Table of Contents

1 Abstract .................................................................................................... 1

2 Introduction ............................................................................................. 2

3 Governing Equation.................................................................................. 3

4 Finite Difference Equation ....................................................................... 4

5 Initial and Boundary Conditions .............................................................. 5

6 Modeling .................................................................................................. 6

7 Discussion ................................................................................................. 7

References ……………………………………………………...................III
1 –Abstract

A model for homogeneous-heterogeneous reactions in boundary-layer

flow is presented in which the homogeneous reaction is represented by
cubic autocatalysis and the heterogeneous reaction by a first-order process.
The flow is of a uniform stream over a flat surface .

It is shown that the surface reaction is the dominant mechanism near the
leading edge.

Numerical solutions of the governing equations reveal that the

homogeneous reaction dominates downstream with this reaction taking
place in a narrow region located well away from the surface. An asymptotic

2-Introduction

We consider a simple model for the interaction between a homogeneous

reaction and a heterogeneous (or surface) reaction involving the two
chemical species A and B in a boundary layer flow.
We assume that the homogeneous reaction can be represented by cubic
autocatalysis, written schematically as

A+2B 3B rate 𝒌𝒄 𝒂𝒃𝟐 (1)

and that the heterogeneous reaction is the simple first-order step

A B rate 𝒌𝒔 a (2)

Here a and b are the concentrations of the species A and B, respectively,

and the ki (i = c,s) are the rate constants.
We also assume that both reaction processes are isothermal.
3-Governing Equation
Many important physical processes in nature are governed by PDEs. For this
reason, it’s important to understand the physical behavior of the model
represented by the PDE. In addition, knowledge of the mathematical
character, properties and solution of the governing equations is required.

The governing equation of this case is the energy equation are below .
𝜕𝑢 𝛿𝑣
+ =0 (3)
𝜕𝑥 𝛿𝑦

𝜕𝑢 𝜕𝑢 𝜕2 𝑢
𝑢 +𝑣 =𝑣 (4)
𝜕𝑥 𝜕𝑦 𝜕𝑦 2

𝜕𝑎 𝜕𝑎 𝜕2 𝑎
𝑢 +𝑣 = 𝐷𝐴 − 𝑘𝑐 𝑎𝑏2 (5)
𝜕𝑥 𝜕𝑦 𝜕𝑦 2

𝜕𝑏 𝜕𝑏 𝜕2 𝑏
𝑢 +𝑣 = 𝐷𝐵 + 𝑘𝑐 𝑎𝑏 2 (6)
𝜕𝑥 𝜕𝑦 𝜕𝑦 2

4-Initial and Boundary Conditions

For the boundary-layer flow, we take the simplest case of a uniform stream
Uo , in which there is a uniform concentration ao, of reactant A and in which
there is no auto catalyst B, over a the flat surface.

We assume that there is no feedback from the chemical reactions to the flow
and, consequently, the boundary-layer flow will be given by the standard
Blasius similarity solution;
𝜕𝑎 𝜕𝑏
𝑢 = 𝑣 = 0 , 𝐷𝐴 = 𝑘𝑠 𝑎 , 𝐷𝐵 = − 𝑘𝑠 𝑎 𝑜𝑛 𝑦 = 0 (𝑥 > 0) (7)
𝜕𝑦 𝜕𝑦

𝑢 Uo , a ao , b 0 as y=∞ (𝑥 > 0) (8)

𝑢 = 𝐔𝐨 , 𝐚 = ao , b=0 0n x=0 ( y> 0 ) (9)

Where x and y are, respectively, streamwise and transverse coordinates with

velocity components 𝑢 and v . v is the kinematic viscosity and 𝑫𝑨 and
𝑫𝑩 are the diffusion coefficients of species A and B.
Here we consider the interplay between surface and homogeneous reactions
as given by (l),(2) We show that the heterogeneous reaction (2) is required to
produce some auto catalyst B, with the consequence that the surface reaction
is the dominant mechanism near the leading edge.

The effects of the homogeneous reaction (1) become increasingly important

as the boundary layer develops and at large distances from the leading edge,
it is this reaction which becomes the dominant mechanism.

At large distances downstream, we find that the surface reaction has been

effectively switched off (as the concentration of the reactant A close to the
surface falls to zero) and that the reaction is confined to a relatively thin
region located at increasingly large distances from the surface.

Now we start by expressing equations (3)-(9) in dimensionless form.

The model is given by equations (3)-(9) has prescribed scales Uo and ao for
velocity and concentration, but does not have a prescribed length scale 𝒍 the
choice for 𝒍 is, to some extent, at our disposal and we find it convenient to
take
𝐔𝐨
𝑙=𝑘 (10)
𝑐 𝑎𝑜2

We note that this choice for 1 gives a ratio of the strength of convective
effects, of O(𝐔𝐨 𝐚𝐨/ 𝒍 ) to the strength of the homogeneous reaction, of
O(𝒌𝒄 𝒂𝒐𝟑 ).

We now write
𝑥
𝑢 = Uo 𝑢̅ , 𝑣 = Uo R 𝑒−0.5 𝑣̅ ̅ , 𝑥̅ =
, 𝑎 = 𝑎𝑜 𝑎 ,
𝑙
𝑦𝑅𝑒 0.5 Uo 𝑙
𝑦̅ = , 𝑅𝑒 = (11)
𝑙 𝑣

with equations (3)-(6) becoming (on dropping the bars for convenience)
𝝏𝒖 𝜹𝒗
+ =𝟎 (12)
𝝏𝒙 𝜹𝒚
𝜕𝑢 𝜕𝑢 𝜕2 𝑢
𝑢 +𝑣 =𝑣 (13)
𝜕𝑥 𝜕𝑦 𝜕𝑦 2

𝜕𝑎 𝜕𝑎 1 𝜕2 𝑎
𝑢 +𝑣 = − 𝑎𝑏 2 (14)
𝜕𝑥 𝜕𝑦 𝑆𝑐 𝜕𝑦 2

𝜕𝑏 𝜕𝑏 ɸ 𝜕2 𝑏
𝑢 +𝑣 = + 𝑎𝑏 2 (15)
𝜕𝑥 𝜕𝑦 𝑆𝑐 𝜕𝑦 2

where 𝑆𝑐 = 𝑣/𝐷𝐴 is the Schmidt number and ɸ = 𝐷𝐵 /𝐷𝐴 is the ratio of the
diffusion coefficients.
So boundary conditions (7)-(9) become

𝜕𝑎 𝜕𝑏
𝑢 =𝑣 =0, = 𝑘𝑠 a , ɸ= − 𝑘𝑠 a 𝑜𝑛 𝑦 = 0 (𝑥 > 0) (16)
𝜕𝑦 𝜕𝑦
𝑢 1 , a 1 , b 0 as y=∞ (𝑥 > 0) (17)
𝐮 = 𝟏 , 𝐚 = 𝟏 ,𝐛 = 𝟎 𝐨𝐧 𝒙 = 𝟎 (𝒚 > 𝟎) (18)

𝒌 𝒗
where 𝑲𝒔 = ( 𝒔 )( )𝟎.𝟓 is a dimensionless measure of the
𝑫𝑨 𝒌𝒄𝒂𝒐𝟐)
heterogeneous reaction.
In general, the diffusion coefficients of chemical species 𝑨 and 𝑩 will not be
equal, though we could expect them to be comparable in size. In the special
case where 𝑫𝑨 = 𝑫𝑩 (i.e., ɸ = l),
equations (14),(15) can be combined to give

𝑎(𝑥, 𝑦) + 𝑏(𝑥, 𝑦) = 1 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑥≥0 ,𝑦 ≥ 0 (19)

leading to the single equation

𝜕𝑎 𝜕𝑎 1 𝜕2 𝑎
𝑢 +𝑣 = − 𝑎(1 − 𝑎)2 (20)
𝜕𝑥 𝜕𝑦 𝑆𝑐 𝜕𝑦 2

still subject to boundary conditions (16)-(18). We start by considering this

case in detail and then assess the modifications that arise to the behaviour of
the solution when ɸ ≠ 𝟏 .
(So when ɸ = 1)
Here we consider the solutions to equations (12), (13)) and (20) subject to
boundary conditions (16)-(18).
We start by making the transformation, motivated by the Blasius solution [l]
𝒚
Ū = 𝒙𝟎.𝟓 𝒇(ƞ), 𝒂 = 𝒂(𝒙, ƞ) ƞ= (21)
𝒙𝟎.𝟓

This leads to the standard Blasius problem for 𝒇(ƞ) and to the equation
1 𝜕2 𝑎 1 𝜕𝑎 𝑑𝑓 𝜕𝑎
2
+ 𝑓 − 𝑥𝑎(1 − 𝑎)2 − 𝑥 =0 (22)
𝑆𝑐 𝜕ƞ 2 𝜕ƞ 𝑑ƞ 𝜕𝒙

subject to the boundary conditions

𝜕𝑎
= 𝑘𝑠 𝑥 0.5 𝑎 𝑜𝑛 ƞ=𝟎 𝒂 1 𝑎𝑠 ƞ ∞ (23)
𝜕ƞ

We start by obtaining a solution of equations (22),(23) valid for small 𝒙.

A) Solution near the Leading Edge, 𝒙 << 1
Boundary condition (23) suggests looking for a solution by expanding

𝑎(𝑥, ƞ) = 𝟏 + 𝒂𝟏(ƞ)𝒙𝟎.𝟓 + 𝒂𝟐(ƞ)𝒙 ………… (24)

The details are straightforward and need not be given here, except to
mention that the surface
reaction constant 𝑲𝒔 can be factored from the equations at each order, and
that the homogeneous reaction term does not enter the expansion until
O(𝒙𝟐 ) .
We find, from a numerical solution of the resulting ordinary differential
equations for the ai(ƞ) that, for 𝑺𝒄 = 1.

𝟑
𝑎𝑠 = 1 − 2.1788𝑲𝒔 𝒙𝟎.𝟓 + 𝟒. 𝟎𝟏𝟔𝟔𝑲𝒔 𝟐 𝒙 − 𝟔. 𝟔𝟐𝟎𝟖𝑲𝒔 𝟑 𝒙𝟐 + (𝟏𝟎. 𝟎𝟑𝟒𝟕𝑲𝒔 𝟒 − 𝟑. 𝟕𝟏𝟐𝑲𝒔 𝟐 )𝒙𝟐

(25)
B) Solution for 𝒙 is Large
Here we consider a solution of equation (20) for large 𝒙. The numerical
results plotted and shows that here the homogeneous reaction is confined to
a relatively narrow region at large distances from the surface with a
balancing of terms in equation (20) suggesting that this
Reaction zone will have a thickness of O(1) in y. This, together with the
result that
Ū~𝐲 − ∆𝒙𝟎.𝟓 𝒇𝒐𝒓 𝒚 >> 𝒙𝟎.𝟓
(26)

where ∆ = 1.72079 (from [l]), leads us to write

𝑦 = 𝑌 + 𝑅(𝑥) , 𝑎 = 𝑎(𝑥, 𝑌) (27)

where 𝒚 = 𝑹(𝒙) >> 𝟏 𝒇𝒐𝒓 𝒙 >> 𝟏 is the location of the reaction zone
Substituting (26),(27) into equation (20) leads us to conclude that 𝑹(𝒙) is
O(𝒙) as 𝒙 ∞ and suggests the expansion
𝑅(𝑥) = 𝑅0 (𝑥) + 𝑅1 𝑥 0.5 (28)
For constants 𝑹𝟎 , 𝑹𝟏 to be determined, and to look for a solution of the form
𝑎(𝑥, 𝑌) = 𝑎0 (𝑌) + 𝑎1 (𝑌)𝑥 −0.5 + ⋯ (29)
At leading order, we obtain the equations.
1
𝑎0′′ + 𝑅𝑜 𝑎𝑜′ − 𝑎𝑜 (1 − 𝑎𝑜 )2 = 0 (30)
𝑆𝑐

Thus
𝑎𝑜 1 , 𝑌 ∞ , 𝑎𝑜 0 𝑌 ∞ (31)
(where primes denote differentiation with respect to( Y). Now equation (30)
is the cubic-Fisher travelling wave equation and has been discussed in detail
by Billingham and Needham [12].

To put the equation in the form treated in [12], we first write ̅𝒀 = 𝒀 𝑺𝒄 𝟎.𝟓
and then, from [12], we have that .
 ̅̅̅
𝑌
1 exp ( )
𝑅0 = , 𝑎0 (𝑌̅) = √2
̅̅̅ (32)
(2𝑆𝑐 0.5 ) 𝑌
1+𝑒𝑥𝑝( )
√2

At O(𝑥 −0.5 ), we obtain the equation

𝑆𝑐 0.5
𝑎1′′ + 𝑅0 𝑆𝑐 0.5 𝑎1′ − (1 − 4𝑎0 + 3(𝑎0 )2 𝑎1 = (∆ − 𝑅1 )𝑎0 (33)
2

Subject to
𝑎1 0 𝑎𝑠 𝑌̅ ∞ (34)
Primes now denote differentiation with respect to 𝒀 ̅ , now equation (33) has
a complementary function 𝒂𝟎 ′ which satisfies boundary conditions (34), and
hence, for the full system to have a solution, the compatibility condition .
∞ 𝑌 ̅
(∆ − 𝑅1 ) ∫−∞(𝑎0 )2 exp ( ) 𝑑 𝑌̅ = 0 (35)
2 √

Must be satisfied. It is clear that the integral in (35) is nonzero, and hence,
we must have
𝑅1 = ∆ (36)
From (32),(36), we then have that
𝑥
𝑅𝑥 ~ + ∆𝑥 0.5 + ⋯ 𝑥 ∞ (37)
2𝑆𝑐 0.5

The reaction zone in our numerical was calculated integrations of equations

(22),(23), which we took to be at 𝒚 = 𝒚𝑹 (𝒙), with 𝒚𝑹 defined by.

∞
𝒚𝑹 (𝑥) = ∫0 (1 − 𝑎)𝑑𝑦 = 0 (38)

We now consider the general case, ɸ ≠ 1 , concentrating on the differences

that arise between this case and the case when ɸ = 1 described above.
General case , ɸ ≠ 1
Here we consider the solution of equations (12)-(18) for , ɸ ≠ 1. We start by
noting that we can obtain a solution valid for , 𝒙 small by first applying
transformation (21) and then expanding 𝑎(𝒙, ƞ) via (24) and 𝑏(𝒙, ƞ) as,

𝑏(𝑥, ƞ) = 𝑏1 (ƞ) 𝑥 0.5 + 𝑏2 (ƞ)𝑥 + ⋯ (39)

3
The equations for the 𝒂𝒊 , 𝒃𝒊 are decoupled up to the term of O(𝑥 2 ), and
hence, the solution for 𝒂𝒊 is the same as for the case when ɸ = 𝟏, being
independent of ɸ up to this order, the solution of the equations for the 𝒃𝒊 do
depend on ɸ. It is only at O (𝒙𝟐 ) that the effects of the homogeneous
reaction enter the solution with both 𝒂𝟒 (ƞ) and 𝒃𝟒 (ƞ) depending on ɸ.
Thus, we expect the initial development of 𝒂𝒔 , to be insensitive to the value
of ɸ, though we expect 𝒃(𝒙, 𝟎) = 𝒃𝒔 , to depend on ɸ for small 𝑥. This
was confirmed from numerical integration of equations (12)-(18), where
we found that 𝒂𝒔 , approached zero relatively rapidly for all the values of ɸ
tried (from ɸ = 0.1 to ɸ = 10), following closely the behaviour for ɸ = 1.
The behaviour for 𝒃𝒔 , is somewhat different and depends on whether ɸ > 1
or ɸ < 1. The results that shown numerically , where we see that for ɸ > 1,
𝒃𝒔 , increases monotonically from zero, approaching its downstream value
(𝒃𝒔 𝟏 only very slowly (as compared to ɸ = 1).
For ɸ < 1, 𝒃𝒔 attains a maximum value (of 𝒃𝒔 ≈ 2.16 at 𝒙 ≈ 1.64 for
ɸ = 0.2) before falling (again very slowly) to its asymptotic value.

We next consider the concentration profiles a(y), b(y) at large distances.

(for ɸ = 10 and 0.2). Both cases show that the reaction zone has become
distant from the wall, and that there is a region relatively close to the wall
where there is no reaction (since a = 0) with b taking values below unity (for
ɸ > 1) or above unity (for ɸ < 1).
The structure of the reaction zone for 𝒙 large is very similar to that described
in the previous
section. at leading order, the reaction zone is located at y =𝑹𝟎 𝒙, with the
equations for at 𝒂𝟎 (Y), 𝒃𝟎 (Y) being
Thus
1
𝑎0′′ + 𝑅0 𝑎0′ − 𝑎0 𝑏0 2 = 0 (40)
𝑆𝐶

ɸ
𝑏0′′ + 𝑅0 𝑏0 + 𝑎0 𝑏 2 0 = 0 (41)
𝑆𝐶

subject to

𝑎0 1 , 𝑏0 0 , 𝑎𝑠 𝑌 ∞ (42)
𝑎0 0 , 𝑏0 1 , 𝑎𝑠 𝑌 −∞ (43)

The problem given by (40)-(43) has been discussed in detail by Billingham

and Needham [13] where the existence of a solution for all ɸ is established.
From (13)1, it follows that 𝑅0 depends on ɸ and has to be determined by
numerical integration.
It finally remains to consider the region near the surface where 𝒃 slowly
approaches its asymptotic value. A consideration of equation (15) suggests
that this region will have a thickness of.
O(𝒙𝟎.𝟓 ). We then put
𝑦
𝑏 = 1 + 𝑥 −𝑚 𝐵(𝑥, ƞ) , ƞ= (44)
𝑥 0.5
When (44) is substituted into equation (15) (with the reaction term put to
zero) and a solution
sought by expanding in inverse powers of 𝒙 , at leading order, we obtain
an eigenvalue problem for the exponent (m), namely.
ɸ 1
𝐵′′ + 𝑓𝑏 ′ + 𝑚𝑓 ′ 𝐵 = 0 (45)
𝑠𝑐 2

𝐵′ (0) = 0 , 𝐵 0 as ƞ ∞ (46)
A consideration of equations (45),(46) reveals that the smallest eigenvalue is
m = l/2 for all ɸ and 𝒔𝒄 , with corresponding solution.

𝒔𝒄 ƞ
𝐵(ƞ) = 𝐵0 exp(− ∫ 𝑓(𝑠)𝑑𝑠 ) (47)
2ɸ 0

for some constant 𝐵0 , which cannot be determined from the asymptotic

analysis. Thus, we have that.

𝒃𝒔 = 𝟏 + 𝑶(𝒙−𝟎.𝟓 ) as 𝒙 ∞ (48)

We have considered a simple model for the interaction between a surface

reaction and a homogeneous reaction in a boundary-layer flow, we have
shown that the initial development is dominated by the surface reaction with
the effects of the homogeneous reaction becoming important only as the
flow develops from the leading edge.
These effects can lead to an effective ignition of the reaction in the fluid,
manifested by a very rapid drop in the surface concentration of reactant A.
We have seen that, at large distances from the leading edge, it is the
homogeneous reaction which dominates and that this reaction is confined to
a narrow region located at large distances from the surface. An asymptotic
analysis for large 𝒙 revealed that this reaction zone is at distances O(𝒙)
from the surface, which is outside the boundary layer (having thickness of
O(𝒙𝟎.𝟓 )).
The present model may prove useful in giving insights into the general
problem of heterogeneous homogeneous catalysis, especially if we exploit
the ‘contact equivalence’ between cubic autocatalysis and the Arrhenius
temperature dependence of reaction rates (and then equate the parameter
(ɸ )with the Lewis number).
The similarity between the two forms motivated, in part, theoriginal
discussion of cubic autocatalysis by Gray and Scott [14,15] and
subsequently used in a
wide variety of applications in both spatially uniform and reaction-diffusion
contexts by them and coworkers; much of this work is detailed in their
recent book [lS].
The modelling of the Arrhenius law by autocatalysis has two significant
advantages in the mathematical treatment of the model. First, the simplicity
of the functional form allows much greater progress to be made analytically,
a feature of the present model. This is important in gaining insight into the
model and in understanding the basic mechanisms involved. Second, the
‘cold boundary problem’ is obviated. This can be seen in the present model,
that by having no autocatalysis in the outer flow, the requirement of having
no reaction outside the boundary
layer region is satisfied automatically. With the Arrhenius law, some
artificial device has to be employed to ensure no reaction outside the region
of interest. There is then the extra problem of deciding whether any
structures seen, particularly if these involve the subtle interplay between
competing mechanisms, are a real feature or are a result of the mathematical
artifact used. There are differences between the two cases with, perhaps, the
most significant for the present model being the heat transfer from the (hot)
reaction region to the (cooler) environment. This effect could be included in
our model by a simple loss term, as was originally done in [14,15], for
example, and the influence of this extra term will need to be assessed.

References III

1.1 Modeling and Simulation of Catalytic Reactors for Petroleum Refining

(by JORGE ANCHEYTA)…

1.2 Chemical-reaction-kinetics-concepts-methods-and-case-studies (2)(by

Jorge Ancheyta Instituto Mexicano del Petróleo Mexico City, Mexico
1.3 1.3 A model for isothermal homogeneous heterogeneous reactions in
boundary-layer flow( J. H. MERKIN Department of Applied Mathematics,
University of Leeds)