𝑡𝑖𝑜𝑛) = = = ( 𝑠) ( 𝑠)
𝑅𝑏 𝑘𝑏 𝐶𝑏 𝑛 𝑘𝑏 𝐶𝑏

𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑏𝑢𝑙𝑘
𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 (𝑤𝑖𝑡ℎ𝑜𝑢𝑡
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛)
78
 On substitution,
1⁄
𝑘𝑏 𝑛
𝑘𝑚 𝑎𝐶𝑏 [1 − ( 𝜂)
𝑘𝑠
] = 𝜂𝑘𝑏 𝐶𝑏 𝑛 (𝐶𝑠 is substituted in terms of 𝐶𝑏 )

1⁄
𝑘𝑏 𝑛 𝑘𝑏 𝐶𝑏 𝑛−1
⇒1−( 𝜂) = 𝜂( )
𝑘𝑠 𝑘𝑚 𝑎

𝑘𝑠 𝐸 𝑇
𝜂= (1 − 𝜂𝐷𝑎 )𝑛 = (1 − 𝜂𝐷𝑎 )𝑛 𝑒𝑥𝑝 (− ( 𝑇𝑏 − 1))
𝑘𝑏 𝑅𝑇𝑏 𝑠

Note: If isothermal, Ts = Tb, the equation is reduced to the isothermal case.

𝑘
To determine ( 𝑠 ) (𝑘𝑠 (𝑇𝑠 ) is unknown), energy balance equation:
𝑘 𝑏

𝑅(+Δ𝐻) = ℎ𝑎 (𝑇𝑏 − 𝑇𝑠 ): 𝑒𝑛𝑒𝑟𝑔𝑦 𝑏𝑎𝑙𝑎𝑛𝑐𝑒 (𝑐𝑎𝑙⁄

𝑠 − 𝑚3 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 )

𝑐𝑎𝑙⁄ 𝑚2⁄
𝑚𝑜𝑙𝑒𝑠⁄ 𝑚𝑜𝑙𝑒𝑠 𝑚3
𝑠 − 𝑚3
𝑐𝑎𝑙⁄
𝑚2 − 𝑠 − 𝐾

𝑅 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) ℎ𝑒𝑎𝑡 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛

𝑇𝑠 −𝑇𝑏 𝐶𝑏 −𝐶𝑠 𝑘𝑚 (−Δ𝐻)𝐶𝑏
𝑟𝑎𝑡𝑒
𝑠ubstitute, = ×
𝑇𝑏 𝐶𝑏 ℎ𝑇𝑏

ℎ𝑒𝑎𝑡 𝑑𝑖𝑠𝑠𝑖𝑝𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒

Define 𝛽 (𝑃𝑟𝑎𝑡𝑒𝑟 #; 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡 𝑜𝑓 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙𝑖𝑡𝑦)
𝑇𝑠 𝐶𝑠
Therefore, ( − 1) = (1 − )𝛽
𝑇𝑏 𝐶𝑏

𝑇𝑏 1
or, = 𝐶
𝑇𝑠 1+(1− 𝑠 )𝛽
𝐶𝑏

Tb 1
It can be shown that = (try yourself as a homework)
Ts 1+βDa η

79
 𝐸 1
Therefore, 𝜂 = (1 − 𝜂𝐷𝑎 )𝑛 𝑒𝑥𝑝 (− (1+𝛽𝐷 − 1))
𝑅𝑇𝑏 𝑎𝜂

𝐸
𝑁𝑜𝑡𝑒: 𝜂 = 𝜂 (𝑛, 𝐷𝑎 , 𝛽, 𝑅𝑇 )
𝑏
𝐴𝑟𝑟ℎ𝑒𝑛𝑖𝑢𝑠 #

Compare to 𝜂𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙 = 𝜂(𝑛, 𝐷𝑎 ) (𝛽 = 0)

In either case (isothermal or non-isothermal), you are interested in calculating

Robs by determining rate based on bulk conditions and then correcting or
modifying it with the effectiveness factor as Robs = 𝜼 Rb

𝑪𝒂𝒕𝒂𝒍𝒚𝒔𝒕 𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏

C𝑏
Δ𝐻 − 𝑣𝑒 T𝑠
T𝑠 > T𝑏
𝛽 > 0 (𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐) C𝑠 < C𝑏
T𝑏 C𝑠
n<0
𝐵𝑜𝑡ℎ
𝑚𝑎𝑠𝑠 & ℎ𝑒𝑎𝑡
𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 ℎ𝑒𝑙𝑝
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒
η 𝑛 = 1⁄2
1 }𝑛 > 0
1.0
𝑛=2
(n>1)
𝑃𝑟𝑜𝑐𝑒𝑠𝑠 𝑖𝑠 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑩𝒆𝒔𝒕 𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏: 𝒎𝒂𝒙𝒊𝒎𝒖𝒎 𝒓𝒂𝒕𝒆
𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑
(𝑺𝒐𝒎𝒆 𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝒊𝒔
(𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒
𝒑𝒓𝒆𝒇𝒇𝒆𝒓𝒓𝒆𝒅 𝒐𝒓 𝒅𝒆𝒔𝒊𝒓𝒂𝒃𝒍𝒆!)
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑖𝑠 𝑙𝑎𝑟𝑔𝑒) ( 𝑖𝑠 ℎ𝑢𝑟𝑡 𝑏𝑦 𝑑𝑟𝑜𝑝 𝑖𝑛 𝐶𝑠 )
𝑎𝑡 ℎ𝑖𝑔ℎ 𝐷𝑎

η𝑎 D𝑎

 𝜂 indicates how effective the catalyst is relative to the bulk conditions

 For isothermal operation D𝑎 < < 1 to get 𝜂 = 1 (𝑛 > 0)
 For non-isothermal condition, operate at same intermediate D𝑎 (bring some
diffusion resistance!) to get 𝜂 > 1.
80
⇒ How about endothermic reaction? 𝛽 < 0 (Δ𝐻; +𝑣𝑒) : Not interesting (results
are straightforward);

Some more analysis:

𝜂 = 𝜂(D𝑎 , 𝑛, 𝛽, 𝜖)
Δ𝑇 Δ𝐶
− = 𝛽 (𝑅𝑒𝑐𝑎𝑙𝑙)
𝑇𝑏 𝐶𝑏

How to find 𝛽 ?

𝑘𝑚 (−Δ𝐻)𝐶𝑏
𝛽= (𝐶𝑏 , 𝑇𝑏 𝑎𝑟𝑒 𝑘𝑛𝑜𝑤𝑛, Δ𝐻 is a thermodynamic property ,
ℎ𝑇𝑏
𝑘𝑚 , ℎ ⇒ 𝑘𝑚 (𝑆ℎ), ℎ(𝑁𝑢) : easy to calculate
−Δ𝐻 𝐶𝑏 𝑘𝑚
= ℎ
𝜌𝐶𝑝 𝑇𝑏 ⁄𝜌𝐶
𝑝

2⁄ 2⁄ 𝑅𝑒𝑓𝑒𝑟 𝐵𝑆𝐿 𝑓𝑜𝑟

𝑘𝑚 𝜌 𝜈 3 ℎ 𝜈 3
𝐽𝐷 =
𝐺
(𝐷) ; 𝐽𝐻 =
𝐺𝐶𝑝 𝛼
( ) ( 𝑎𝑛𝑎𝑙𝑜𝑔𝑦 𝑏𝑒𝑤𝑒𝑒𝑛 )
𝑆𝑐 #
ℎ𝑒𝑎𝑡 & 𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡
𝑃𝑟 #
2⁄
𝑘𝑚 𝐽𝐷 𝐷 3 𝛼
Take the ratio: ℎ⁄ = (𝛼 ) (𝐿𝑒 = 𝐷 = 𝐿𝑒𝑤𝑖𝑠 #)
𝜌𝐶𝑝 𝐽𝐻

(−Δ𝐻) 𝐶𝑏 𝐽𝐷 −2⁄
Therefore, 𝛽 = (𝐿𝑒 ) 3 ∶ 𝐷𝑎𝑚𝑘𝑜ℎ𝑙𝑒𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
𝜌𝐶𝑝 𝑇𝑏 𝐽𝐻

( 𝐽𝐷 , 𝐽𝐻 ≡ 𝑡ℎ𝑒𝑟𝑒 𝑎𝑟𝑒 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛𝑠 𝑟𝑒𝑝𝑜𝑟𝑡𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑙𝑖𝑡𝑒𝑟𝑎𝑡𝑢𝑟𝑒)

𝑓
For many practical cases (ℎ𝑖𝑔ℎ 𝑅𝑒 ): 𝐽𝐷 = 𝐽𝐻 = ⁄2

(−Δ𝐻) 𝐶𝑏 𝑃𝑟 ~ 1 − 1000
Therefore, 𝛽 = ( )
𝜌𝐶𝑝 𝑇𝑏 𝑆𝑐 ~ 1 − 1000
𝑇𝑠 −𝑇𝑏 Δ𝐻
and ( )=− (thus simplified)
𝐶𝑏 −𝐶𝑠 𝜌𝐶𝑝

HW (SMITH, 3rd ed): 10.3, 10.5

81
 Lecture 17
Non-porous catalyst (interphase + reaction)

-----continued

Different cases (non-isothermal):

𝑘𝑠 𝐶𝑠 𝑛
𝜂 = ( )( )
𝑘𝑏 𝐶𝑏

(1) 𝛽 > 0 and 𝑛 < 0 ⇒ 𝜂 > 1 ∶ 𝑅 ℎ𝑒𝑎𝑡 ↑ 𝑅 𝑚𝑎𝑠𝑠 ↑

𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡

(2) 𝛽 > 0 and 𝑛 > 0 ⇒ 𝜂 > 1 𝑜𝑟 𝜂 < 1 depending upon 𝐷𝑎 ∶ 𝑅𝐻 ↑ 𝑅𝑚 ↓

(3) β < 0 and n > 0 ⇒ η < 1 ∶ RH ↓ Rm ↓
(4) 𝛽 < 0 and 𝑛 < 0 ⇒ 𝜂 > 1 𝑜𝑟 𝜂 < 1 ∶ 𝑅𝐻 ↓ 𝑅𝑚 ↑

Special Cases: Non - isothermal (1st order)

Mass Flux: (moles⁄ r = 𝑘𝑚 (Cb − Cs ) = 𝑘𝑠 Cs

s − m2 )

(moles⁄ ) 𝑅 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑎𝑘𝑠 𝐶𝑠 = 𝑘𝑠 ′𝐶𝑠

s − m3

𝑚2⁄
𝑚3
Therefore,
𝑘𝑚
𝐶𝑠 = ( ) 𝐶𝑏
𝑘𝑚 +𝑘𝑠

𝑎𝑘𝑠 𝑘𝑚 (𝑠𝑎𝑚𝑒 𝑎𝑠 𝑏𝑒𝑓𝑜𝑟𝑒,

𝑅=( ) 𝐶𝑏
𝑘𝑚 +𝑘𝑠 𝑖. 𝑒. 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

82
Heat Transport:

𝑄
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛/𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 = 𝑅 (−𝛥𝐻) (𝑐𝑎𝑙⁄ )
𝑠−𝑚3 𝑜𝑓 𝑐𝑎𝑡 𝑣𝑜𝑙
𝑚𝑜𝑙𝑒⁄ 𝑐𝑎𝑙⁄
𝑠 − 𝑚3 𝑠 − 𝑚𝑜𝑙𝑒
𝑄
{ 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟𝑟𝑒𝑑/𝑟𝑒𝑚𝑜𝑣𝑒𝑑 = 𝑎ℎ(𝑇𝑠 −𝑇𝑏 ) (𝑐𝑎𝑙⁄ )
𝑠−𝑚3 𝑜𝑓 𝑐𝑎𝑡 𝑣𝑜𝑙
𝑐𝑎𝑙⁄
𝑠 − 𝑘 − 𝑚2

−Δ𝐻 𝑎 𝑘𝑚 𝑘𝑠 (−Δ𝐻)𝑎 𝐶𝑏
𝑜𝑟, 𝑄𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 = ( ) 𝐶𝑏 ≡ [ ] (𝑐𝑎𝑙⁄ )
𝑘𝑚 + 𝑘𝑠 1
𝑘𝑚
+
1
𝐸
𝑠 − 𝑚3
𝐴 𝑒𝑥𝑝(−𝑅𝑇 )
𝑠

𝑆𝑆 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛

𝐻𝑜𝑡 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
Investigate under SS:

 Multiplicity (3 solutions) 𝑄𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟𝑟𝑒𝑑 (𝑙𝑖𝑛𝑒𝑎𝑟)

cold/hot/unstable solutions 𝑄
𝑢𝑛𝑠𝑡𝑎𝑏𝑙𝑒( 𝑖𝑔𝑛𝑖𝑡𝑖𝑜𝑛 𝑝𝑡)
 Mass transport: 𝑐𝑜𝑙𝑑
weak/linear dependence on T 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
𝑄𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 (𝑠𝑖𝑔𝑚𝑜𝑖𝑑𝑎𝑙 𝑠ℎ𝑎𝑝𝑒)
 Heat transport (reaction):
Exponential dependence
𝑇𝑠 − 𝑇𝑏
𝑇𝑏

Recall: Similar situation in the exothermic CSTR operation: instability is an issue

83
Example 1: A well-mixed basket type reactor is used for carrying out the reaction
𝐴 → 𝑃 over a non-porous catalyst under constant flow-condition. The following
data are available:

𝑣𝑜 = 𝑐𝑜𝑛𝑠𝑡
𝑇 = 𝑐𝑜𝑛𝑠𝑡 𝑏𝑢𝑙𝑘 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠
𝑑𝑝 W 𝑋𝐴 𝐶𝐴𝑜 = 𝑐𝑜𝑛𝑠𝑡
1 𝑚𝑜𝑙𝑒/𝑙𝑡
2 2 0.6
6 1 0.2
𝐶𝐴 ,
𝑋𝐴

What controls the reaction? Interphase transport or kinetic rate?

Ans: Write down the species balance equation across the reactor.
𝑚𝑜𝑙𝑒𝑠
: (−𝑟)𝑜𝑏𝑠 𝑊 = 𝑣𝑜 𝐶𝐴𝑜 𝑋𝐴 𝐶𝐴 = 𝐶𝐴𝑜 (1 − 𝑋𝐴 )
𝑠

𝑚𝑜𝑙𝑒𝑠⁄ (𝑣𝑜 𝐶𝐴𝑜 − 𝑣𝑜 𝐶𝐴 )

𝑠 − 𝑔𝑚 𝑔𝑚
𝑜𝑓 𝑐𝑎𝑡 𝐼𝑛 𝑂𝑢𝑡

𝑣𝑜 𝐶𝐴𝑜 𝑋𝐴
(−𝑟)𝑜𝑏𝑠 =
𝑊

(−𝑟)𝑜𝑏𝑠,1 𝑋 𝑊 0.6 1
(−𝑟)𝑜𝑏𝑠,2
= ( 𝐴) × ( ) = × = 3⁄2 (from the measurements)
𝑊 1 𝑋𝐴 2 2 0.2

Recall: If external mass transfer/interphase rate controls or kinetics is fast

𝑟𝑜𝑏𝑠 = 𝑘𝑚 𝑎𝐶𝑏 (𝐶𝑠 → 0)

(𝑟𝑜𝑏𝑠 )1 𝑎1 (1−𝑋𝐴 )1 𝑑𝑝2 (1−𝑋𝐴 )1 6 0.4 𝜎

(𝑟𝑜𝑏𝑠 )2
= = = × = 3⁄2 (𝑎 = )
𝑎2 (1−𝑋𝐴 )2 𝑑𝑝1 (1−𝑋𝐴 )2 2 0.8 𝑑𝑝

(𝐴𝑠𝑠𝑢𝑚𝑒 𝑘𝑚1 = 𝑘𝑚2 𝑢𝑛𝑑𝑒𝑟 𝑡𝑤𝑜 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠) 𝑚2⁄

𝑚3

𝑨𝒏𝒔: Interphase mass transfer controls

84
 𝐾𝑝1 (𝑘 𝑎)1 𝑎1 𝑑𝑝2 6
Alternatively, = (𝑘𝑚 = = = =3
𝐾𝑝2 𝑚 𝑎)2 𝑎2 𝑑𝑝1 2
recall def n of } (𝑟𝑜𝑏𝑠 )1 𝐾𝑝 𝐶𝑏 1 𝐾𝑝 𝐶𝐴𝑜 (1−𝑋1 ) 0.4
overall mass transfer (𝑟𝑜𝑏𝑠 )2
= 1
= 1
=3× = 3⁄2
𝐾𝑝 𝐶𝑏 2 𝐾𝑝 𝐶𝐴𝑜 (1−𝑋2 ) 0.8
2 2
coefficient

Ans: (Interphase mass transfer controls)

𝐄𝐱𝐚𝐦𝐩𝐥𝐞 𝟐: 1 𝑐𝑎𝑡
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛: 𝑆𝑂2 + 𝑂2 → 𝑆𝑂3 (𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙)
(𝑆𝑚𝑖𝑡ℎ 10.1) 2

𝐶𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠
6.42% 𝑆𝑂2 𝑃 = 790 𝑚𝑚 𝐻𝑔
93.50% 𝐴𝑖𝑟 𝑇 = 480𝑜 𝐶
𝐺 = 0.199 𝑘𝑔/𝑚2 − 𝑠 𝑛𝑜𝑛𝑝𝑜𝑟𝑜𝑢𝑠 𝐴𝑙2 𝑂3 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
′′ ′′
(1⁄8 × 1⁄8 𝑐𝑦𝑙𝑖𝑛𝑑𝑟𝑖𝑐𝑎𝑙
𝑘𝑔
𝑑𝑖𝑓𝑓𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑓𝑖𝑥𝑒𝑑 𝑏𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝑝𝑒𝑙𝑙𝑒𝑡𝑠, 𝜌 = 950 ⁄𝑚3 )
𝜖𝑏 = 0.43
𝐴𝑠𝑠𝑢𝑚𝑒 𝑆𝑐 = 1.2
𝑘𝑔
𝜌(𝑎𝑖𝑟) = 0.487 ⁄𝑚3
{𝜇 (𝑎𝑖𝑟) = 3.72 × 10−5 𝑃𝑎 − 𝑠

Data:

𝑋𝑆𝑂 𝑟(
𝑔 𝑚𝑜𝑙𝑒𝑠 𝑆𝑂2
)
𝑝𝑏 , 𝑎𝑡𝑚
2 ℎ−𝑔 𝑜𝑓 𝑐𝑎𝑡
𝑺𝑶𝟐 𝑺𝑶𝟑 𝑶𝟐
0.1 0.0956 0.0603 0.0067 0.201

(𝑝𝑏 − 𝑝𝑠 )⁄
Determine (𝐶𝑏 − 𝐶𝑠 ) 𝑜𝑟 (𝑝𝑏 − 𝑝𝑠 ) 𝑜𝑟 𝑝𝑏

Ans: 𝑟𝑜𝑏𝑠 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) 𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚3

(= 𝑘 𝐶𝑠 𝑛 )

85
Note: 𝐷𝑎 is large, >>1 ; transport or diffusion step controls (slow) and kinetics is fast.

𝐷𝑎 ≪ 1 ⇒ kinetic controlled (slow) or diffusion rate is fast.

1⁄
Or 𝐷𝑎 𝜂 + 𝜂 𝑛 − 1 = 0; 0 ≤ 𝐷𝑎 ≤ ∞

Plot: 𝑛 < 0 (diffusion helps the reaction)

𝑛 = 0 (no − dependence on
1.0 concentration)
𝜂
𝑛 > 0 (diffusion hurts the reaction)
kinetic controlled

𝐷𝑎
One can also write: 𝑛<0
1.0
𝜂 = (1 − 𝜂𝐷𝑎 )𝑛 𝑛=0 (𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑜𝑛𝑙𝑦 𝑜𝑛𝑒 𝑝𝑜𝑖𝑛𝑡)
𝜂
45𝑜 (𝐶ℎ𝑒𝑐𝑘) 𝑛>0

Concentration profiles: 𝜂 𝐷𝑎

(𝑟 = 𝑘𝐶 𝑛 ; 𝐴 → 𝑅)

𝑅𝑒,𝑝
𝐶𝑏

𝐶𝑠

′ 𝑘𝑚 ′

𝑁𝑜𝑡𝑒: 𝐼𝑓 𝑅𝑒 ↑ 𝑘𝑚 𝑎 ↑ Δ𝐶 ↓ 𝐶𝑠 → 𝐶𝑏 𝑟 ↑ 𝑎𝑛𝑑 𝑟 → 𝑘𝐶𝑏 𝑛

}
𝐷𝑎 ↓ 𝜂 → 1 𝐶𝑠 → 𝐶𝑏 𝑜𝑟 𝑟 → 𝑘𝐶𝑏 𝑛

76
 Lecture 16
Special Cases (continued…)

1st order ′𝑅′ = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑘𝐶𝑠 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 )

= 𝐾𝑝 𝐶𝑏 (Definition of overall mass transfer coefficient)

1 1 1
Therefore, = +
𝐾𝑝 𝑘𝑚 𝑎 𝑘

𝐸
𝑎𝑘𝑘𝑚 𝑎𝑘𝑚 𝐴𝑒 − ⁄𝑅𝑇 𝐸′⁄
𝐾𝑝 = 𝑎𝑘 = −𝐸⁄𝑅𝑇
= 𝐴′𝑒 − 𝑅𝑇
𝑚 +𝑘 𝑎𝑘𝑚 + 𝐴𝑒

(𝑁𝑜𝑡𝑒: 𝑘𝑚 ℎ𝑎𝑠 𝑎 𝑤𝑒𝑎𝑘 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 𝑜𝑛 𝑇)

If 𝑘𝑚 𝑎 > > 𝑘 𝐾𝑝 = 𝑘 ; 𝑅 ≡ 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 (𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑖𝑠 𝑓𝑎𝑠𝑡) = 𝑘𝐶𝑏

In other words, 𝐶𝑏 − 𝐶𝑠 ≈ 0 (No ′Δ𝐶′ or concentration 𝑑𝑟𝑜𝑝)

𝐶𝑠 → 𝐶𝑏

𝐶𝑏
𝐶𝑠

If 𝑘𝑚 𝑎 < < 𝑘 𝐾𝑝 = 𝑘𝑚 𝑎 ; 𝑅 ≡ 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 (𝑘𝑖𝑛𝑒𝑡𝑖𝑐

𝑖𝑠 𝑓𝑎𝑠𝑡) = 𝑘𝑚 𝑎𝐶𝑏

𝐶𝑠 → 0 𝑜𝑟 𝐶𝑏 − 𝐶𝑠 ≈ 𝐶𝑏

𝐶𝑏

𝐶𝑠 --> 0

Thus, the effects (mass transfer vis a vis kinetics) earlier explained through Da have now been
explained using or defining Kp.

77
Temp - dependence:

~𝑡𝑟𝑢𝑒 𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛
𝐻𝑖𝑔ℎ 𝑇
𝑒𝑛𝑒𝑟𝑔𝑦
ln(𝐾𝑝 ) (𝑑𝑖𝑓𝑓. 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 𝐿𝑜𝑤 𝑇
(𝑒𝑥𝑝𝑡) 𝐸𝑎𝑝𝑝 ~ 0 (𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
(𝑤𝑒𝑎𝑘 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒)

1⁄
𝑇

Non-isothermal + interphase transport

(𝛿𝑚 , 𝛿ℎ ) ≡ 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡 𝛿𝑚 = 𝛿𝑚 (𝑅𝑒 , 𝑆𝑐)

( )
𝛿ℎ = 𝛿ℎ (𝑅𝑒 , 𝑃𝑟)

Mass – transfer always hurts the rate

𝐶𝑏 𝐶𝑠 < 𝐶𝑏
𝑇𝑠
Heat – transfer may or may not hurt the rate
𝐶𝑠
𝑇𝑠 > 𝑇𝑏 (𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐): ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 ℎ𝑒𝑙𝑝𝑠 𝑟𝑎𝑡𝑒
𝑇𝑏
𝑇𝑠

𝑇𝑠 < 𝑇𝑏 (𝑒𝑛𝑑𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐): ℎ𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 ℎ𝑢𝑟𝑡𝑠 𝑟𝑎𝑡𝑒

nth order reaction rate

𝑅 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑘𝑠 𝐶𝑠 𝑛 (𝑚𝑜𝑙𝑒𝑠⁄ )
𝑠 − 𝑚3 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
𝑅𝑜𝑏𝑠 ⁄𝑅𝑔𝑙𝑜𝑏𝑎𝑙 𝑘𝑠 𝐶𝑠 𝑛 𝑘 𝐶 𝑛
𝜂 (𝑑𝑒𝑓𝑖𝑛𝑡𝑖𝑜𝑛) = = = ( 𝑠) ( 𝑠)
𝑅𝑏 𝑘𝑏 𝐶𝑏 𝑛 𝑘𝑏 𝐶𝑏

𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑏𝑢𝑙𝑘
𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 (𝑤𝑖𝑡ℎ𝑜𝑢𝑡
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛)
78
Special Case: 1st order reaction (analytical solution is available)

𝐷𝑒 ∇2 𝑓 = 𝑘𝑓
𝑑2𝑓 2 𝑑𝑓
𝐷𝑒 (
𝑑𝑧 2
+
𝑧 𝑑𝑧
) = 𝑘𝐿2 𝑓

𝑑2𝑓 2 𝑑𝑓 𝑘 𝑟𝑜
+ = 𝜙 2 𝑓; 𝜙 = 𝐿√ (𝐿 = 𝑓𝑜𝑟 𝑠𝑝ℎ𝑒𝑟𝑒)
𝑑𝑧 2 𝑧 𝑑𝑧 𝐷𝑒 3

𝐵𝐶𝑠 𝑧 = 0, ∇𝑓 = 0 𝑟
(𝑧 = 𝑟⁄𝐿 = 𝑜⁄𝑟𝑜 = 3)
𝑧=3 𝑓=1 ⁄3
𝑔
Let 𝑔 = 𝑓𝑧 ; 𝑓 = ⁄𝑧

Differentiate f twice wrt z, simplify and substitute in the equation above to

show the following transformation:

𝐵𝐶𝑠: 𝑧=0 , 𝑔=0

𝑑2𝑔 2 }
−𝜙 𝑔 =0 𝑧=3 , 𝑔=3
𝑑𝑧 2

𝑔 = 𝐶1 𝑠𝑖𝑛ℎ𝜙𝑧 + 𝐶2 𝑐𝑜𝑠ℎ𝜙𝑧

𝐶2 = 0, 𝐶1 = 3⁄𝑠𝑖𝑛ℎ3𝜙 (apply BCs)

3𝑠𝑖𝑛ℎ(𝜙𝑧)
𝑔=
𝑠𝑖𝑛ℎ(3𝜙)

3𝑠𝑖𝑛ℎ(𝜙𝑧)
𝑓= ∶ (dimensionless solution)
𝑧 𝑠𝑖𝑛ℎ(3𝜙)

𝐶𝐴 3 𝑠𝑖𝑛ℎ(𝜙𝑟⁄𝐿) 𝑟𝑜 𝑠𝑖𝑛ℎ(𝜙3𝑟⁄𝑟𝑜 )
= 𝑟⁄ 𝑠𝑖𝑛ℎ(3𝜙) = ⇒ Thus, 𝐶𝐴 (𝑟) is determined.
𝐶𝐴,𝑏 𝐿 𝑟 𝑠𝑖𝑛ℎ(3𝜙)

Calculate the reaction rate by equating the rate of species arriving at the surface
of the catalyst to that of species consumed/reacted in the catalyst volume under
SS conditions.

89
 𝑚𝑜𝑙𝑒𝑠 𝑟
𝑅( ) = ∫0 𝑜 𝑘𝐶𝐴 (𝑟) 4𝜋 𝑟 2 𝑑𝑟 ("A" is 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 𝑝𝑜𝑟𝑒𝑠)
𝑠

𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 𝑚3
𝐶𝑏
𝑟
𝑑𝐶𝐴 2
= −𝐷𝑒 | 4𝜋𝑟𝑜 ;
𝑑𝑟 𝑟=𝑟𝑜
(𝑓𝑙𝑢𝑥 𝑜𝑓 A 𝑎𝑟𝑟𝑖𝑣𝑖𝑛𝑔 𝑎𝑡 𝑡ℎ𝑒 𝑠𝑢𝑟𝑓𝑎𝑐𝑒) 𝑟𝑜
Δ𝑟

(Note that one can also write:

R global = kCA (r), based on the volume average concentration inside the catalyst)

It is easier to differentiate than to integrate (use the second expression)

𝑑𝐶
𝑅𝑔𝑙𝑜𝑏𝑎𝑙 −𝐷𝑒 𝑑𝑟𝐴 | 4𝜋𝑟𝑜 2
𝑟=𝑟𝑜
𝜂𝑖𝑛𝑡𝑟𝑎 = = 4
𝑅𝑏 (𝑘𝐶𝐴,𝑏 )3𝜋𝑟𝑜 3

= 𝐷𝑒
𝑑(𝐶𝐴 ⁄𝐶𝐴,𝑏 )
| x
1 𝑟
(𝐿 = 𝑜⁄ )
𝑑(𝑟⁄𝐿) 𝑟=𝑟
𝑜
2
𝐿 𝑘 3
1 𝑑𝑓
= |
𝜙2 𝑑𝑧 𝑧=3

Already we have the solution 𝑓(𝑧) from the previous exercise; substitute to obtain:

1 1 1 𝑘
𝜂𝑖𝑛𝑡𝑟𝑎 = [ − ] , 𝜙 = 𝐿√
1⁄ (1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟)
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙 𝐷𝑒
𝑠

𝑚 𝑚2⁄
𝑆

Once we know 𝜼𝒊𝒏𝒕𝒓𝒂 ⇒ 𝒄𝒂𝒍𝒄𝒖𝒍𝒂𝒕𝒆 𝑹𝒈𝒍𝒐𝒃𝒂𝒍 = 𝜼𝒊𝒏𝒕𝒓𝒂 × (𝒌𝑪𝑨,𝒃 ) 𝒎𝒐𝒍𝒆⁄

𝒔 − 𝒎𝟑

90
 𝐸 1
Therefore, 𝜂 = (1 − 𝜂𝐷𝑎 )𝑛 𝑒𝑥𝑝 (− (1+𝛽𝐷 − 1))
𝑅𝑇𝑏 𝑎𝜂

𝐸
𝑁𝑜𝑡𝑒: 𝜂 = 𝜂 (𝑛, 𝐷𝑎 , 𝛽, 𝑅𝑇 )
𝑏
𝐴𝑟𝑟ℎ𝑒𝑛𝑖𝑢𝑠 #

Compare to 𝜂𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙 = 𝜂(𝑛, 𝐷𝑎 ) (𝛽 = 0)

In either case (isothermal or non-isothermal), you are interested in calculating

Robs by determining rate based on bulk conditions and then correcting or
modifying it with the effectiveness factor as Robs = 𝜼 Rb

𝑪𝒂𝒕𝒂𝒍𝒚𝒔𝒕 𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏

C𝑏
Δ𝐻 − 𝑣𝑒 T𝑠
T𝑠 > T𝑏
𝛽 > 0 (𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐) C𝑠 < C𝑏
T𝑏 C𝑠
n<0
𝐵𝑜𝑡ℎ
𝑚𝑎𝑠𝑠 & ℎ𝑒𝑎𝑡
𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 ℎ𝑒𝑙𝑝
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒
η 𝑛 = 1⁄2
1 }𝑛 > 0
1.0
𝑛=2
(n>1)
𝑃𝑟𝑜𝑐𝑒𝑠𝑠 𝑖𝑠 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑩𝒆𝒔𝒕 𝑶𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏: 𝒎𝒂𝒙𝒊𝒎𝒖𝒎 𝒓𝒂𝒕𝒆
𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑
(𝑺𝒐𝒎𝒆 𝒅𝒊𝒇𝒇𝒖𝒔𝒊𝒐𝒏 𝒊𝒔
(𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒
𝒑𝒓𝒆𝒇𝒇𝒆𝒓𝒓𝒆𝒅 𝒐𝒓 𝒅𝒆𝒔𝒊𝒓𝒂𝒃𝒍𝒆!)
𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑖𝑠 𝑙𝑎𝑟𝑔𝑒) ( 𝑖𝑠 ℎ𝑢𝑟𝑡 𝑏𝑦 𝑑𝑟𝑜𝑝 𝑖𝑛 𝐶𝑠 )
𝑎𝑡 ℎ𝑖𝑔ℎ 𝐷𝑎

η𝑎 D𝑎

 𝜂 indicates how effective the catalyst is relative to the bulk conditions

 For isothermal operation D𝑎 < < 1 to get 𝜂 = 1 (𝑛 > 0)
 For non-isothermal condition, operate at same intermediate D𝑎 (bring some
diffusion resistance!) to get 𝜂 > 1.
80
 Note:

Graphical solutions are available for different shapes of the materials and different orders
of reaction.

𝑛 = 0 (two solutions?)
𝑆𝑙𝑎𝑏
1.0 1.0
1 1
1⁄
𝐶𝑦𝑙𝑖𝑛𝑑𝑒𝑟 2
𝑎𝑛𝑑 𝜂
2
𝑆𝑝ℎ𝑒𝑟𝑒

1.0

𝜙 10 𝑛+1 𝑘 10
𝜙 = 𝐿√ 𝐶𝐴 𝑛−1
2 𝐷𝑒

𝑛
Again, calculate 𝜙 first and then 𝜂𝑖𝑛𝑡𝑟𝑎 ⇒ 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒 𝑅𝑔𝑙𝑜𝑏𝑎𝑙 = 𝜂𝑖𝑛𝑡𝑟𝑎 × (𝐶𝐴,𝑏 ) 𝑚𝑜𝑙𝑒⁄𝑠 − 𝑚3

We will close the discussion on intraphase transport + kinetics in this lecture before taking up the
examples in the next few lectures.

Smith has devoted two-three pages (507-510) on the effect of internal transport on selectivity in
chapter 11. It is not difficult for you to go through this section yourself. Briefly, Smith categorizes
the effects through Type 1, 2 and 3 series or parallel reactions.

Type 1: A ⤏ B (desired) and R ⤏ S (undesired)

Type 2: A ⤏ B (desired) and A ⤏ C (undesired)

Type 3: A ⤏ B ⤏ D, where B is the desired product.

Examples are from hydrogenation, dehydrogenation, dehydration, and oxidation of some

hydrocarbons.

Type 1 and 2 are easy to handle and quantify. Type 3 requires some thinking and calculations.
For Type 1 reactions, internal transport or diffusion decreases the selectivity. For Type 2 (both
first order irreversible reactions), selectivity is unaffected by diffusion. Type 3 is slightly trickier
to understand because the product B is the reactant for the second consecutive/series reaction.
Therefore, pore-diffusion coefficient for B becomes as important as the pore-diffusion coefficient
of the reactant A! May be there are two Thiele modulus? And, there may be so many different
situations which you should be able to recognize……

92
 Lecture 19-20
Example 1: Explain 𝑛 = 0, mathematically
𝑇𝑤𝑜 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠:

( 𝜙 ≤ 1, 𝜙 > 1)
1.0

𝜂
𝜂=1 𝜂<1

1.0 𝜙

Hint: For a zero-order reaction, analytically integrate the non-dimensional species conservation
differential equation to derive a general expression for concentration in the pellet. Note that
concentration is minimum at the center; however it cannot be zero. With this constraint you
should be able to derive an expression for concentration profiles inside the pellet with a limit for
𝜙 < Ø*. In this case, the process is kinetic controlled and 𝜂 = 1 (observed rate = intrinsic rate for
zero-order reaction). For sphere Ø* = 1 and for slab Ø* = √2. For 𝜙  Ø*, concentration, however,
decreases to zero at some distance (r*) away from the center inside the pellet, beyond which
region 𝜂 = 1/𝜙 and the process is diffusion controlled. Below that distance, there is no gas
(reactant) and no reaction. See Rodrigues and Orfao (1984) in Chem. Eng. Commun. 27, pp. 327-
33.

𝑘1
Ex 2. For reversible, 𝐴 ⇌ 𝐵 ∶ calculate 𝜂(𝜙).
𝑘−1
𝑑𝐶𝐴
Ans: 𝑟=− = 𝑘1 𝐶𝐴 − 𝑘−1 𝐶𝐵
𝑑𝑡

(For a reversible reaction, recall your previous calculations: 𝐶𝐴 + 𝐶𝐵 = 𝐶 = 𝐶𝐴𝑜 + 𝐶𝐵𝑜

= 𝐶𝐴𝑒 + 𝐶𝐵𝑒 )
𝑘1 𝐶𝐵𝑒
= (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) 𝑤ℎ𝑒𝑟𝑒, =
𝑘−1 𝐶𝐴𝑒

𝑅𝑜𝑏𝑠 𝑅𝑜𝑏𝑠
𝜂= = (𝑘
𝑅𝑏 1 +𝑘−1 )(𝐶𝐴,𝑏 −𝐶𝐴𝑒 )
(𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑏𝑢𝑙𝑘 𝑝ℎ𝑎𝑠𝑒)

93
 𝑑𝐶𝐴
, 𝑎𝑛𝑑 𝑅𝑜𝑏𝑠 = −𝐷𝑒 | 𝑎 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 𝑣𝑜𝑙𝑢𝑚𝑒)
𝑑𝑟 𝑟=𝑟𝑜

𝑚2 ⁄𝑚3

Equation (species balance):

𝐶𝐴,𝑏 𝑟
′Δ𝑟′
𝑟𝑜

𝐷𝑒 ∇2 𝐶𝐴 − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 )

Or,
𝑑 2 𝐶𝐴 2 𝑑𝐶𝐴
𝐷𝑒 ( 2 + ) − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0
𝑑𝑟 𝑟 𝑑𝑟
𝑑𝐶𝐴
BCs. 𝑟 = 0, = 0; 𝑟 = 𝑟𝑜 , 𝐶𝐴 = 𝐶𝐴,𝑏
𝑑𝑟

Modify the species balance equation:

𝑑 2 (𝐶𝐴 − 𝐶𝐴𝑒 ) 2 𝑑(𝐶𝐴 − 𝐶𝐴𝑒 )

𝐷𝑒 ( + ) − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0
𝑑𝑟 2 𝑟 𝑑𝑟
𝑑(𝐶𝐴 −𝐶𝐴𝑒 )
BCs 𝑟=0 =0; 𝑟 = 𝑟𝑜 (𝐶𝐴 − 𝐶𝐴𝑒 ) = (𝐶𝐴𝑏 − 𝐶𝐴𝑒 )
𝑑𝑟

Compare the equation and BCs to the previous situation (irreversible reaction):
𝑘1
𝐴 →𝐵

1 1 1 𝑘1 +𝑘−1
Solution is: 𝜂 = [ − ], 𝜙 = 𝐿√
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙 𝐷𝑒

(No need to solve again)

Ex 3. Determine the effect of temperature on 𝜙, 𝜂 𝑎𝑛𝑑 𝑅𝑜𝑏𝑠 for the catalytic

reaction 𝐴 → 𝐵, 𝑟 = 𝑘𝐶𝐴 (Intrinsic rate)

Ans:

94
 𝑘 𝑘 ∝ exp (−1/𝑇)
𝜙 = 𝐿√
𝐷𝑒 𝐷𝑒 ∝ √𝑇

Therefore, if 𝑇 ↑ (𝑘 ⁄𝐷𝑒 ) ↑ 𝜙 ↑ 𝜂 ↓ (𝑚𝑜𝑣𝑒𝑠 𝑡𝑜𝑤𝑎𝑟𝑑𝑠 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙)

𝐸 1 𝐸
𝑅𝑜𝑏𝑠 = 𝜂 × 𝑅𝑖𝑛 = 𝜂𝑘𝑜 𝑒𝑥𝑝 (− ) 𝐶𝐴 = 𝜂 (𝜙) 𝑘𝑜 𝑒𝑥𝑝 (− 𝑅𝑇) 𝐶𝐴
𝑅𝑇

1 𝐸
𝑅𝑜𝑏𝑠 = 𝜂 𝑘𝑜 𝑒𝑥𝑝 (− )𝐶
𝑅𝑇 𝐴
√𝑒𝑥𝑝 (− 𝐸 )
( 𝑅𝑇 )

4. Determine observed activation energy

If (a) the reaction is kinetic controlled (b) diffusion controlled

(a) 𝜂 = 1, 𝐸𝑜𝑏𝑠 = 𝐸𝑖𝑛𝑡 (𝐸𝑎𝑝𝑝 = 𝐸𝑖𝑛𝑡 )

𝐸 𝐸
[𝑅𝑜𝑏𝑠 = 𝜂𝑘𝑜 𝑒𝑥𝑝 (− ) 𝐶𝐴 = 𝑘𝑜 𝑒𝑥𝑝 (− 𝑅𝑇) 𝐶𝐴 = 𝑘𝑎𝑝𝑝 𝐶𝐴 ]
𝑅𝑇
𝑘𝑎𝑝𝑝 = 𝑘𝑖𝑛𝑡
1
(b) 𝜂 =
𝜙
1 1 𝐷𝑒 1
𝑅𝑜𝑏𝑠 = (𝑘𝐶𝐴 ) = √ (𝑘𝐶𝐴 ) = √𝐷𝑒 𝑘 𝐶𝐴 = 𝑘𝑜𝑏𝑠 𝐶𝐴
𝜙 𝐿 𝑘 𝐿
1 𝐸
𝑘𝑜𝑏𝑠 𝑜𝑟⁄𝑘𝑎𝑝𝑝 = √𝐷𝑒 𝑘 ⇒ 𝐸𝑎𝑝𝑝 = 𝑖𝑛𝑡
𝐿 2
𝐸𝑜𝑏𝑠 = {𝐸⁄2 , 𝐸} compare this to 𝐸𝑜𝑏𝑠 = {0, 𝐸} 𝑓𝑜𝑟 𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒 + 𝑘𝑖𝑛𝑒𝑡𝑖𝑐𝑠

(𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
(𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)

−𝐸⁄
2𝑅
(ℎ𝑖𝑔ℎ 𝑇 ∶ 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
𝑊𝑒 𝑠ℎ𝑜𝑢𝑙𝑑 𝑝𝑒𝑟𝑓𝑜𝑟𝑚
ln(𝑘𝑎𝑝𝑝 ) 𝑒𝑥𝑝𝑡 𝑎𝑡 𝑙𝑜𝑤 𝑇
−𝐸⁄ :
𝑡𝑜 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 𝑡𝑟𝑢𝑒
𝑅
𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦
(𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
𝐿𝑜𝑤 𝑇

1⁄
𝑇

95
⇒ How about endothermic reaction? 𝛽 < 0 (Δ𝐻; +𝑣𝑒) : Not interesting (results
are straightforward);

Some more analysis:

𝜂 = 𝜂(D𝑎 , 𝑛, 𝛽, 𝜖)
Δ𝑇 Δ𝐶
− = 𝛽 (𝑅𝑒𝑐𝑎𝑙𝑙)
𝑇𝑏 𝐶𝑏

How to find 𝛽 ?

𝑘𝑚 (−Δ𝐻)𝐶𝑏
𝛽= (𝐶𝑏 , 𝑇𝑏 𝑎𝑟𝑒 𝑘𝑛𝑜𝑤𝑛, Δ𝐻 is a thermodynamic property ,
ℎ𝑇𝑏
𝑘𝑚 , ℎ ⇒ 𝑘𝑚 (𝑆ℎ), ℎ(𝑁𝑢) : easy to calculate
−Δ𝐻 𝐶𝑏 𝑘𝑚
= ℎ
𝜌𝐶𝑝 𝑇𝑏 ⁄𝜌𝐶
𝑝

2⁄ 2⁄ 𝑅𝑒𝑓𝑒𝑟 𝐵𝑆𝐿 𝑓𝑜𝑟

𝑘𝑚 𝜌 𝜈 3 ℎ 𝜈 3
𝐽𝐷 =
𝐺
(𝐷) ; 𝐽𝐻 =
𝐺𝐶𝑝 𝛼
( ) ( 𝑎𝑛𝑎𝑙𝑜𝑔𝑦 𝑏𝑒𝑤𝑒𝑒𝑛 )
𝑆𝑐 #
ℎ𝑒𝑎𝑡 & 𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡
𝑃𝑟 #
2⁄
𝑘𝑚 𝐽𝐷 𝐷 3 𝛼
Take the ratio: ℎ⁄ = (𝛼 ) (𝐿𝑒 = 𝐷 = 𝐿𝑒𝑤𝑖𝑠 #)
𝜌𝐶𝑝 𝐽𝐻

(−Δ𝐻) 𝐶𝑏 𝐽𝐷 −2⁄
Therefore, 𝛽 = (𝐿𝑒 ) 3 ∶ 𝐷𝑎𝑚𝑘𝑜ℎ𝑙𝑒𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
𝜌𝐶𝑝 𝑇𝑏 𝐽𝐻

( 𝐽𝐷 , 𝐽𝐻 ≡ 𝑡ℎ𝑒𝑟𝑒 𝑎𝑟𝑒 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛𝑠 𝑟𝑒𝑝𝑜𝑟𝑡𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑙𝑖𝑡𝑒𝑟𝑎𝑡𝑢𝑟𝑒)

𝑓
For many practical cases (ℎ𝑖𝑔ℎ 𝑅𝑒 ): 𝐽𝐷 = 𝐽𝐻 = ⁄2

(−Δ𝐻) 𝐶𝑏 𝑃𝑟 ~ 1 − 1000
Therefore, 𝛽 = ( )
𝜌𝐶𝑝 𝑇𝑏 𝑆𝑐 ~ 1 − 1000
𝑇𝑠 −𝑇𝑏 Δ𝐻
and ( )=− (thus simplified)
𝐶𝑏 −𝐶𝑠 𝜌𝐶𝑝

HW (SMITH, 3rd ed): 10.3, 10.5

81
 Lecture 17
Non-porous catalyst (interphase + reaction)

-----continued

Different cases (non-isothermal):

𝑘𝑠 𝐶𝑠 𝑛
𝜂 = ( )( )
𝑘𝑏 𝐶𝑏

(1) 𝛽 > 0 and 𝑛 < 0 ⇒ 𝜂 > 1 ∶ 𝑅 ℎ𝑒𝑎𝑡 ↑ 𝑅 𝑚𝑎𝑠𝑠 ↑

𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡

(2) 𝛽 > 0 and 𝑛 > 0 ⇒ 𝜂 > 1 𝑜𝑟 𝜂 < 1 depending upon 𝐷𝑎 ∶ 𝑅𝐻 ↑ 𝑅𝑚 ↓

(3) β < 0 and n > 0 ⇒ η < 1 ∶ RH ↓ Rm ↓
(4) 𝛽 < 0 and 𝑛 < 0 ⇒ 𝜂 > 1 𝑜𝑟 𝜂 < 1 ∶ 𝑅𝐻 ↓ 𝑅𝑚 ↑

Special Cases: Non - isothermal (1st order)

Mass Flux: (moles⁄ r = 𝑘𝑚 (Cb − Cs ) = 𝑘𝑠 Cs

s − m2 )

(moles⁄ ) 𝑅 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) = 𝑎𝑘𝑠 𝐶𝑠 = 𝑘𝑠 ′𝐶𝑠

s − m3

𝑚2⁄
𝑚3
Therefore,
𝑘𝑚
𝐶𝑠 = ( ) 𝐶𝑏
𝑘𝑚 +𝑘𝑠

𝑎𝑘𝑠 𝑘𝑚 (𝑠𝑎𝑚𝑒 𝑎𝑠 𝑏𝑒𝑓𝑜𝑟𝑒,

𝑅=( ) 𝐶𝑏
𝑘𝑚 +𝑘𝑠 𝑖. 𝑒. 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛)

82
Heat Transport:

𝑄
𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛/𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 = 𝑅 (−𝛥𝐻) (𝑐𝑎𝑙⁄ )
𝑠−𝑚3 𝑜𝑓 𝑐𝑎𝑡 𝑣𝑜𝑙
𝑚𝑜𝑙𝑒⁄ 𝑐𝑎𝑙⁄
𝑠 − 𝑚3 𝑠 − 𝑚𝑜𝑙𝑒
𝑄
{ 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟𝑟𝑒𝑑/𝑟𝑒𝑚𝑜𝑣𝑒𝑑 = 𝑎ℎ(𝑇𝑠 −𝑇𝑏 ) (𝑐𝑎𝑙⁄ )
𝑠−𝑚3 𝑜𝑓 𝑐𝑎𝑡 𝑣𝑜𝑙
𝑐𝑎𝑙⁄
𝑠 − 𝑘 − 𝑚2

−Δ𝐻 𝑎 𝑘𝑚 𝑘𝑠 (−Δ𝐻)𝑎 𝐶𝑏
𝑜𝑟, 𝑄𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 = ( ) 𝐶𝑏 ≡ [ ] (𝑐𝑎𝑙⁄ )
𝑘𝑚 + 𝑘𝑠 1
𝑘𝑚
+
1
𝐸
𝑠 − 𝑚3
𝐴 𝑒𝑥𝑝(−𝑅𝑇 )
𝑠

𝑆𝑆 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛

𝐻𝑜𝑡 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
Investigate under SS:

 Multiplicity (3 solutions) 𝑄𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟𝑟𝑒𝑑 (𝑙𝑖𝑛𝑒𝑎𝑟)

cold/hot/unstable solutions 𝑄
𝑢𝑛𝑠𝑡𝑎𝑏𝑙𝑒( 𝑖𝑔𝑛𝑖𝑡𝑖𝑜𝑛 𝑝𝑡)
 Mass transport: 𝑐𝑜𝑙𝑑
weak/linear dependence on T 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛
𝑄𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 (𝑠𝑖𝑔𝑚𝑜𝑖𝑑𝑎𝑙 𝑠ℎ𝑎𝑝𝑒)
 Heat transport (reaction):
Exponential dependence
𝑇𝑠 − 𝑇𝑏
𝑇𝑏

Recall: Similar situation in the exothermic CSTR operation: instability is an issue

83
Example 1: A well-mixed basket type reactor is used for carrying out the reaction
𝐴 → 𝑃 over a non-porous catalyst under constant flow-condition. The following
data are available:

𝑣𝑜 = 𝑐𝑜𝑛𝑠𝑡
𝑇 = 𝑐𝑜𝑛𝑠𝑡 𝑏𝑢𝑙𝑘 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠
𝑑𝑝 W 𝑋𝐴 𝐶𝐴𝑜 = 𝑐𝑜𝑛𝑠𝑡
1 𝑚𝑜𝑙𝑒/𝑙𝑡
2 2 0.6
6 1 0.2
𝐶𝐴 ,
𝑋𝐴

What controls the reaction? Interphase transport or kinetic rate?

Ans: Write down the species balance equation across the reactor.
𝑚𝑜𝑙𝑒𝑠
: (−𝑟)𝑜𝑏𝑠 𝑊 = 𝑣𝑜 𝐶𝐴𝑜 𝑋𝐴 𝐶𝐴 = 𝐶𝐴𝑜 (1 − 𝑋𝐴 )
𝑠

𝑚𝑜𝑙𝑒𝑠⁄ (𝑣𝑜 𝐶𝐴𝑜 − 𝑣𝑜 𝐶𝐴 )

𝑠 − 𝑔𝑚 𝑔𝑚
𝑜𝑓 𝑐𝑎𝑡 𝐼𝑛 𝑂𝑢𝑡

𝑣𝑜 𝐶𝐴𝑜 𝑋𝐴
(−𝑟)𝑜𝑏𝑠 =
𝑊

(−𝑟)𝑜𝑏𝑠,1 𝑋 𝑊 0.6 1
(−𝑟)𝑜𝑏𝑠,2
= ( 𝐴) × ( ) = × = 3⁄2 (from the measurements)
𝑊 1 𝑋𝐴 2 2 0.2

Recall: If external mass transfer/interphase rate controls or kinetics is fast

𝑟𝑜𝑏𝑠 = 𝑘𝑚 𝑎𝐶𝑏 (𝐶𝑠 → 0)

(𝑟𝑜𝑏𝑠 )1 𝑎1 (1−𝑋𝐴 )1 𝑑𝑝2 (1−𝑋𝐴 )1 6 0.4 𝜎

(𝑟𝑜𝑏𝑠 )2
= = = × = 3⁄2 (𝑎 = )
𝑎2 (1−𝑋𝐴 )2 𝑑𝑝1 (1−𝑋𝐴 )2 2 0.8 𝑑𝑝

(𝐴𝑠𝑠𝑢𝑚𝑒 𝑘𝑚1 = 𝑘𝑚2 𝑢𝑛𝑑𝑒𝑟 𝑡𝑤𝑜 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠) 𝑚2⁄

𝑚3

𝑨𝒏𝒔: Interphase mass transfer controls

84
 𝐾𝑝1 (𝑘 𝑎)1 𝑎1 𝑑𝑝2 6
Alternatively, = (𝑘𝑚 = = = =3
𝐾𝑝2 𝑚 𝑎)2 𝑎2 𝑑𝑝1 2
recall def n of } (𝑟𝑜𝑏𝑠 )1 𝐾𝑝 𝐶𝑏 1 𝐾𝑝 𝐶𝐴𝑜 (1−𝑋1 ) 0.4
overall mass transfer (𝑟𝑜𝑏𝑠 )2
= 1
= 1
=3× = 3⁄2
𝐾𝑝 𝐶𝑏 2 𝐾𝑝 𝐶𝐴𝑜 (1−𝑋2 ) 0.8
2 2
coefficient

Ans: (Interphase mass transfer controls)

𝐄𝐱𝐚𝐦𝐩𝐥𝐞 𝟐: 1 𝑐𝑎𝑡
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛: 𝑆𝑂2 + 𝑂2 → 𝑆𝑂3 (𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙)
(𝑆𝑚𝑖𝑡ℎ 10.1) 2

𝐶𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠
6.42% 𝑆𝑂2 𝑃 = 790 𝑚𝑚 𝐻𝑔
93.50% 𝐴𝑖𝑟 𝑇 = 480𝑜 𝐶
𝐺 = 0.199 𝑘𝑔/𝑚2 − 𝑠 𝑛𝑜𝑛𝑝𝑜𝑟𝑜𝑢𝑠 𝐴𝑙2 𝑂3 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
′′ ′′
(1⁄8 × 1⁄8 𝑐𝑦𝑙𝑖𝑛𝑑𝑟𝑖𝑐𝑎𝑙
𝑘𝑔
𝑑𝑖𝑓𝑓𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑓𝑖𝑥𝑒𝑑 𝑏𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝑝𝑒𝑙𝑙𝑒𝑡𝑠, 𝜌 = 950 ⁄𝑚3 )
𝜖𝑏 = 0.43
𝐴𝑠𝑠𝑢𝑚𝑒 𝑆𝑐 = 1.2
𝑘𝑔
𝜌(𝑎𝑖𝑟) = 0.487 ⁄𝑚3
{𝜇 (𝑎𝑖𝑟) = 3.72 × 10−5 𝑃𝑎 − 𝑠

Data:

𝑋𝑆𝑂 𝑟(
𝑔 𝑚𝑜𝑙𝑒𝑠 𝑆𝑂2
)
𝑝𝑏 , 𝑎𝑡𝑚
2 ℎ−𝑔 𝑜𝑓 𝑐𝑎𝑡
𝑺𝑶𝟐 𝑺𝑶𝟑 𝑶𝟐
0.1 0.0956 0.0603 0.0067 0.201

(𝑝𝑏 − 𝑝𝑠 )⁄
Determine (𝐶𝑏 − 𝐶𝑠 ) 𝑜𝑟 (𝑝𝑏 − 𝑝𝑠 ) 𝑜𝑟 𝑝𝑏

Ans: 𝑟𝑜𝑏𝑠 = 𝑘𝑚 𝑎(𝐶𝑏 − 𝐶𝑠 ) 𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚3

(= 𝑘 𝐶𝑠 𝑛 )

85
 2𝜋𝑑 2
Determine 𝑑𝑝 first for the pellet: 𝜋𝑑𝑝2 = 𝜋𝑑𝐿 +
4
(𝑠𝑝ℎ𝑒𝑟𝑒) (𝑐𝑦𝑙𝑖𝑛𝑑𝑒𝑟)

𝑑𝑝 𝐺 6
𝑑𝑝 = 0.0039 𝑚 → 𝑅𝑒 = = 21, 𝑆𝑐 = 1.2, 𝑎 = = 1538 𝑚−1
𝜇 𝑑𝑝

2⁄
2 𝑘𝑚 𝜌 𝜇 3
Recall: Definition for 𝐽𝐷 = 𝑆𝑡 𝑆𝑐 ⁄3 = (𝜌𝐷)
𝐺

0.458
= R e,p −0.407 (from correlation) = 0.31
ϵb

0.199
Calculate 𝑘𝑚 from here: 0.31 × 2 = 0.112 𝑚/𝑠
1.2 ⁄3 ×0.487

Therefore,
𝑟𝑜𝑏𝑠 0.0956/3600 x 950
(𝐶𝑏 − 𝐶𝑠 ) = = = 1.46x10−4 𝑘𝑔 𝑚𝑜𝑙/𝑚3
𝑘𝑚 𝑎 0.112 x 1538

(𝑝𝑏 − 𝑝𝑠 )𝑆𝑂2 = 𝑅𝑇(𝐶𝑏 − 𝐶𝑠 ) = 0.009 𝑎𝑡𝑚

(𝑝𝑏 − 𝑝𝑠 )𝑆𝑂2
⁄𝑝 = 15% (𝑙𝑎𝑟𝑔𝑒 𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑟𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒)
𝑏,𝑆𝑂2

-Mass transfer controls!

Example 3: A liquid phase oxidation reaction is carried out over non-porous spherical catalyst
particles (W = 500 g) in a well-mixed basket type of reactor. Some oxidation also takes place over
the inner walls (A = 0.1 m2) of the reactor. The following kinetic and transport conditions are
known:

Intrinsic rate Da

Catalyst particles: 1.0C moles/min-kg 1

Inner walls of reactor: 1.0C moles/min-m2 2

Determine the exit conversion of the reactant if the volumetric flowrate of the reacting mixture
is 0.5 m3/min.

Try this example yourself. The answer is 36%. First, write down the species balance equation for
the reactor. Then, you can use either of “two methods/approaches” to solve this problem. One
is to use effectiveness factor, and the other is to use Cs. You should get the same result, i.e., 36%.

86
 Lecture 18
Porous Catalyst (intraphase transport + kinetics)

𝐴→𝑅 - Interphase transport is fast

(Assume nth order) 𝑜𝑟, 𝐶𝑠 → 𝐶𝑏 (𝑘𝑚 𝑎 ↑) (the surface sees the bulk phase concentration)

- Reaction occurs within the material as the reacting species ‘A’

diffuses inside the pores.
𝐶𝑏
𝑟 𝑟 = 𝑘𝐶𝐴 𝑛
moles⁄s − m3
(𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑟𝑎𝑡𝑒) of catalyst volume
(If mole⁄s − m2 of
Δ𝑟 BET area, or
𝑟0 moles⁄s − g of catalyst,
the value & unit
of 𝑘 will change
accordingly) }

Isothermal case

Take a CV (4πr 2 Δr) between r and r + Δr ∶

∂CA
𝑆𝑝𝑒𝑐𝑖𝑒𝑠 𝑏𝑎𝑙𝑎𝑛𝑐𝑒: + Vr . ∇CA ≡ Deff ∇2 CA − kCA n
∂t

𝑚𝑜𝑙𝑒𝑠
⇒ 𝐷𝑒𝑓𝑓 ∇2 𝐶𝐴 = 𝑘𝐶𝐴 𝑛 ( 𝑠−𝑚3 ) (0 ≤ 𝑟 ≤ 𝑟𝑜 )

𝐷𝑒𝑓𝑓 𝑑 2 𝑑𝐶𝐴
2
(𝑟 ) = 𝑘𝐶𝐴 𝑛
𝑟 𝑑𝑟 𝑑𝑟
𝑑𝐶𝐴
𝐵𝑐 1 𝑟 = 0, =0 (symmetric BC)
𝑑𝑟
2 𝑟 = 𝑟𝑜 , 𝐶𝐴 = 𝐶𝐴,𝑏 (𝐶𝐴 |𝑟=𝑟𝑜 = 𝐶𝑏 )

87
Non-dimensionalize:
𝒓 𝐶𝐴
𝒛= & 𝒇=
𝐿 𝐶𝐴,𝑏

𝐿 = 𝑐ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝑙𝑒𝑛𝑔𝑡ℎ
𝑟𝑜
= 1⁄𝑎 = (𝑠𝑝ℎ𝑒𝑟𝑒)
3

(𝑚2 ⁄𝑚3 )
Substitute,
𝐷𝑒𝑓𝑓 𝐶𝐴,𝑏
∇2 𝑓 = 𝑘𝑓 𝑛 𝐶𝐴,𝑏 𝑛
𝐿2

𝐿2 𝑘
2
∇ 𝑓= 𝐶𝐴,𝑏 𝑛−1 𝑓 𝑛
𝐷𝑒

2 2 𝑛 𝑘𝐶𝐴,𝑏 𝑛−1
∇ 𝑓= ϕ 𝑓 ; 𝜙 ≡ 𝑇ℎ𝑖𝑒𝑙𝑒 𝑚𝑜𝑑𝑢𝑙𝑢𝑠 = 𝐿√
𝐷𝑒

𝐵𝐶𝑠. 𝑧 = 0 ∇𝑓 = 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑦)
𝑧=3 𝑓=1 (𝑠𝑢𝑟𝑓𝑎𝑐𝑒)

𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑟𝑎𝑡𝑒 𝑘𝐶𝐴,𝑏 𝑛−1

1. 𝜙 ≡ = 𝐿√
𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑟𝑎𝑡𝑒 𝐷𝑒
𝑖𝑛𝑡𝑟𝑎𝑝ℎ𝑎𝑠𝑒
(𝑝𝑜𝑟𝑒 − 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛)
𝑅𝑜𝑏𝑠/𝑔𝑙𝑜𝑏𝑎𝑙/ 𝑎𝑝𝑝𝑎𝑟𝑒𝑛𝑡
Define, 𝜂 = effectiveness factor = ≤ 1 for + ve order
𝑅𝑏
2.

where, R b = rate w⁄o diffusion (max at the surface)

𝑅𝑜𝑏𝑠 = 𝜂(𝑘𝐶𝐴,𝑏 𝑛 ) ⇒ 𝜂𝑖𝑛𝑡𝑟𝑎 = 𝜂(𝜙, 𝑛) compare 𝑡𝑜 𝜂𝑖𝑛𝑡𝑒𝑟 = 𝜂(𝐷𝑎 , 𝑛)

- First solve the equation (numerically) ⇒ 𝑓(𝑧) 𝑜𝑟 𝐶𝐴 (𝑟)

- Calculate ‘𝑅𝑜𝑏𝑠 ’
Alternatively, calculate ϕ , ηintra and then R obs
88
Special Case: 1st order reaction (analytical solution is available)

𝐷𝑒 ∇2 𝑓 = 𝑘𝑓
𝑑2𝑓 2 𝑑𝑓
𝐷𝑒 (
𝑑𝑧 2
+
𝑧 𝑑𝑧
) = 𝑘𝐿2 𝑓

𝑑2𝑓 2 𝑑𝑓 𝑘 𝑟𝑜
+ = 𝜙 2 𝑓; 𝜙 = 𝐿√ (𝐿 = 𝑓𝑜𝑟 𝑠𝑝ℎ𝑒𝑟𝑒)
𝑑𝑧 2 𝑧 𝑑𝑧 𝐷𝑒 3

𝐵𝐶𝑠 𝑧 = 0, ∇𝑓 = 0 𝑟
(𝑧 = 𝑟⁄𝐿 = 𝑜⁄𝑟𝑜 = 3)
𝑧=3 𝑓=1 ⁄3
𝑔
Let 𝑔 = 𝑓𝑧 ; 𝑓 = ⁄𝑧

Differentiate f twice wrt z, simplify and substitute in the equation above to

show the following transformation:

𝐵𝐶𝑠: 𝑧=0 , 𝑔=0

𝑑2𝑔 2 }
−𝜙 𝑔 =0 𝑧=3 , 𝑔=3
𝑑𝑧 2

𝑔 = 𝐶1 𝑠𝑖𝑛ℎ𝜙𝑧 + 𝐶2 𝑐𝑜𝑠ℎ𝜙𝑧

𝐶2 = 0, 𝐶1 = 3⁄𝑠𝑖𝑛ℎ3𝜙 (apply BCs)

3𝑠𝑖𝑛ℎ(𝜙𝑧)
𝑔=
𝑠𝑖𝑛ℎ(3𝜙)

3𝑠𝑖𝑛ℎ(𝜙𝑧)
𝑓= ∶ (dimensionless solution)
𝑧 𝑠𝑖𝑛ℎ(3𝜙)

𝐶𝐴 3 𝑠𝑖𝑛ℎ(𝜙𝑟⁄𝐿) 𝑟𝑜 𝑠𝑖𝑛ℎ(𝜙3𝑟⁄𝑟𝑜 )
= 𝑟⁄ 𝑠𝑖𝑛ℎ(3𝜙) = ⇒ Thus, 𝐶𝐴 (𝑟) is determined.
𝐶𝐴,𝑏 𝐿 𝑟 𝑠𝑖𝑛ℎ(3𝜙)

Calculate the reaction rate by equating the rate of species arriving at the surface
of the catalyst to that of species consumed/reacted in the catalyst volume under
SS conditions.

89
 𝑚𝑜𝑙𝑒𝑠 𝑟
𝑅( ) = ∫0 𝑜 𝑘𝐶𝐴 (𝑟) 4𝜋 𝑟 2 𝑑𝑟 ("A" is 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 𝑝𝑜𝑟𝑒𝑠)
𝑠

𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 𝑚3
𝐶𝑏
𝑟
𝑑𝐶𝐴 2
= −𝐷𝑒 | 4𝜋𝑟𝑜 ;
𝑑𝑟 𝑟=𝑟𝑜
(𝑓𝑙𝑢𝑥 𝑜𝑓 A 𝑎𝑟𝑟𝑖𝑣𝑖𝑛𝑔 𝑎𝑡 𝑡ℎ𝑒 𝑠𝑢𝑟𝑓𝑎𝑐𝑒) 𝑟𝑜
Δ𝑟

(Note that one can also write:

R global = kCA (r), based on the volume average concentration inside the catalyst)

It is easier to differentiate than to integrate (use the second expression)

𝑑𝐶
𝑅𝑔𝑙𝑜𝑏𝑎𝑙 −𝐷𝑒 𝑑𝑟𝐴 | 4𝜋𝑟𝑜 2
𝑟=𝑟𝑜
𝜂𝑖𝑛𝑡𝑟𝑎 = = 4
𝑅𝑏 (𝑘𝐶𝐴,𝑏 )3𝜋𝑟𝑜 3

= 𝐷𝑒
𝑑(𝐶𝐴 ⁄𝐶𝐴,𝑏 )
| x
1 𝑟
(𝐿 = 𝑜⁄ )
𝑑(𝑟⁄𝐿) 𝑟=𝑟
𝑜
2
𝐿 𝑘 3
1 𝑑𝑓
= |
𝜙2 𝑑𝑧 𝑧=3

Already we have the solution 𝑓(𝑧) from the previous exercise; substitute to obtain:

1 1 1 𝑘
𝜂𝑖𝑛𝑡𝑟𝑎 = [ − ] , 𝜙 = 𝐿√
1⁄ (1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟)
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙 𝐷𝑒
𝑠

𝑚 𝑚2⁄
𝑆

Once we know 𝜼𝒊𝒏𝒕𝒓𝒂 ⇒ 𝒄𝒂𝒍𝒄𝒖𝒍𝒂𝒕𝒆 𝑹𝒈𝒍𝒐𝒃𝒂𝒍 = 𝜼𝒊𝒏𝒕𝒓𝒂 × (𝒌𝑪𝑨,𝒃 ) 𝒎𝒐𝒍𝒆⁄

𝒔 − 𝒎𝟑

90
Q: 1. What conditions will lead 𝜂 > 1 always

a) 𝑛 < 0 b) exothermic

2. Is it practical to operate catalytic reaction with 𝜂 > 1

-Yes (𝜂 > 1 𝑖𝑠 𝑔𝑒𝑛𝑢𝑖𝑛𝑒)

𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐 (𝑟𝑒𝑎𝑙 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠)

Interphase plus intraphase + reaction (isothermal): 𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑟𝑎𝑡𝑒

𝑟 = 𝑘𝐶 𝑛
𝐶𝑏 𝐶𝑝 (r)
𝐶𝑝 (r)

𝐶𝑠
𝑟𝑝
𝐷𝑒 ∇2 𝐶𝑝 = 𝑘𝐶𝑝 𝑛
𝑚𝑜𝑙𝑒𝑠
𝑅𝑜𝑏𝑠 (
𝑠−𝑚3
) = 𝑎 𝑘𝑚 (𝐶𝑏 − 𝐶𝑠 ) eq. (1)
2
𝑚 ⁄ 𝑚⁄ 𝑚𝑜𝑙𝑒𝑠⁄
𝑚3 𝑠
𝑚3

𝑑𝐶𝑝
= −𝐷𝑒 𝑑𝑟 | 𝑎 (Note that Cp @ r = R is Cs)
𝑟=𝑟𝑝

𝑅𝑜𝑏𝑠
= (𝑘𝐶𝑠 𝑛 )𝜂𝑖𝑛𝑡𝑟𝑎 (𝜂𝑖𝑛𝑡𝑟𝑎 = )
𝑅𝑠

𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛s
1
𝑅𝑜𝑏𝑠 𝑛
𝑹𝒐𝒃𝒔 = 𝑎 𝑘𝑚 [𝐶𝑏 − ( ) ] 𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑒 𝑖𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝐶𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑒𝑞𝑛)
𝑘𝜂 𝑖𝑛𝑡𝑟𝑎

where, 𝜂𝑖𝑛𝑡𝑟𝑎 = 𝜂𝑖𝑛𝑡𝑟𝑎 (𝑛, 𝜙) and

𝑘𝐶𝑠 𝑛−1
𝜙 = 𝐿√ eq. (2)
{ 𝐷𝑒
One can also define,

𝑅𝑜𝑏𝑠 (𝑘𝐶𝑠 𝑛 )𝜂𝑖𝑛𝑡𝑟𝑎 𝐶 𝑛

𝜂𝑡𝑜𝑡𝑎𝑙 = = (𝑘𝐶𝑏 𝑛) = 𝜂𝑖𝑛𝑡𝑟𝑎 ( 𝑠 ) eq. (3)
𝑅𝑚𝑎𝑥 𝐶 𝑏

(𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑏𝑢𝑙𝑘 𝑝ℎ𝑎𝑠𝑒 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛)

106
 Plot:
𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑

1.0

𝜂
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑

𝜂 ~ 1⁄𝜙 (𝜙 𝑖𝑠 ℎ𝑖𝑔ℎ)

1 𝜙 10

(The curve is hand-drawn. See the textbook for an accurate representation of the data)

𝐶𝑏

𝜙 ↓ 𝑜𝑟 𝐷𝑒 ↑↑ 𝑜𝑟 𝐿 ↓
𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑡𝑜 ′𝑘′ 𝐶𝑏
(𝑙𝑎𝑟𝑔𝑒 𝑝𝑜𝑟𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛) 𝑠𝑚𝑎𝑙𝑙 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑠𝑖𝑧𝑒
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑖𝑠 𝑓𝑎𝑠𝑡 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔
𝜂→1 ′𝜙′
𝑅𝑜𝑏𝑠 → 𝑅𝑏 (𝑘𝐶𝐴,𝑏 ) 𝐶𝑏
{ − 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 }

∇𝐶𝐴 (𝑟) → 0 }

Alternatively, 𝐶𝑏

𝜙 ↑ 𝑜𝑟 𝐷𝑒 ↓↓ 𝑜𝑟 𝐿 ↑ 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔
(𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑡𝑜′𝑘 ′ ) ′𝜙′
𝑜𝑟 𝑘 𝑖𝑠 𝑙𝑎𝑟𝑔𝑒 𝑙𝑎𝑟𝑔𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝐶𝑏
I
𝑜𝑟 𝑝𝑜𝑟𝑒 𝑖𝑠 𝑛𝑎𝑟𝑟𝑜𝑤
𝜂 → 𝑠𝑚𝑎𝑙𝑙 𝑜𝑟 𝜂 ≪ 1
𝑅𝑜𝑏𝑠 ≪ 𝑅𝑏
{ − 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑 } 𝐶𝑏
∇𝐶𝐴 (𝑟) 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑒𝑠
𝑤𝑖𝑡ℎ 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 ′𝜙′

I
91
 Note:

Graphical solutions are available for different shapes of the materials and different orders
of reaction.

𝑛 = 0 (two solutions?)
𝑆𝑙𝑎𝑏
1.0 1.0
1 1
1⁄
𝐶𝑦𝑙𝑖𝑛𝑑𝑒𝑟 2
𝑎𝑛𝑑 𝜂
2
𝑆𝑝ℎ𝑒𝑟𝑒

1.0

𝜙 10 𝑛+1 𝑘 10
𝜙 = 𝐿√ 𝐶𝐴 𝑛−1
2 𝐷𝑒

𝑛
Again, calculate 𝜙 first and then 𝜂𝑖𝑛𝑡𝑟𝑎 ⇒ 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒 𝑅𝑔𝑙𝑜𝑏𝑎𝑙 = 𝜂𝑖𝑛𝑡𝑟𝑎 × (𝐶𝐴,𝑏 ) 𝑚𝑜𝑙𝑒⁄𝑠 − 𝑚3

We will close the discussion on intraphase transport + kinetics in this lecture before taking up the
examples in the next few lectures.

Smith has devoted two-three pages (507-510) on the effect of internal transport on selectivity in
chapter 11. It is not difficult for you to go through this section yourself. Briefly, Smith categorizes
the effects through Type 1, 2 and 3 series or parallel reactions.

Type 1: A ⤏ B (desired) and R ⤏ S (undesired)

Type 2: A ⤏ B (desired) and A ⤏ C (undesired)

Type 3: A ⤏ B ⤏ D, where B is the desired product.

Examples are from hydrogenation, dehydrogenation, dehydration, and oxidation of some

hydrocarbons.

Type 1 and 2 are easy to handle and quantify. Type 3 requires some thinking and calculations.
For Type 1 reactions, internal transport or diffusion decreases the selectivity. For Type 2 (both
first order irreversible reactions), selectivity is unaffected by diffusion. Type 3 is slightly trickier
to understand because the product B is the reactant for the second consecutive/series reaction.
Therefore, pore-diffusion coefficient for B becomes as important as the pore-diffusion coefficient
of the reactant A! May be there are two Thiele modulus? And, there may be so many different
situations which you should be able to recognize……

92
 Lecture 19-20
Example 1: Explain 𝑛 = 0, mathematically
𝑇𝑤𝑜 𝑠𝑖𝑡𝑢𝑎𝑡𝑖𝑜𝑛𝑠:

( 𝜙 ≤ 1, 𝜙 > 1)
1.0

𝜂
𝜂=1 𝜂<1

1.0 𝜙

Hint: For a zero-order reaction, analytically integrate the non-dimensional species conservation
differential equation to derive a general expression for concentration in the pellet. Note that
concentration is minimum at the center; however it cannot be zero. With this constraint you
should be able to derive an expression for concentration profiles inside the pellet with a limit for
𝜙 < Ø*. In this case, the process is kinetic controlled and 𝜂 = 1 (observed rate = intrinsic rate for
zero-order reaction). For sphere Ø* = 1 and for slab Ø* = √2. For 𝜙  Ø*, concentration, however,
decreases to zero at some distance (r*) away from the center inside the pellet, beyond which
region 𝜂 = 1/𝜙 and the process is diffusion controlled. Below that distance, there is no gas
(reactant) and no reaction. See Rodrigues and Orfao (1984) in Chem. Eng. Commun. 27, pp. 327-
33.

𝑘1
Ex 2. For reversible, 𝐴 ⇌ 𝐵 ∶ calculate 𝜂(𝜙).
𝑘−1
𝑑𝐶𝐴
Ans: 𝑟=− = 𝑘1 𝐶𝐴 − 𝑘−1 𝐶𝐵
𝑑𝑡

(For a reversible reaction, recall your previous calculations: 𝐶𝐴 + 𝐶𝐵 = 𝐶 = 𝐶𝐴𝑜 + 𝐶𝐵𝑜

= 𝐶𝐴𝑒 + 𝐶𝐵𝑒 )
𝑘1 𝐶𝐵𝑒
= (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) 𝑤ℎ𝑒𝑟𝑒, =
𝑘−1 𝐶𝐴𝑒

𝑅𝑜𝑏𝑠 𝑅𝑜𝑏𝑠
𝜂= = (𝑘
𝑅𝑏 1 +𝑘−1 )(𝐶𝐴,𝑏 −𝐶𝐴𝑒 )
(𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑏𝑢𝑙𝑘 𝑝ℎ𝑎𝑠𝑒)

93
 𝑑𝐶𝐴
, 𝑎𝑛𝑑 𝑅𝑜𝑏𝑠 = −𝐷𝑒 | 𝑎 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡 𝑣𝑜𝑙𝑢𝑚𝑒)
𝑑𝑟 𝑟=𝑟𝑜

𝑚2 ⁄𝑚3

Equation (species balance):

𝐶𝐴,𝑏 𝑟
′Δ𝑟′
𝑟𝑜

𝐷𝑒 ∇2 𝐶𝐴 − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0 (𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑚3 )

Or,
𝑑 2 𝐶𝐴 2 𝑑𝐶𝐴
𝐷𝑒 ( 2 + ) − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0
𝑑𝑟 𝑟 𝑑𝑟
𝑑𝐶𝐴
BCs. 𝑟 = 0, = 0; 𝑟 = 𝑟𝑜 , 𝐶𝐴 = 𝐶𝐴,𝑏
𝑑𝑟

Modify the species balance equation:

𝑑 2 (𝐶𝐴 − 𝐶𝐴𝑒 ) 2 𝑑(𝐶𝐴 − 𝐶𝐴𝑒 )

𝐷𝑒 ( + ) − (𝑘1 + 𝑘−1 )(𝐶𝐴 − 𝐶𝐴𝑒 ) = 0
𝑑𝑟 2 𝑟 𝑑𝑟
𝑑(𝐶𝐴 −𝐶𝐴𝑒 )
BCs 𝑟=0 =0; 𝑟 = 𝑟𝑜 (𝐶𝐴 − 𝐶𝐴𝑒 ) = (𝐶𝐴𝑏 − 𝐶𝐴𝑒 )
𝑑𝑟

Compare the equation and BCs to the previous situation (irreversible reaction):
𝑘1
𝐴 →𝐵

1 1 1 𝑘1 +𝑘−1
Solution is: 𝜂 = [ − ], 𝜙 = 𝐿√
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙 𝐷𝑒

(No need to solve again)

Ex 3. Determine the effect of temperature on 𝜙, 𝜂 𝑎𝑛𝑑 𝑅𝑜𝑏𝑠 for the catalytic

reaction 𝐴 → 𝐵, 𝑟 = 𝑘𝐶𝐴 (Intrinsic rate)

Ans:

94
 𝑘 𝑘 ∝ exp (−1/𝑇)
𝜙 = 𝐿√
𝐷𝑒 𝐷𝑒 ∝ √𝑇

Therefore, if 𝑇 ↑ (𝑘 ⁄𝐷𝑒 ) ↑ 𝜙 ↑ 𝜂 ↓ (𝑚𝑜𝑣𝑒𝑠 𝑡𝑜𝑤𝑎𝑟𝑑𝑠 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙)

𝐸 1 𝐸
𝑅𝑜𝑏𝑠 = 𝜂 × 𝑅𝑖𝑛 = 𝜂𝑘𝑜 𝑒𝑥𝑝 (− ) 𝐶𝐴 = 𝜂 (𝜙) 𝑘𝑜 𝑒𝑥𝑝 (− 𝑅𝑇) 𝐶𝐴
𝑅𝑇

1 𝐸
𝑅𝑜𝑏𝑠 = 𝜂 𝑘𝑜 𝑒𝑥𝑝 (− )𝐶
𝑅𝑇 𝐴
√𝑒𝑥𝑝 (− 𝐸 )
( 𝑅𝑇 )

4. Determine observed activation energy

If (a) the reaction is kinetic controlled (b) diffusion controlled

(a) 𝜂 = 1, 𝐸𝑜𝑏𝑠 = 𝐸𝑖𝑛𝑡 (𝐸𝑎𝑝𝑝 = 𝐸𝑖𝑛𝑡 )

𝐸 𝐸
[𝑅𝑜𝑏𝑠 = 𝜂𝑘𝑜 𝑒𝑥𝑝 (− ) 𝐶𝐴 = 𝑘𝑜 𝑒𝑥𝑝 (− 𝑅𝑇) 𝐶𝐴 = 𝑘𝑎𝑝𝑝 𝐶𝐴 ]
𝑅𝑇
𝑘𝑎𝑝𝑝 = 𝑘𝑖𝑛𝑡
1
(b) 𝜂 =
𝜙
1 1 𝐷𝑒 1
𝑅𝑜𝑏𝑠 = (𝑘𝐶𝐴 ) = √ (𝑘𝐶𝐴 ) = √𝐷𝑒 𝑘 𝐶𝐴 = 𝑘𝑜𝑏𝑠 𝐶𝐴
𝜙 𝐿 𝑘 𝐿
1 𝐸
𝑘𝑜𝑏𝑠 𝑜𝑟⁄𝑘𝑎𝑝𝑝 = √𝐷𝑒 𝑘 ⇒ 𝐸𝑎𝑝𝑝 = 𝑖𝑛𝑡
𝐿 2
𝐸𝑜𝑏𝑠 = {𝐸⁄2 , 𝐸} compare this to 𝐸𝑜𝑏𝑠 = {0, 𝐸} 𝑓𝑜𝑟 𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒 + 𝑘𝑖𝑛𝑒𝑡𝑖𝑐𝑠

(𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
(𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)

−𝐸⁄
2𝑅
(ℎ𝑖𝑔ℎ 𝑇 ∶ 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
𝑊𝑒 𝑠ℎ𝑜𝑢𝑙𝑑 𝑝𝑒𝑟𝑓𝑜𝑟𝑚
ln(𝑘𝑎𝑝𝑝 ) 𝑒𝑥𝑝𝑡 𝑎𝑡 𝑙𝑜𝑤 𝑇
−𝐸⁄ :
𝑡𝑜 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒 𝑡𝑟𝑢𝑒
𝑅
𝑎𝑐𝑡𝑖𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑛𝑒𝑟𝑔𝑦
(𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
𝐿𝑜𝑤 𝑇

1⁄
𝑇

95
𝐶𝐴 (𝑟):
𝐶𝐴,𝑏 𝐶𝑖

𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 𝑇 (𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)

5. Determine the pellet size effects on 𝜙 𝑎𝑛𝑑 𝜂

𝑘 1
𝜙 = 𝐿√ ; 𝜂 = 𝜂( )
𝐷𝑒 𝜙

If 𝐿 ↓ , 𝜙 ↓ , 𝜂 ↑ 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑
}
𝐴𝑙𝑠𝑜, 𝑙𝑎𝑟𝑔𝑒 𝑝𝑜𝑟𝑒 𝑠𝑖𝑧𝑒, 𝐷𝑒 ↑ , 𝜙 ↓ , 𝜂 ↑ (𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑖𝑠 𝑓𝑎𝑠𝑡)

(𝑅𝑜𝑏𝑠 → 𝑅𝑖𝑛𝑡 )

If a catalyst is given to you to find intrinsic rate (or true ‘E’), make it powder, and
also, conduct reaction at low temperature (see the finding in Q4).

𝑑𝑝 = 0.4 𝑐𝑚
𝐴 → 𝑅 ( 1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟, 𝑖𝑟𝑟𝑒𝑣𝑒𝑟𝑠𝑖𝑏𝑙𝑒) 𝐷𝑒𝑓𝑓 = 0.015 𝑐𝑚2 ⁄𝑠
Ex 4. 𝑜𝑣𝑒𝑟 𝑝𝑜𝑟𝑜𝑢𝑠 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡
𝑘𝑖𝑛𝑡 (100𝑜 𝐶) = 0.93 𝑠 −1
{ 𝐸 = 20,000 𝑐𝑎𝑙/𝑚𝑜𝑙𝑒

𝑖𝑛𝑡𝑒𝑟𝑝ℎ𝑎𝑠𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑟𝑎𝑡𝑒 𝑖𝑠 fast

Q1: Determine 𝑅𝑎 𝑜𝑟 𝑅𝑜𝑏𝑠 𝑎𝑡 100𝑜 𝐶, 𝑖𝑓 𝐶𝐴𝑜 = 3.25 × 10−2 𝑚𝑜𝑙𝑒𝑠⁄𝐿

𝑅𝑜𝑏𝑠 = 𝜂 (𝑅𝑏 ) = 𝜂(𝑘𝐶𝐴𝑜 )

𝑘 𝑑𝑝 𝑘 0.4 0.93
𝜙 = 𝐿√ = √𝐷 = √ = 0.525
𝐷 𝑒 6 𝑒 6 0.015

1 1 1
𝜂= [ − ] ⇒ 𝜂 = 0.866
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙

96
𝑅𝑜𝑏𝑠 = 0.866 × 0.93 × 3.25 × 10−2 = 2.61 × 10−2 𝑚𝑜𝑙𝑒𝑠⁄𝐿 − 𝑠

Q2: For 𝐶𝐴𝑜 = 3.25 × 10−2 𝑚𝑜𝑙𝑒𝑠⁄𝐿 , 𝑤ℎ𝑎𝑡 𝑖𝑠 𝑅𝑜𝑏𝑠 𝑎𝑡 150𝑜 𝐶 ?

𝐸 1 1
𝑘150 = 𝑘100 𝑒𝑥𝑝 (− ((150+273) − (100+273)))
𝑅

20,000 1 1
= 0.93 𝑒𝑥𝑝 (− ( − ))
1.98 423 373

= 22.83 𝑠 −1
1⁄
150+273 2
𝐷𝑒𝑓𝑓 = 0.015 × ( ) = 0.016
100+273

0.4 22.83
Therefore, 𝜙 = √ = 2.51
6 0.016

1 1 1
𝜂= [ − ] = 0.345
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙

𝑅𝑜𝑏𝑠 = 𝜂(𝑘𝐶𝐴 ) = 0.345(22.83 × 3.25 × 10−2 )

= 2.52 × 10−1 𝑚𝑜𝑙𝑒𝑠⁄𝐿 − 𝑠

Q3: Why 𝜂(150𝑜 𝐶) < 𝜂(100𝑜 𝐶)

- Temperature effects (kinetics is faster at 150𝑜 𝐶; pore-diffusion controls)

(catalyst is less effective if there is a pore-diffusion) (𝜂 = 0.45); catalyst will be

more effective at relatively lower temperatures.

To make the catalyst relatively more effective at a fixed high temperature, you can
make smaller pellets; see if the material pore size distribution can be tailored to
increase pore diffusion coefficient (𝐷𝑒𝑓𝑓 ), although the latter is not trivial.

97
Q4: Determine 𝐸𝑜𝑏𝑠 (you have two data points)
𝑘150 𝑅150
( ) =( ) ( 𝐶𝐴,𝑏 = 𝑐𝑜𝑛𝑠𝑡)
𝑘100 𝑎𝑝𝑝 𝑜𝑟 𝑜𝑏𝑠 𝑅100 𝑎𝑝𝑝 𝑜𝑟 𝑜𝑏𝑠
𝑏𝑢𝑙𝑘 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖𝑠 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑖𝑛 𝑡𝑤𝑜 𝑐𝑎𝑠𝑒𝑠

2.52×10−1 𝐸𝑜𝑏𝑠 1 1
= = 𝑒𝑥𝑝 (− ( − ))
2.61×10−2 𝑅 𝑇2 𝑇1

𝐸 1 1 2.267×1.98
𝑜𝑟, 9.655 = 𝑒𝑥𝑝 (− 1.98
𝑜𝑏𝑠
(423 − 373)) - 𝐸𝑜𝑏𝑠 = = 14.16 𝑘𝑐𝑎𝑙⁄𝑚𝑜𝑙𝑒.
3.17×10−4

98
 Lecture 20-21
Examples continued…

Q5: Consider a catalytic reaction over a porous catalyst. The data for the catalyst grain (particle)
are given as follows:

𝑇 = 803 𝐾, 𝑑𝑝 = 0.32 𝑐𝑚, 𝑘𝑜𝑏𝑠 = 0.94 𝑐𝑚3 ⁄𝑠 − 𝑔𝑚 𝑜𝑓 𝑐𝑎𝑡

𝑣𝑝𝑜𝑟𝑒 ≠
𝑉𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑎 = 110𝑜 𝐴, 𝜏 = 3.0, 𝑣𝑝𝑜𝑟𝑒 = 0.35 𝑐𝑚3 ⁄𝑔𝑚 , 𝑀 (𝑟𝑒𝑎𝑐𝑡𝑖𝑛𝑔 𝑠𝑝𝑒𝑐𝑖𝑒𝑠) = 58
{ }
Determine/Calculate 𝜂. Assume 𝐷𝑘 is dominant.

𝐷𝑐⁄
Ans: 𝐷𝑒𝑓𝑓 = 𝜖 𝜏 ; 𝐷𝑐 = 𝐷𝑘

𝑇(𝐾)
𝐷𝑘 = 9700 × 𝑎 (𝑐𝑚)√ 𝑀

803
𝑉𝑝𝑜𝑟𝑒 = 9700 × 110 × 10−8 √ 58 = 3.97 × 10−2 𝑐𝑚2 ⁄𝑠

𝜖 = 0.35 × 𝜌𝑝 (𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑖𝑠 𝑛𝑜𝑡 𝑝𝑟𝑜𝑣𝑖𝑑𝑒𝑑)

𝑐𝑚3
( ) 𝑔𝑚
𝑔𝑚
𝑐𝑚3
(𝑡𝑜𝑡𝑎𝑙)

0.35 × 𝜌𝑝 ×3.97×10−2
𝐷𝑒𝑓𝑓 = = 4.6 × 10−3 𝜌𝑝 𝑐𝑚2 ⁄𝑠
3

1. Check 𝑘𝑜𝑏𝑠 ⇒ 𝑅𝑜𝑏𝑠 = 𝑘𝑜𝑏𝑠 𝐶𝐴,𝑏 = 𝜂(𝑅𝑏 ) = 𝜂(𝑘𝑖𝑛𝑡 𝐶𝐴,𝑏 )

𝑑𝑒𝑓 𝑛 𝑓𝑜𝑟 𝑘𝑜𝑏𝑠 𝑤⁄𝑜 𝑝𝑜𝑟𝑒 − 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛/

𝑎𝑡 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑐𝑜𝑛𝑑𝑖𝑡𝑜𝑛
𝑘𝑜𝑏𝑠⁄
(𝑘𝑜𝑏𝑠 = 𝜂 𝑘𝑖𝑛𝑡 ⇒ 𝑘𝑖𝑛𝑡 = 𝜂)

2. Check the unit of 𝑘𝑜𝑏𝑠 ⇒ 𝑐𝑚3 ⁄𝑠 − 𝑔𝑚 𝑜𝑓 𝑐𝑎𝑡 (1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟)

𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑅 ≡ 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑔𝑚 𝑜𝑓 𝑐𝑎𝑡
} 𝑘 ≡ 𝑐𝑚3 ⁄𝑠 − 𝑔𝑚 𝑜𝑓 𝑐𝑎𝑡
= 𝑘 𝐶𝐴 𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3

99
 𝑘𝑖𝑛𝑡 𝑑𝑝 𝑘𝑖𝑛𝑡
The equation 𝜙 = 𝐿√ = √𝐷 was derived for 𝑅 = 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑐𝑚3
𝐷𝑒𝑓𝑓 6 𝑒𝑓𝑓
𝑜𝑓 𝑐𝑎𝑡 𝑣𝑜𝑙𝑢𝑚𝑒
1⁄ (1st order)
𝑠 =

Therefore,

𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑/
𝑘𝑜𝑏𝑠 to 𝑘𝑜𝑏𝑠 × 𝜌𝑝 → (𝑘𝑜𝑏𝑠 × 𝜌): 1⁄𝑠
𝑐𝑜𝑛𝑣𝑒𝑟𝑡𝑒𝑑
𝑔𝑚
3 ⁄𝑐𝑚3
𝑐𝑚 ⁄
𝑠 − 𝑔𝑚

𝑑𝑝 𝑘𝑜𝑏𝑠 × 𝜌𝑝 1
Therefore, 𝜙= √( )×( )
6 𝜂 𝐷𝑒𝑓𝑓 × 𝜌𝑝

𝑑𝑝 𝑘𝑜𝑏𝑠 0.32 0.94

𝜙= √𝜂 𝐷 = √4.6×10−3 × 𝜂 − (1)
6 𝑒𝑓𝑓 6

1 1 1
𝜂= [ − ] − (2)
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙

Solve to get 𝜂 𝑜𝑟 𝜙 (requires iteration)

In the exam, just do one or two iterations.

Q6: Oxidation of 𝐶𝑂 occurs over the porous 𝑃𝑡⁄𝐴𝑙2 𝑂3 catalyst at 𝑇 = 400 𝑜 𝐶, 𝑃𝐶𝑂 = 0.15 atm

(film transport is negligible or 𝑘𝑚 is very high):

Intrinsic rate: 𝑟𝑖 = 10−6 𝐶𝐴 (𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚2 − 𝑠)

Data for the catalyst: 𝑆 = 20 𝑚2⁄𝑔𝑚 , 𝜖 = 0.5 𝑐𝑚3 ⁄𝑐𝑚3

𝜌𝑝 = 5.3 𝑔⁄𝑐𝑚3 , 𝑑𝑝 = 8 𝑚𝑚
𝑔𝑟𝑎𝑖𝑛 (𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒)𝑑𝑒𝑛𝑠𝑖𝑡𝑦 ⁄

𝐷𝑚 𝑎𝑡 0𝑜 𝐶 = 0.04 𝑐𝑚2 ⁄𝑠

Determine ‘𝑟’, the rate of reaction in one particle, i.e., or Robs in

𝑚𝑜𝑙𝑒𝑠⁄
𝑠−′ 𝑜𝑛𝑒 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 ′

Ans: First modify 𝑟𝑖 = 10−6 𝐶𝐴 × 𝑆 × 𝜌𝑝 ( 𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3 − 𝑠)

100
 = 10−6 × 20 × 10 4 × 5.3 𝐶𝐴 ( 𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3 − 𝑠)

= 1.06 𝐶𝐴 ( 𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3 − 𝑠)

1. 𝑘𝑖𝑛𝑡 = 1.06 1/𝑠

3⁄
𝑇1 2 673 1.5
𝐷400 = 0.04 ( ) = 0.04 ( )
𝑇 2 273

𝐷400 = 0.155 𝑐𝑚2 ⁄𝑠

𝑇 673
2. 𝐷𝑘 = 9700 𝑎 (𝑐𝑚)√ = 9700 𝑎 √
𝑀 28
2 𝑉𝑝 𝜖
𝑎 = (𝑅𝑒𝑐𝑎𝑙𝑙)
𝑆

𝑎𝑠𝑠𝑢𝑚𝑒 𝑐𝑦𝑙𝑖𝑛𝑑𝑟𝑖𝑐𝑎𝑙 𝑝𝑜𝑟𝑒𝑠:

𝑆 = 2𝜋𝑟𝐿
𝑣𝑝 = 𝜋 𝑟 2 𝐿 } 𝜖𝑝 = 𝜖𝑔 (𝜖)
2𝑣𝑝 2𝑉𝑝 𝜖𝑝
𝑟= = 𝑔𝑟𝑎𝑖𝑛 𝑜𝑟 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦
𝑆 𝑆
2𝜖𝑝 𝑉𝑝 = 𝑔𝑟𝑎𝑖𝑛⁄𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑣𝑜𝑙𝑢𝑚𝑒
=
𝑎𝑣𝑔 𝑝𝑜𝑟𝑒 𝑠
𝑟𝑎𝑑𝑖𝑢𝑠
𝑆 = 𝑝𝑜𝑟𝑒 (𝑠𝑢𝑟𝑓𝑎𝑐𝑒) 𝑎𝑟𝑒𝑎
𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑎𝑟𝑒𝑎
𝑣 = 𝑝𝑜𝑟𝑒 − 𝑣𝑜𝑙𝑢𝑚𝑒 = 𝑉𝑝 𝜖𝑝 𝑜𝑟 𝜖𝑔
𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 𝑣𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 ) { 𝑝
𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑜𝑟 𝑔𝑟𝑎𝑖𝑛)

Therefore,

2×0.5 2×0.5
𝑎 = = = 9.4 × 10−7 𝑐𝑚
𝑠×𝜌𝑝 20×104 ×5.3

𝑐𝑚2⁄ 𝑔𝑚⁄
𝑔𝑚 𝑐𝑐
673
𝐷𝑘 = 9700 × 9.4 × 10−7 √ = 0.045 𝑐𝑚2 ⁄𝑠
28

𝜖 𝐷𝑐𝑜𝑚𝑏 2
𝐷𝑒𝑓𝑓 = = 𝜖 2 𝐷𝑐𝑜𝑚𝑏 = 𝜖 1
𝜏 ⁄1 1
( + )
𝐷 𝐷𝑘

101
 2
= 0.5 1
⁄ 1 1 = 0.0087 𝑐𝑚2 ⁄𝑠
(0.045+0.155)

𝑏𝑜𝑡ℎ 𝑎𝑟𝑒
𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑡

𝑘 (1⁄𝑠) 0.8 1.06

3. 𝜙 = 𝐿√ 2 = √ = 1.471
𝐷𝑒𝑓𝑓 (𝑐𝑚 ⁄𝑠) 6 0.0087

1 1 1
Therefore, 𝜂 = [ − ] = 0.526
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙

𝑝𝐴 0.15
4. 𝐶𝐴 = ⁄𝑅𝑇 = = 2.7 × 10−6 𝑚𝑜𝑙𝑒𝑠⁄𝑐𝑚3
82×673

Robs:
𝑅𝑎𝑡𝑒 𝑚𝑜𝑙𝑒𝑠 𝜋𝑑𝑝 3
= 0.526 x 1.06 𝐶𝐴 ( )×( )
𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑠−𝑐𝑚3 6

𝜂 cm3 of particle volume

= (0.526 x 1.06 × 2.7 × 10−6 × 0.268) = 0.4 × 10−6 𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒

H.W.: 3.8, 3.9 (FB)

102
 Lecture 22
nth order reaction, intraphase transport + reaction, non-isothermal (interphase is fast)

𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐 (Δ𝐻 < 0)

𝐶𝑏 𝐶𝑝 (𝑟) 𝑇(𝑟)
𝑇𝑏

𝑒𝑛𝑑𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐 (Δ𝐻 > 0)

Write down transport equations:

𝑛
Species balance: 𝐷𝑒 ∇2 𝐶𝑝 = 𝑘𝐶𝑝

𝑟=0 ∇𝐶𝑝 = 0 (symmetric boundary condition)

𝑟 = 𝑟0 𝐶𝑝 = 𝐶𝑏 (surface condition)
𝑛
Energy balance: 𝜆𝑒 ∇2 𝑇 = (𝑘𝐶𝑝 )(−Δ𝐻) 𝜆𝑒 = 𝑡ℎ𝑒𝑟𝑚𝑎𝑙 𝑐𝑜𝑛𝑑𝑢𝑐𝑡𝑖𝑣𝑖𝑡𝑦

𝑟=0 ∇𝑇 = 0 (symmetric boundary condition)

𝑟 = 𝑟0 𝑇 = 𝑇𝑏 (surface condition)
or 𝜆𝑒 ∇2 𝑇 = 𝐷𝑒 ∇2 𝐶𝑝 (−Δ𝐻) (𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑒 𝑘𝑖𝑛𝑒𝑡𝑖𝑐𝑠 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑒𝑞𝑛)

−Δ𝐻𝐷𝑒
or ∇2 𝑇 = ( ) ∇2 𝐶𝑝
𝜆𝑒

To solve: non-dimensionalize
𝑇 𝑟
𝑓 = 𝐶𝑝 ⁄𝐶𝑏 , 𝜃= , 𝑧=
𝑇𝑏 𝐿

On substitution in the species balance equation

𝑘
∇2 𝑓 = ϕ2 𝑓 𝑛 ( ) Note: k ≠ kb and T(r) ≠ Tb (non-isothermal condition)
𝑘 𝑏

𝑘𝑏 𝐶𝑏 𝑛−1
𝜙 = 𝐿√ (See the text for the generalized Theile modulus for n = 0 and –ve order
𝐷𝑒
reactions; term within the square root contains (n+1)/2 )

103
 (𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠)

𝑅𝑜𝑏𝑠 − 𝐸⁄2𝑅
𝑙𝑛 ( )
𝐶𝐴

or ln(𝐾𝑝 ) − 𝐸⁄𝑅 (𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠)

1⁄
𝑇

H.W. (Froment & Bischoff): 3.10, 3.11.

120
 Lecture 26-27
Tubular packed bed reactor

So far, we did a single-particle analysis:

𝐶𝑏

𝑇𝑏

𝜂𝑡𝑜𝑡𝑎𝑙 = 𝜂𝑡𝑜𝑡𝑎𝑙 (𝑛, 𝐷𝑎 , 𝜙) ∶ 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙

= 𝜂𝑡𝑜𝑡𝑎𝑙 (𝑛, 𝐷𝑎 , 𝜙, 𝛽, 𝛽 ′ , 𝐴𝑟) ∶ 𝑛𝑜𝑛 − 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙

𝑅𝑜𝑏𝑠 == 𝜂𝑡𝑜𝑡𝑎𝑙 𝑅𝑏 ( 𝐶𝑏 , 𝑇𝑏 )

𝑏𝑢𝑙𝑘 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠

Now, let us consider a packed bed reactor of catalyst particles, and determine the performance or design
equation.

𝐿 𝐶𝑉
𝑄, 𝑊 (𝑘𝑔/𝑠)
𝐼𝐷 𝐹𝐴𝑓
𝑙𝑝𝑚 𝑜𝑟 𝑠𝑙𝑝𝑚
𝑍 𝐶𝐴𝑓 𝑜𝑟𝑋𝐴𝑓 =?
𝑜𝑟 𝐹𝐴𝑜 , 𝐶𝐴𝑜 Δ𝑍
(𝑚𝑜𝑙𝑒𝑠/𝑠) 𝜖𝑏 (𝑏𝑒𝑑 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦)

Rate expression: 𝑟𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 (𝑇, 𝐶). Let us assume isothermal condition.

Particles: 𝑑𝑝 , 𝜖𝜇 , 𝜖𝑚 , 𝜌𝑝 … all physical properties are determined a priori.

Reactor: 𝐿, 𝐼𝐷, 𝜖𝑏 or catalyst amount = (1 − 𝜖𝑏 )𝑉, 𝑤ℎ𝑒𝑟𝑒 𝑉 𝑖𝑠 𝑡ℎ𝑒 𝑟𝑒𝑎𝑐𝑡𝑜𝑟 𝑣𝑜𝑙𝑢𝑚𝑒

Let us focus on the 1D species conservation balance equation over CV

⇒ 𝑣𝑧 is determined from momentum balance equation (𝑣𝑧 = 𝑣𝑧 𝑖𝑛 𝑝𝑎𝑐𝑘𝑒𝑑 𝑏𝑒𝑑)

𝜌𝑧 ≠ 𝜌𝑧 (𝑍)
𝑣𝑧 ≠ 𝑣𝑧 (𝑟): 𝑎 𝑔𝑜𝑜𝑑
1D species balance on 𝐶𝐴 : 𝑎𝑠𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑎𝑡 ℎ𝑖𝑔ℎ 𝑅𝑒𝑦𝑛𝑜𝑙𝑑𝑠 𝑛𝑢𝑏𝑚𝑒𝑟

121
 𝜕𝐶𝐴
𝜖𝑏 + 𝑣𝑧 . ∇𝐶𝐴 = 𝐷𝑧 ∇2 𝐶𝐴 − (1 − 𝜖𝑏 )(𝑅𝑜𝑏𝑠 )
𝜕𝑡
𝑚𝑜𝑙𝑒𝑠/𝑠 − 𝑚3
𝑚3 𝑜𝑓 𝑣𝑜𝑖𝑑 𝑜𝑓 𝑐𝑎𝑡 − 𝑣𝑜𝑙𝑢𝑚𝑒
𝑎𝑥𝑖𝑎𝑙 𝑑𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 𝑐𝑜𝑒𝑓𝑓.
𝑚3 𝑜𝑓 𝑡𝑜𝑡𝑎𝑙 𝐶𝑉

0, 𝑆𝑆
𝜕𝐶𝐴 𝑑𝐶𝐴 𝑑 2 𝐶𝐴
Or, 𝜖𝑏 + 𝑣𝑧 = 𝐷𝑧 − (1 − 𝜖𝑏 )(𝑘𝐶𝐴 𝜂𝑡𝑜𝑡𝑎𝑙 )
𝜕𝑡 𝑑𝑧 𝑑𝑧 2

𝑑𝑖𝑠𝑝𝑒𝑟𝑖𝑠𝑜𝑛 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡
(𝐶𝐴 ≠ 𝐶𝐴 (𝑟): narrow tube ; +𝑟𝐴,𝑖 = 𝑘𝐶𝐴 ; 𝑅𝑜𝑏𝑠 = 𝜂𝑡𝑜𝑡𝑎𝑙 (𝑘𝐶𝐴 (𝑧)))

Note that the traditional 1D species conservation equation written for a homogenous reaction system is
modified by just one term, i.e. (1 − 𝜖𝑏 ) 𝜂𝑡𝑜𝑡𝑎𝑙 ! There is no other modification required!!

If you assume 𝑅𝑒𝑝 is large and there is no film resistance,

𝜂𝑡𝑜𝑡𝑎𝑙 = 𝜂𝑖𝑛𝑡𝑟𝑎 = 𝜂𝑖𝑛𝑡𝑟𝑎 (𝑛, 𝜙)

𝐶𝑏
𝐶 𝑛
In general: 𝜂𝑡𝑜𝑡𝑎𝑙 = 𝜂𝑖𝑛𝑡𝑟𝑎 (𝐶 𝑠 ) 𝐶𝑠
𝑏

𝐶𝑠 𝐶𝑠
and = (𝐷𝑎 )
𝐶𝑏 𝐶𝑏

Alternatively, recall from previous lecture:

1
𝑅𝑜𝑏𝑠 𝑛
𝑹𝒐𝒃𝒔 = 𝑎 𝑘𝑚 [𝐶𝑏 − ( ) ] (𝑠𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑒 𝑖𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝐶𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑒𝑞𝑛)
𝑘 𝜂𝑖𝑛𝑡𝑟𝑎
𝜂𝑖𝑛𝑡𝑟𝑎 = 𝜂𝑖𝑛𝑡𝑟𝑎 (𝑛, 𝜙)
𝑘𝐶𝑠 𝑛−1
𝜙 = 𝐿√
{ 𝐷𝑒

Therefore, at this stage with the single-particle analysis already done a priori, the conservation equation
can be integrated with the necessary boundary conditions.

𝑑𝐶𝐴 𝑑 2 𝐶𝐴
𝑣𝑧 𝑑𝑧
= 𝐷𝑧 𝑑𝑧 2
− (1 − 𝜖𝑏 )(𝑘𝐶𝐴 ) 𝜂𝑡𝑜𝑡𝑎𝑙

𝜕𝐶𝐴
𝑧 = 0, 𝑣𝐶𝐴𝑜 = 𝑣𝐶𝐴 −𝐷𝑧 |
𝜕𝑧 0+
(𝐶𝐴 ≠ 𝐶𝐴𝑜 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛)

𝜕𝐶𝐴
= 𝐿, 𝜕𝑧
= 0 (long-tube approximation)

122
 𝜕𝐶𝐴
𝑜𝑟 (𝑣𝐶𝐴 − 𝐷𝑧 )| − = 𝑣𝐶𝐴,𝐿 (interfacial flux balance)
𝜕𝑧 𝐿

Solve to get 𝐶𝐴 (𝑧) 𝑜𝑟 𝐶𝐴 (𝑧 = 𝐿) or 𝑋𝐴𝑓 : This is what a performance or design equation yields.

At this stage, you should have realized that the RTD study has taught us how to calculate or determine Dz.

Some discussion on 1-D vs 2-D:

𝜕2 𝐶𝐴
𝐷𝑧 2 𝐵𝐶𝑠
𝜕𝑧 2
2
D∇ 𝐶𝐴
𝐷𝑟 𝜕 𝜕𝐶𝐴
𝑟 𝜕𝑟
(𝑟 𝜕𝑟
) 2 𝐵𝐶𝑠

𝜕𝐶𝐴
𝑟=0 =0; 𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝐵𝐶
𝜕𝑟
𝜕𝐶
=𝑅 − 𝐷𝑟 𝜕𝑟𝐴 = 0; 𝑛𝑜𝑛 − 𝑟𝑒𝑎𝑐𝑡𝑖𝑣𝑒/𝑖𝑚𝑝𝑒𝑟𝑣𝑖𝑜𝑢𝑠 𝑤𝑎𝑙𝑙𝑠

𝑟𝑎𝑑𝑖𝑎𝑙 𝑑𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡.

How about if the reaction is non-isothermal? Let us say that it is exothermic and heat is removed from the
walls of the tube by convection or some cooling jacket provided on the reactor.

In such case a thermal energy balance equation is required:

𝑟𝐴 = 𝑟𝐴 (𝐶𝐴 , 𝑘(𝑇)) and 𝑇(𝑧) depends on 𝑟𝐴 (−Δ𝐻)

𝑒𝑥𝑜𝑡ℎ𝑒𝑟𝑚𝑖𝑐 ℎ𝑒𝑎𝑡 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛

Therefore, both species and energy balance equations are required, which must be solved simultaneously.

(1−𝜖𝑏 )
𝑣𝑧 . ∇𝑇 = 𝛼∇2 𝑇 + (𝑅𝑜𝑏𝑠 )(−Δ𝐻)
𝜌𝐶𝑝
𝑐𝑎𝑙⁄ :
𝑠 − 𝑚3

𝑑𝑇 𝑑2𝑇 (1−𝜖𝑏 )
1-D: 𝑣𝑧 =𝛼 + (𝑘𝐶𝐴 ) 𝜂𝑡𝑜𝑡𝑎𝑙 (−Δ𝐻)
𝑑𝑧 𝑑𝑧 2 𝜌𝐶𝑝

𝜂𝑡𝑜𝑡𝑎𝑙 (𝑛, 𝐶𝐴 , 𝑇, 𝛽, 𝛽B′ , 𝐴𝑟)

𝑎𝑥𝑖𝑎𝑙 𝑡ℎ𝑒𝑟𝑚𝑎𝑙
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛/𝑑𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡

BCs
𝑑𝑇
𝑧=0 𝑇 = 𝑇𝑜 𝑜𝑟 𝜌 𝐶𝑝 𝑣 𝑇𝑜 = (𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 𝑑𝑧 )|
0+
𝑑𝑇 𝜕𝑇
} ∶ 𝐹𝑙𝑢𝑥 balance
=𝐿 𝑑𝑧
=0 𝑜𝑟 (𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 𝜕𝑧 )| − = 𝜌 𝐶𝑝 𝑣 𝑇𝑒𝑥𝑖𝑡
𝐿

(long-tube approximation)

123
 𝛼𝑟 and 2 BCs are requried
Here also, if 2-D 𝛼
𝛼𝑧 𝑎𝑛𝑑 2 𝐵𝐶𝑠 𝑎𝑟𝑒 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑

𝜕𝑇
Bc 1: 𝑟 = 0 = 0 (symmetric condition)
𝜕𝑟

𝜕𝑇
𝑟 = 𝑅 −𝑘 (𝑐𝑎𝑙⁄𝑠 − 𝑚2 ) = ℎ(𝑇 − 𝑇𝑎 )
𝜕𝑟

= 𝑞𝐶𝑜𝑜𝑙𝑖𝑛𝑔

𝑜𝑟 𝑇 = 𝑇𝑤 (𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑒𝑑)

(Recall the constant wall temperature and constant heat flux cases in Transport Phenomena)

It is clear that you require the knowledge of transport conservation equation, in conjunction with kinetics
to design a tubular reactor or 𝑋(𝑧 = 𝐿) =?

How about the empty tubular reactors whose walls are coated with a non-porous catalyst?

𝐷 catalyst 𝐶𝐴 @ 𝐿 =?
𝐹𝐴𝑜,
𝑄, 𝐶𝐴𝑜 𝑍 Δ𝑍

Assume that 𝑅𝑒 𝑖𝑠 𝑙𝑎𝑟𝑔𝑒 ≫ 2100, 𝑣 ≠ 𝑣𝑧 (𝑟)

𝑚𝑜𝑙𝑒𝑠⁄
𝑠 − 𝑚2 : 𝑟 = 𝑘𝐶𝐴 (1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟) {
(𝑎𝑡 𝑡ℎ𝑒 𝑤𝑎𝑙𝑙𝑠) 𝐶𝐴 ≠ 𝐶𝐴 (𝑟): 1𝐷
∶ 𝑖𝑠𝑜𝑡ℎ𝑒𝑟𝑚𝑎𝑙
𝑚𝑜𝑙𝑒𝑠⁄ : 𝑣𝑧 ∇𝐶𝐴 = 𝐷∇2 𝐶𝐴 − (𝑅𝑜𝑏𝑠 ) × 𝑎
𝑠 − 𝑚3 𝑚2⁄ 𝑜𝑓 𝑡𝑢𝑏𝑢𝑙𝑎𝑟 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
𝑚3
𝑚𝑜𝑙𝑒𝑠⁄ 2𝜋𝑅Δ𝑧
𝑠 − 𝑚2 𝑎=
𝜋𝑅 2 Δ𝑧
𝑑𝐶𝐴 𝑑 2 𝐶𝐴 4
𝑣𝑧 = 𝐷𝑧 − (𝑘𝐶𝐴 ) 4⁄𝐷 × 𝜂𝑖𝑛𝑡𝑒𝑟 ∶ 𝒆𝒒. (𝟏) = 2⁄𝑅 =
𝑑𝑧 𝑑𝑧 2
𝐷
𝜂𝑖𝑛𝑡𝑒𝑟 = 𝜂𝑖𝑛𝑡𝑒𝑟 (𝐷𝑎 )

(No intra – diffusional resistance; there is a thin coating of catalyst)

Alternatively,

𝜕𝐶𝐴 𝜕2 𝐶𝐴 𝐷𝑟 𝜕 𝜕𝐶
𝑣𝑧 𝜕𝑧
= 𝐷𝑧 𝜕𝑧 2
+ ( (𝑟 𝜕𝑟𝐴 ))
𝑟 𝜕𝑟
(no reaction/source term because the reaction occurs
at the walls)

124
BCs on z: same as before

𝜕𝐶𝐴
BCs: 𝑟=0 𝜕𝑟
= 0 (𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐𝑖𝑡𝑦)

𝜕𝐶𝐴
= 𝑅 − 𝐷𝑟 𝜕𝑟
= 𝑘𝑚 (𝐶𝐴 − 𝐶𝑆 ) (interphase/film)

𝑚𝑜𝑙𝑒𝑠⁄
𝑠 − 𝑚2 : = 𝑘𝐶𝑆 @ 𝑜𝑟 𝑛𝑒𝑎𝑟 𝑠𝑢𝑟𝑓𝑎𝑐𝑒
𝑘𝑘𝑚
= 𝐶
𝑘+𝑘𝑚 𝐴

If you consider 𝐶𝐴 ≠ 𝐶𝐴 (𝑟), the governing equation is to be modified as a 1D model:

𝑑𝐶𝐴 𝑑2 𝐶𝐴 𝑘𝑘 2
𝑣𝑧 = 𝐷𝑧 − (𝑘+𝑘𝑚 ) 𝐶𝐴 × 𝑎 (𝑚 ⁄ 3 𝑜𝑓 𝑡𝑢𝑏𝑢𝑙𝑎𝑟 𝑟𝑒𝑎𝑐𝑡𝑜𝑟)
𝑑𝑧 𝑑𝑧 2 𝑚 𝑚
𝑚𝑜𝑙𝑒𝑠⁄
𝑠 − 𝑚2 (𝑠𝑎𝑚𝑒 𝑡𝑒𝑟𝑚 𝑎𝑝𝑝𝑒𝑎𝑟𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑎𝑖𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛)
𝑑𝐶𝐴 𝑑2 𝐶𝐴 4 𝑘𝑘
𝑣𝑧 = 𝐷𝑧 − 𝐷 (𝑘+𝑘𝑚 ) 𝐶𝐴 - eq. (2)
𝑑𝑧 𝑑𝑧 2 𝑚

You must realize equations (1) & (2) are the same!

What about installing a pre-heater and/or post-cooler?

𝑙1 𝑙2 𝑙3
𝐹𝐴𝑜 (𝑤)
𝐶𝐴 =?
𝐶𝐴𝑜
𝑇𝑜 𝑞𝑤 ↑
𝑝𝑎𝑐𝑘𝑒𝑑 𝑏𝑒𝑑 𝑞𝑤 ↓

Species balance in the segment 𝑙2 will be the same as before?

no reaction
CA (z = l1 ) = CAo
in pre − heater
One can also assume ∇CA |l=l1 +l2 = 0 } & post − cooler.
CA |z=(l1 +l2 +l3 ) = CA |z=(l1 +l2 )
CA = CA (T)

Energy balance over 𝒍𝟏 :

𝜌𝐶𝑝 𝑣𝑧 . ∇𝑇 = 𝑘∇2 𝑇 (no source term)

𝑑𝑇 𝑑2 𝑇
1-D: 𝑣𝑧 =𝛼 + 𝑞𝑤,1 × 𝑎/(𝜌𝐶𝑝)
𝑑𝑧 𝑑𝑧 2

ℎ𝑒𝑎𝑡 𝑓𝑙𝑢𝑥 𝑚2⁄ 𝑜𝑓 𝑡𝑢𝑏𝑒

𝑇 ≠ 𝑇(𝑟) 𝑚3
(𝑐𝑎𝑙⁄ )
𝑠 − 𝑚2

125
BCs: 𝑧 = 𝑙1 𝑇 = 𝑇𝑖𝑛𝑙𝑒𝑡 𝑡𝑜 𝑟𝑒𝑎𝑐𝑡𝑜𝑟

(known; to be specified/calculated)

𝜕𝑇
In case of 2D, −𝑘 = −𝑞𝑤 (ℎ𝑒𝑎𝑡 𝑖𝑠 𝑠𝑢𝑝𝑝𝑙𝑖𝑒𝑑): 𝑐𝑎𝑙⁄
𝜕𝑟 𝑠 − 𝑚2
↑ ↑
𝑛𝑜𝑡𝑒 𝑡ℎ𝑒 𝑠𝑖𝑔𝑛

Energy balance over 𝒍𝟑 :

𝑑𝑇 𝑑2 𝑇
𝑣𝑧 =𝛼 − 𝑞𝑤,3 × 𝑎/(𝜌𝐶𝑝)
𝑑𝑧 𝑑𝑧 2 Toutlet/exit
from the reactor
BCs 1: 𝑇 + (𝑧 = 𝑙1 + 𝑙2 ) = 𝑇 − (𝑧 = 𝑙1 + 𝑙2 )
(solve from the energy balance
over l2 )
2: ∇𝑇 = 0 @ 𝑧 = (𝑙1 + 𝑙2 + 𝑙3 )

In such problems interfacial boundary conditions are important @𝑧 = 𝑙1 , 𝑙1 + 𝑙2 .

The general BC at interface:

𝑑𝐶 𝑑𝐶
(𝑣𝑐 − 𝐷 𝑑𝑧 )| = (𝑣𝑐 − 𝐷 𝑑𝑧 )|
𝑧− 𝑧+
𝑑𝑇 𝑑𝑇
} Flux (mass or thermal)
(𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 𝑑𝑧 )| − = (𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 𝑑𝑧 )| +
𝑧 𝑧

(Considering a broad applicability of such physical models, it is difficult to suggest any assignment or
homework on this topic, yet you may try 11.2 of F&B….it is the simplest of all to solve!)

126
Many situations and BCs at the interfaces for 𝐶𝐴 𝑎𝑛𝑑 𝑇:

1. 𝑚𝑜𝑠𝑡 𝑔𝑒𝑛𝑒𝑟𝑎𝑙
𝑑𝐶 𝑑𝐶
(𝑣𝑐 − 𝐷 )| = (𝑣𝑐 − 𝐷 )|
𝑑𝑧 𝑧 − 𝑑𝑧 𝑧 +
𝜕𝑇 𝜕𝑇
(𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 )| = (𝜌 𝐶𝑝 𝑣 𝑇 − 𝑘 )|
𝜕𝑧 𝑧− 𝜕𝑧 𝑧+ }

𝑚𝑜𝑠𝑡 𝑠𝑖𝑚𝑝𝑙𝑒
2.
∇𝑇 = ∇𝐶 = 0
𝑇+ = 𝑇−
𝐶+ = 𝐶−

3.

Mixed boundary condition with the

gradient at the left of the interface set to 0

4.

Mixed boundary condition with the

gradient at the right of the interface set to 0

To this end, this lecture should teach you how to write transport equations in conjunction with
kinetics, with the latter role limited to the calculations for Dz (from RTD study) and Robs (or
𝜼𝒕𝒐𝒕𝒂𝒍 ) 𝐨𝐧𝐥𝐲!

127
 𝑁𝑜𝑡𝑒: 𝑟𝑝 𝑖𝑠 𝑡ℎ𝑒 𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 𝑟𝑎𝑡𝑒.
𝐴𝑡 𝑡ℎ𝑒 𝑖𝑛𝑐𝑖𝑝𝑖𝑒𝑛𝑡 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛
𝑤ℎ𝑒𝑛 𝑡ℎ𝑒 𝑖𝑛𝑡𝑟𝑎𝑝ℎ𝑎𝑠𝑒 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛
𝑗𝑢𝑠𝑡 𝑠𝑡𝑎𝑟𝑡𝑠 𝑡𝑜 𝑏𝑒𝑐𝑜𝑚𝑒 𝑖𝑚𝑝𝑜𝑟𝑡𝑎𝑛𝑡
𝑟𝑜 = 0.26 𝑐𝑚 𝑟𝑜𝑏𝑠 = 𝑘𝑖 𝐶𝑏

𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛
(𝑖𝑡 𝑖𝑠 𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑)
{(𝑟𝑒 − 𝑟𝑒𝑎𝑑 𝑡ℎ𝑒 𝑝𝑟𝑜𝑏𝑙𝑒𝑚 𝑠𝑡𝑎𝑡𝑒𝑚𝑒𝑛𝑡)

𝐶𝑏 𝐶𝑏

𝑟𝑜𝑏𝑠 = 𝑟𝑖 𝑟𝑜𝑏𝑠 ≈ 𝑟𝑖
(𝑘𝑖𝑛𝑒𝑡𝑖𝑐 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑑) (𝑎𝑡 𝑡ℎ𝑒 𝑖𝑛𝑐𝑖𝑝𝑖𝑒𝑛𝑐𝑒 𝑜𝑓 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑟𝑜𝑙)
(Δ𝐶𝑝𝑜𝑟𝑒 = 0) (Δ𝐶𝑝𝑜𝑟𝑒 𝑏𝑒𝑔𝑖𝑛𝑠 𝑡𝑜 𝑑𝑒𝑣𝑒𝑙𝑜𝑝)

(c) What 𝑘𝑜𝑏𝑠 𝑜𝑟 𝑘𝑎𝑝𝑝 will be measured at 510𝑜 𝐶 for the particles of 5 mm
diameter?

𝑟𝑜𝑏𝑠 = 𝜂 (𝑟𝑖𝑛𝑡 ) = 𝜂(𝑘𝑖𝑛𝑡 𝐶𝑏 ) ∶ defintion for 𝜂

𝑟𝑜𝑏𝑠 = 𝑘𝑜𝑏𝑠 𝐶𝑏 = 𝜂 𝑘𝑖𝑛𝑡 𝐶𝑏 : definition for 𝑘𝑜𝑏𝑠 ⇒ 𝑘𝑜𝑏𝑠 = 𝜂𝑘𝑖𝑛𝑡 = 𝜂 × 0.04

(𝑤ℎ𝑎𝑡 𝑖𝑓 𝑖𝑡 𝑖𝑠 𝑛𝑜𝑡 𝑎 1𝑠𝑡 𝑜𝑟𝑑𝑒𝑟 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑟 𝑠𝑜𝑚𝑒 𝑐𝑜𝑚𝑝𝑙𝑒𝑥 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒? 𝑇ℎ𝑒𝑛, 𝑖𝑡 𝑖𝑠

𝑎 𝑝𝑠𝑒𝑢𝑑𝑜 𝑓𝑖𝑟𝑠𝑡 𝑜𝑟𝑑𝑒𝑟 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛? )

𝑑𝑝 𝑘′′′ (𝑘 ′′′ = ? 𝑣𝑜𝑙𝑢𝑚𝑒 𝑏𝑎𝑠𝑖𝑠? )

𝜙= √𝐷
6 𝑒 𝑟𝑖𝑛 𝑖𝑠 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑖𝑛 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑐𝑚3 𝑜𝑓 𝑐𝑎𝑡 − 𝑣𝑜𝑙𝑢𝑚𝑒)

0.5 0.04
= √ = 0.33; 𝑘 ′′′ 𝑚𝑢𝑠𝑡 𝑏𝑒 𝑖𝑛 1⁄𝑠
6 0.0026

1 1 1
𝜂= [ − ] = 0.94
𝜙 𝑡𝑎𝑛ℎ3𝜙 3𝜙

𝑘𝑜𝑏𝑠 = 0.04 × 0.94 = 0.0376 𝑠 −1

114
Example 2: Isothermal, 1st order reaction, reaction in a porous catalyst, pore
diffusion control; How does 𝜙, 𝜂, 𝑅𝑜𝑏𝑠 (𝑇) = ?

𝑟𝑜 𝑘 𝑘𝑜 𝑒𝑥𝑝(−𝐸 ⁄𝑅𝑇 )
a) 𝜙 = √ = 𝐿√
3 𝐷 𝐷

𝐿𝑘𝑜 1⁄2 𝐸
or 𝑙𝑛 𝜙 = 𝑙𝑛 ( ) − 2𝑅𝑇 ⇒ ln 𝜙
𝐷1⁄2 −𝐸⁄
2𝑅
𝑇↑𝜙↑
1⁄
𝑇 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛
1
1 1 1 𝐷 ⁄2 𝑐𝑜𝑛𝑡𝑟𝑜𝑙
𝑏) 𝜂 = ⇒ = (𝑘 )
𝜙 𝜙 𝐿
𝐸⁄
1 2𝑅
𝐷 ⁄2 𝐸 ln(𝜂)
ln(𝜂 ) = 𝑙𝑛 ( 1 ) + 2𝑅𝑇
𝐿𝑘𝑜 ⁄2

1⁄
𝑇
𝑇↑ 𝜂↓

(There may be some effects of ‘𝑇’ 𝑜𝑛 ‘𝐷’: some non-linearity in plots)

𝑐) 𝑅𝑜𝑏𝑠 = 𝜂 (𝑘𝐶𝐴 ) 𝑠𝑢𝑟𝑓𝑎𝑐𝑒

𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 ⁄𝑖𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐

𝑅𝑜𝑏𝑠
= 𝜂 𝑘 = 𝜂(𝑇)𝑘(𝑇) (𝑘𝑜𝑏𝑠 = 𝜂 𝑘𝑖𝑛 )
𝐶𝑏

(There is a dependence of 𝑅𝑜𝑏𝑠 𝑜𝑛 ′𝑇′ through both 𝜂 & 𝑘)

𝑙𝑛(𝑅𝑜𝑏𝑠 ⁄𝐶𝑏 ) = ln 𝜂 + 𝑙𝑛 𝑘 = ln(𝑐𝑜𝑛𝑠𝑡) − 𝐸⁄2𝑅𝑇

(𝑙𝑛 𝑘 = ln 𝑘𝑜 − 𝐸⁄𝑅𝑇)

𝑅𝑜𝑏𝑠 −𝐸⁄
𝑙𝑛 ( ) 2𝑅 𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛
( )
𝐶𝑏 𝑐𝑜𝑛𝑡𝑟𝑜𝑙

1⁄
𝑇

115
Recall previous lectures: overall (reactive) mass transfer coefficient

− 𝐸⁄2𝑅 (𝑂𝑣𝑒𝑟 𝑡ℎ𝑒 𝑒𝑛𝑡𝑖𝑟𝑒 𝑟𝑎𝑛𝑔𝑒)

ln(𝐾𝑝 )
𝑑𝑖𝑓𝑓𝑢𝑠𝑖𝑜𝑛
( ) − 𝐸⁄𝑅
𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
( )
𝑐𝑜𝑛𝑡𝑟𝑜𝑙
ℎ𝑖𝑔ℎ 𝑇
𝑙𝑜𝑤 𝑇

1⁄
𝑇
or
𝐸⁄
2𝑅

ln(𝐾𝑝 )

𝐸⁄
𝑅

1⁄
𝑇

Example 3: Basket reactor (continuous operation)

𝐶𝑖𝑛

𝐶𝑜𝑢𝑡

𝑐𝑎𝑠𝑒 𝑁𝑜. 𝑝𝑒𝑙𝑙𝑒𝑡 𝑑𝑖𝑎 𝑟𝑝𝑚 𝐶𝑜𝑢𝑡 𝑅𝑜𝑏𝑠

1 1 ↑ (ℎ𝑖𝑔ℎ) 1 3
2 3 Low 1 1
3 3 ↑ 1 1

What controls ?

⇒ Case 2 & 3 (choose two conditions in which one variable is varied at a time)

116
 - Interphase transport or diffusion does not have influence. 𝑘𝑚 𝑖𝑠 high;
interphase resistance is negligible at high or low stirring speed over that range;
No Δ𝐶 across film; 𝐶𝑠 → 𝐶𝑏 .

Case 1 & 3: 𝑅𝑜𝑏𝑠 ↓ with 𝑑𝑝 ↑ : indicates the influence of intraphase. Do some

calculations.
1
If pore-diffusion controls, 𝜂 =
𝜙

𝜙1 𝑑𝑝1 1
= =
𝜙3 𝑑𝑝3 3

𝑅𝑜𝑏𝑠,1 𝜂1 (𝑘𝐶𝑏 )1 𝜂1 3 𝑁𝑜𝑡𝑒 𝑖𝑡 𝑖𝑠 𝑎 𝐶𝑏 = 𝐶𝑒𝑥𝑖𝑡 )

= = = ( :)
𝑅𝑜𝑏𝑠,3 𝜂3 (𝑘𝐶𝑏 )3 𝜂3 1 𝑤𝑒𝑙𝑙 − 𝑚𝑖𝑥𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟

- Pore-diffusion controls 𝜂
= 1⁄𝜙

𝜙
Example 4:

𝐶𝐴,𝑖𝑛 = 1 𝑚𝑜𝑙𝑒⁄𝐿 𝑊𝑡. 𝑜𝑓 𝑐𝑎𝑡𝑎𝑙𝑦𝑠𝑡

𝑣𝑜 = 𝑓𝑖𝑥𝑒𝑑
(𝑛𝑒𝑔𝑙𝑖𝑔𝑖𝑏𝑙𝑒 𝑓𝑖𝑙𝑚
𝑟𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑠, 𝑑𝑜 𝑊 𝑋𝐴
𝑐𝑜𝑛𝑠𝑡. 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒) 2 2 0.6
𝑋𝐴 , 𝐶𝐴,𝑒𝑥𝑖𝑡
6 1 0.2

⇒ 1st order

⇒ porous catalyst What controls ?

117
 ← 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑖𝑜𝑛𝑠 →

S. No. 𝑑𝑜 𝑊 𝑋𝐴 𝐶𝐴,𝑒𝑥𝑖𝑡 𝑅𝑜𝑏𝑠 𝜙 𝜂

1 2 2 0.6 0.4 3 1 3
2 6 1 0.2 0.8 2 3 1

Ans: 𝐶𝐴 = 𝐶𝐴𝑜 (1 − 𝑋𝐴 ) ⇒ 𝐶𝐴,1 = 0.4, 𝐶𝐴,2 = 0.8

Species balances equation across the reactor (design equation/performance

equation:

𝐹𝐴𝑜 𝑋𝐴 = (−𝑟𝐴 )𝑜𝑏𝑠 𝑊 𝑜𝑟, 𝑣𝑜 (𝐶𝐴𝑜 − 𝐶𝐴 ) = (−𝑟𝑜𝑏𝑠 ) × 𝑊

𝑚𝑜𝑙𝑒𝑠⁄𝑠 𝑚𝑜𝑙𝑒𝑠⁄𝑠 − 𝑔 𝑜𝑓
𝑐𝑎𝑡. 𝑔 𝑜𝑓
𝑐𝑎𝑡.

𝐹𝐴𝑜 𝑋𝐴
(−𝑟𝐴 )𝑜𝑏𝑠 =
𝑊

𝑁𝑜𝑡𝑒 𝐶𝐴,𝑏𝑢𝑙𝑘 = 𝐶𝐴,𝑒𝑥𝑖𝑡

𝑟𝐴,1 𝑥𝐴,1 𝑊2 0.6 1 3
(𝑟 ) = × = × = ( 𝑖𝑛 𝑎 )
𝐴,2 𝑥𝐴,2 𝑊1 0.2 2 2
𝑜𝑏𝑠
𝑤𝑒𝑙𝑙 − 𝑚𝑖𝑥𝑒𝑑 𝑟𝑒𝑎𝑐𝑡𝑜𝑟
Theory:
𝑟 𝜂1 𝐶𝐴,1
(𝑟𝐴,1) =
𝜂2
×(
𝐶𝐴,2
) (isothermal)
𝐴,2 𝑜𝑏𝑠 𝑏𝑢𝑙𝑘

𝜙2 0.4 3
= × = (if diffusion control, 𝜂 ∝ 1⁄𝜙)
𝜙1 0.8 2

𝜙2 𝐿2 𝑑𝑝,2 6 3
( = = = = )
𝜙1 𝐿1 𝑑𝑝,1 2 1

- Diffusion controls

118
 ChE TRIAD e-LECTURE SERIES
About the Author
Nishith Verma is currently the Chevron Corporation Chair Professor of
chemical engineering at IIT Kanpur (India). Having pursued B. Tech in
Lectures can be downloaded from www.iitk.ac.in/che/nv.htm

chemical engineering (1982-86) from IIT Kharagpur (India) and Ph.D. in

chemical engineering (1991-95) from the University of Arizona, Tuscon,
USA, Prof. Verma worked as a process engineer (1995-97) at BOC Gases
(now Linde Engineering), NJ, USA, before joining IIT Kanpur as an assistant professor. His
research interests are environmental remediation techniques, nanomaterials, catalysts,
sensors, and lattice Boltzmann methods-based modelling. Prof. Verma is the recipient of
Alexander von Humboldt Research Fellowship and Fulbright-Nehru Academic and Professional
Excellence Fellowship. He was Head, Chemical Engineering Department, 2011 – 13 and
Coordinator, Center for Environmental Science and Engineering, 2011 – 14, both at IIT Kanpur.

From the Author

ChE Triad e-Lecture Series materials were prepared during the pandemic when faculty were
quarantined at home, and classes were held in online mode. It was a unique and first-hand
experience for me to deliver online lectures. The online lectures ran for almost three academic
years or six semesters. During that period, we also uploaded lecture materials at the Institute’s
e-server, which students accessed. As an instructor, I taught Transport Phenomena, Finite
Difference-based Numerical Methods, and Heterogeneous Chemical Reaction Engineering at
the graduate level. The lecture materials that I prepared during the online classes have
culminated in ChE Triad e-Lecture Series. Most of the Indian as well as foreign universities
consider these three courses to be compulsory for graduate students. I hope that this e-lecture
series are useful to readers, but I must caution students that they should not wholly depend on
the materials covered in the series. Rather, they should refer the standard text books
prescribed by their course instructors in regular offline classes. I have my opinion that a
comprehensive understanding of the topics is possible only through the offline lectures. Thus, I
have a plan to develop audio lectures in the near future, which will be instructive and recorded
in sync with the present e-lectures. To this end, this effort is painfully dedicated to all the Covid
warriors who risked their health and life in serving the people during the pandemic, while we
enjoyed delivering online lectures at home.
December 2022, IIT Kanpur

Nishith Verma
Indian Institute of Technology Kanpur
(nishith@iitk.ac.in, vermanishith@gmail.com)