Вы находитесь на странице: 1из 15

CHE 3167 (JUN 10)

CHE 3167: Transport Phenomena


& Numerical Methods

Final Exam - Solution

Date: 06/2010 Total Marks: 100

1) In one common class of polymerization processes, monomer is converted to solid polymer


on the surface of a catalyst particle. As the polymer is produced, it forms a porous layer
around the catalyst particle. The monomer must therefore diffuse through the solid polymer
layer before reacting at the catalyst particle surface. The situation is depicted in the figure
below for a spherical catalyst particle, where the monomer is species A.

The catalyst particle has a radius R. The porous polymer coating has thickness t. The
concentration of species A at the outer edge of the coating is cA0 . The transport of species
A in the porous polymer is described by a constant effective diffusivity DA . We can assume
that (1) the reaction at the catalyst surface is infinitely fast, i.e cA = 0. (2) the thickness
t changes sufficiently slowly that the concentration profile reaches steady state (the quasi-
steady state assumption), and (3) the concentration of species A does not depend on or
(spherical coordinates).

(a) Set up the differential equation that governs the concentration distribution in the poly-
mer coating by making a shell balance.

(b) Integrate the differential equation twice to obtain the concentration distribution in terms
of two integration constants.

(c) Use the boundary conditions to determine the integration constants.

1
CHE 3167 (JUN 10)

(d) Show that this leads to the following concentration profile for the quasi-steady-state
concentration profile cA (r), ! "
1 1

cA r R
= ! "
cA0 1 1

R+t R

(e) What is the total rate of consumption, WA , of species A on the surface of a catalyst
particle.
(20 marks)

Solution: (a) Using a shell control volume we can write the shell balance as,
# #
(Jr 4r 2 )#r (Jr 4r 2 )#r+r = 0

Dividing throughout by 4r and taking the limit as r goes to zero, we have

# #
(r 2 Jr )#r (r 2 Jr )#r+r
lim =0
r0 r
Which implies,
d 2
(r Jr ) = 0 (1.1)
dr
Substituting Ficks law into eqn (1.1),
$ ! "% $ %
d 2 dcA dcA
r D =0 since Jr = D
dr dr dr

The governing differential equation is consequently,


! "
d 2 dcA
r D =0
dr dr

(b) If we integrate the differential equation once, we get,

dcA
r2D =A
dr

dcA A
Or, =
dr Dr 2
Integrating the equation one more time,
A
cA = +B
Dr

2
CHE 3167 (JUN 10)

(c) The boundary conditions for this problem are as follows:

(i) BC 1: At r = R, the reaction at the catalyst surface is infinitely fast, so cA = 0


(ii) BC 2: On the surface, r = R + t, we are prescribed the surface concentration, cA = cA 0

As a result, applying BC 1,
A
0= +B
DR
A
Or, B =
DR
which implies, ! "
A 1 1
cA =
D r R
Applying BC 2, ! "
A 1 1
cA 0 =
D R+t R
Or,
A cA0
=! "
D 1 1

R+t R
(d) The solution of the differential equation is therefore,
! "
1 1

cA r R
=! "
cA0 1 1

R+t R

(e) The total rate of consumption, WA , of species A on the surface of a catalyst particle is
given by
WA = (Jr 4r 2 )#R
#

We know that,
# dcA ## A D cA0
Jr #R = D = = ! "
dr R R2 1 1
R2
R+t R
As a result,
4D cA0
WA = ! "
1 1

R+t R

3
CHE 3167 (JUN 10)

2) Consider a spherical ball, initially at a uniform temperature of T0 = 98 C. The sphere is


suddenly quenched by dropping it into a large bath held at a constant uniform temperature
Tbath = 23 C. The ball is made of lead, with the following properties:

Radius R = 10 mm, density = 11, 340 kg/m3 , Cp = 129 J/kgK, k = 34 W/mK

The coefficient of heat transfer between the bath and the sphere is known to be h = 5
W/m2 K.

(a) Simplify the full heat equation, stating all assumptions that have been made to eliminate
any terms.

(b) How many boundary/initial conditions are necessary to solve this simplified equation.
Justify your answer.
(c) Identify these boundary/initial conditions.

(d) The full solution to this equation is very complicated. Therefore we make one further
assumption, that the sphere is nearly uniform in temperature (i.e T (r) = T (R) for
all r), and that the heat loss is governed by the rate of heat loss into the fluid at the
spheres surface. By carrying out an averaging procedure, the simplified heat equation
then becomes,
T
Cp V = hA (T Tbath )
t
where, V is the volume and A the surface area of the sphere. (You do not need to
derive this).

(i) Solve this equation for the temperature as a function of time, T (t).
(ii) Under this assumption, how long would it take for the sphere to reach 60 C?

Solution:
(a) The full heat equation in spherical coordinates is,
! " $ ! "
T T v T v T 1 2 T
Cp + vr + + =k 2 r +
t r r r sin r r r

2T
! " %
1 T 1
sin + 2 2 + v
r 2 sin r sin 2
The assumptions made are as follows:

There is no external velocity field, therefore vr , v and v are all zero


From the geometry we do not expect any dependence on either or , therefore
any terms containing these derivatives are zero
The final term (the dissipation function) is usually zero except in the presence of
very large velocity gradients. As we have zero velocity, this term is also zero.

4
CHE 3167 (JUN 10)

This results in the final equation


! ! ""
2 T 2 T
Cp r =k r
t r r

(b) The above equation is first order in time and second order in space, therefore we would
expect 3 boundary conditions to solve this equation.

(c) (i) Initial Condition (at t = 0): T (r, 0) = T0 = 98 C.


T ##
(iii) Boundary Condition (symmetry at r = 0): =0
r r=0
#
T ## & '
(iii) Boundary Condition (heat flux at r = R): k = h T (R, t) Tbath
r r=R
#

(d) (i) The differential equation can be solved for T as follows:

T
Cp V = hA (T Tbath )
t
T ( t
1
(
Cp V dT = hA dt
T0 T Tbath 0
! "
T Tbath
Cp V ln = h A t
T0 Tbath

Therefore, " !
hA
T = Tbath + (T0 Tbath ) exp t
Cp V
The spherical ball cools exponentially in time.
(ii) Substituting T0 = 98 + 273 = 371, Tbath = 23 + 273 = 296, A = 4R2 = 0.001257,
4
V = R3 = 4.1888 106 , and T = 60 + 273 = 333, we get
3
11340 129 4.189 106 (333 296)
t= ln
5 0.001257 (371 296)

Therefore t = 688.91 s = 11 min 29 sec

5
CHE 3167 (JUN 10)

3) The figure above illustrates the fully developed Couette flow of a Newtonian fluid hav-
ing viscosity , density and thermal conductivity k, between two infinite parallel plates
separated by a distance L. The upper plate at y = L moves with a constant velocity u1
while the lower plate remains stationary. The lower and upper plates are kept at uniform
temperatures T0 and T1 respectively. The flow is thermally developed i.e. the temperature
varies in the y direction only. Physical properties of the fluid can be assumed to be inde-
pendent of temperature. The heat transfer problem characterised with this simple model
is important for a journal and its bearing in which the clearance between the surfaces is
small, and is filled with a high viscosity lubricant oil. The temperature rise in the fluid
due to friction may become considerable even at moderate flow velocities, and consequently
the viscous energy dissipation is considered to be a significant contributor in this problem.
The temperature rise and the amount of heat transfer through the walls are of interest in
engineering applications.

(a) What are the boundary conditions for the fluid velocity field? Check if the boundary
conditions are satisfied by the velocity profile given by:
y
u(y) = u1 (3.1)
L

(b) What boundary conditions must the fluid temperature satisfy?

(c) Simplify the appropriate form of the steady state energy equation for this problem. Note
that the viscous energy dissipation term cannot be neglected.

(d) Use the velocity profile given in (a) and the energy equation obtained in (c) to obtain
a differential equation for T in terms of , k, u1 , L and y.

(e) Solve this differential equation and show that the temperature profile is given by:
$ %
T (y) T0 y 1 ) y*
= 1 + Pr E 1 (1)
T1 T0 L 2 2

6
CHE 3167 (JUN 10)

cp u1 2
where, Pr = is the Prandtl number and E = is the Eckert number.
k cp (T1 T0 )

(f) Show that the heat flux at the upper wall (y = L) is given by:
! "
T1 T0 1
q|y=L = k 1 Pr E (2)
L 2

(20 marks)

Solution:
(a) The velocity of the fluid at the upper and the lower plates is specified in the problem
statement for all x as: +
0 at y = 0 for all x,
u=
u1 at y = L for all x.

If we substitute y = 0 and y = L in eqn (3.1), we get u|y=0 = 0 and u|y=L = u1


respectively. Thus eqn (3.1) satisfies the boundary conditions for the velocity field.
(b) The temperatures at the lower and the upper walls are specified in the problem statement
for all x as: +
T0 at y = 0 for all x,
T =
T1 at y = L for all x.

(c) The energy equation in Cartesian coordinates is:


$ 2
2T 2T
! " %
T T T T T
cp + ux + uy + uz =k + + 2 + u
t x y z x2 y 2 z
where u is called the dissipation function and is given by:
,! "2 ! "2 ! "2 - ! "2 ! "2
ux uy uz uy ux uz uy
u = 2 + + + + + +
x y z x y y z
! "2 ! "2
ux uz 2 ux uy uz
+ + + +
z x 3 x y z
Since uy = uz = 0, u = ux , and the flow is fully developed,
! "2 ! "2
ux u
u = =
y y
For the present problem, the temperature varies in the y direction only and hence the
derivatives of T with respect to x will vanish in the energy equation. Substituting u
in the energy equation yields the following ordinary differential equation:
! "2
d2 T (y) du
=
dy 2 k dy

7
CHE 3167 (JUN 10)

(d) Using the velocity profile given in eqn (3.1), the differential equation takes the form:

2 T (y) u1 2
= (3.2)
y 2 k L2

(e) Integration of the differential equation (3.2) twice leads to:

u1 2 y 2
T = + C1 y + C2 (3.3)
kL2 2
Using the boundary conditions given in (b) we can evaluate C1 and C2 as:

T1 T0 u1 2
C1 = +
L 2kL
C2 = T0

Substituting C1 and C2 into eqn (3.3) gives the temperature profile

u12 u1 2
$ %
y y
T T0 = (T1 T0 ) 1+
L 2k (T1 T0 ) L 2k (T1 T0 )
which can be rearranged to be,
$ %
T (y) T0 y 1 ) y*
= 1 + Pr E 1 (3.4)
T1 T0 L 2 L

cp u1 2
where, Pr = is the Prandtl number and E = is the Eckert number.
k cp (T1 T0 )
(f) The heat flux at the upper wall is determined from its definition,
#
dT (y) ##
q|y=L = k
dy #y=L

The derivative is obtained from eqn (3.4):


$ % $ %
dT (y) T1 T0 Pr E ) y* y Pr E 1
= 1+ 1 + (T1 T0 )
dy L 2 L L 2 L
which simplifies at y = L to,
# ! "
dT (y) ## T1 T0 1
= 1 Pr E
dy #y=L L 2

Then the heat flux at the upper wall becomes:


! "
T1 T0 1
q|y=L = k 1 Pr E
L 2

8
CHE 3167 (JUN 10)

4) The figure above shows the velocity and temperature profiles in a thick paste, flowing
between two parallel plates, distance 2B apart. The length and width of the slit are such
that end effects can be neglected. Since the paste flows nearly as a solid plug, the velocity
profile is approximately flat over the slit cross-section, so vz = V = constant, may be
assumed. For z < 0, the fluid temperature is uniform at the inlet and equal to T1 . For z > 0,
heat is added at a uniform constant flux q0 through both the slit walls. The temperature
profile in the paste changes in both the x and z direction as it flows between the plates.
It is known, however, that when the distance L (indicated as the right boundary in the
figure) is sufficiently large, the temperature profile becomes linear in the z direction, and
T ##
the flux becomes constant and equal to q0 . Since the velocity profile is flat, viscous
z z=L
#
dissipation effects may be neglected. Further, physical properties can be assumed to be
independent of temperature. It can be shown that at steady state, the partial differential
equation that governs the temperature distribution in the narrow slit is,
! 2
2T
"
T T
Cp vz =k + 2 (4.1)
z x2 z

We are interested in developing a numerical finite difference scheme to predict the tempera-
ture in the domain B x B and 0 z L.

(a) Show that the governing equation can be written in the following dimensionless form:
! 2
2
"
1
= + (4.2)
Re Pr 2 2

V B CP
where, Re = is the Reynolds number, Pr = is the Prandtl number and,
k
x z T T1
= ; = ; =
B B q0 B/k

9
CHE 3167 (JUN 10)

(b) What class of PDEs does eqn (4.1) belong to?

(c) The boundary conditions for this problem are,

T T
B.C. 1 : k = q0 , at x = +B; B.C. 2 : k = q0 , at x = B
x x
T
B.C. 3 : T = T1 , at z = 0; B.C. 4 : k = q0 , at z = L
z
Rewrite the boundary conditions in terms of the dimensionless variables defined above.

(d) Considering Re = Pr = 1, set up the O(h2 ) finite difference equations for a numerical
solution of this problem to determine the temperature, (i) at a typical interior point
within the domain, B x B and 0 z L, (ii) at a typical point on each of the
bounding surfaces, and (iii) at the domain corners. Note that h is the step-size in the
x and z directions.

Solution:

(a) From the definition of the non-dimensional variables,

T T1 x z
= ; = ; =
q0 B/k B B

we can write the derivatives of non-dimensional variables as follows:


dT dx dz
d = ; d = ; d =
q0 B/k B B

Now from eqn (4.1) we get,

1 2 2
! " ! "
(q0 B/k) q0 B
Cp V =k 2 +
B k B 2 2
! 2
2
"
k
= +
Cp V B 2 2
"! 2
2
! "! "
k
= +
Cp V B 2 2
! 2
2
"
1
= +
Re Pr 2 2

(b) We know that for a linear second-order PDE, the sign of the expression B 2 4AC
(where A and C are coefficients of the second-order derivatives with respect to the
independent variables and B is the coefficient of the mixed second-order derivative),
determines the classification of the PDE. In this case, A = C > 0, and B = 0. As a
result, B 2 4AC < 0, which implies this PDE belongs to the class of elliptic PDEs.

10
CHE 3167 (JUN 10)

(c) In terms of the dimensionless variables, the boundary conditions are

B (q0 B/k) d d
B.C. 1 : at = = 1, k = q0 , which implies, =1
B B d d
B d
B.C. 2 : at = = 1, = 1
B d
0 T1 T1
B.C. 3 : at = = 0, = =0
B (q0 B/k)
L (q0 B/k) d d
B.C. 4 : at = , k = q0 , which implies, =1
B B d d

Figure 1: Discretization of the domain. The empty circles are interior nodes, the gray circles
represent boundary nodes, while the black circles are the corner-nodes

(d) To implement a finite-difference scheme to solve the problem, we use second-order


central-difference approximations of the partial derivatives, given by:

2 i+1,j 2i,j + i1,j


=
2
()2
2 i,j+1 2i,j + i,j1
=
2
()2
i+1,j i1,j
=
2 ()

Considering = = h, and Re = Pr = 1, the nodal equations are developed as


follows. For all internal nodes (1 < i < m + 1, 1 < j < n + 1), eqn (4.2) is valid, and
hence,
i+1,j i1,j i+1,j 2i,j + i1,j i,j+1 2i,j + i,j1
= +
2h h2 h2

11
CHE 3167 (JUN 10)

For the non-corner nodes on the left boundary ( = 0), we simply have the Dirichlet
condition
1,j = 0
The Neumann B.C. for any node 1 < i < m + 1 at the top-boundary ( = 1) can be
written with the help of the backward difference approximation as,
# $ %
## 1
= (3i,n+1 4i,n + i,n1) = 1
#=1 2h

while for nodes on the bottom boundary ( = 1), for any 1 < i < m + 1, a forward
difference leads to,
# $ %
## 1
= (3i,1 + 4i,2 i,3 ) = 1
#=1 2h

Similarly, for the nodes on the right-boundary ( = L/B), for any 1 < j < n + 1,
# $ %
## 1
= (3m+1,j 4m,j + m1,j ) = 1
#=L/B 2h

Finally, the equations for the corner nodes are obtained by assuming that the temper-
ature at any corner-node is the arithmetic average of the temperatures of the adjoining
surface nodes. Thus,
2,1 + 1,2
1,1 =
2
1,n + 2,n+1
1,n+1 =
2
m,1 + m+1,2
m+1,1 =
2
m+1,n + m,n+1
m+1,n+1 =
2

12
CHE 3167 (JUN 10)


(a) (b)

(c)

5) Two plates are 10 cm apart as shown in figure (a) above. Initially, both the plates and
the fluid are still. At time t = 0, the top plate is moved at a constant velocity V cm/s. The
equations governing the motions of the fluids are,
voil 2 voil vwater 2 vwater
= oil and = water
t x2 t x2
where x is in the direction from the bottom to the top plate, as indicated in the figure. The
following relationships hold true at the oil-water interface,

voil = vwater

and
voil vwater
oil = water
x x
(a) Using the computational molecule depicted in figure (b), develop finite difference equa-
tions for the respective governing equations in each domain.
(b) By dividing the distance between the plates into 5 intervals such that x = 2 (as shown
in figure (c)), write down,

(i) The initial conditions at each node

13
CHE 3167 (JUN 10)

(ii) The boundary conditions at the nodes corresponding to the two plates, and the
node at the fluid-fluid interface (using backward or forward differences, as appro-
priate)
(iii) The corresponding finite difference equations at each of the internal nodes in both
the domains

(c) List all the unknowns, and indicate if the number of equations equals the number of
unknowns.
(20 marks)

Solution: The equations governing the motions of the fluids are,


voil 2 voil
= oil (3)
t x2
vwater 2 vwater
= water (4)
t x2
The following relationships hold true at the oil-water interface,
voil = vwater (5)
and
voil vwater
oil = water (6)
x x
(a) Treating the time derivative to first order, and the spatial derivative to second order in
accuracy, we can construct the finite difference approximations for the derivatives:
%+1 %
vf vf,i vf,i

=
t t
%+1 %+1 %+1
2
vf vf,i+1 2vf,i + vf,i1

=
x2 (x)2
where f is the type of fluid (water or oil). Substituting the approximations in the
governing equations 3 and 4 we can get the explicit finite difference form at the ith
node at t%+1 :
. /
%+1 % %+1 %+1 %+1
voil,i voil,i voil,i+1 2voil,i + voil,i1
= oil
t (x)2
. /
%+1 % %+1 %+1 %+1
vwater,i vwater,i vwater,i+1 2vwater,i + vwater,i1
= water
t (x)2
which can be rearranged to give:
%+1 % oil t & %+1 %+1 %+1
'
voil,i = voil,i + 2
voil,i+1 2voil,i + voil,i1
(x)
%+1 % water t & %+1 %+1 %+1
'
vwater,i = vwater,i + 2
vwater,i+1 2vwater,i + vwater,i1
(x)

14
CHE 3167 (JUN 10)

Defining f = f t/(x)2 , and rearranging the equations, we get the governing finite
difference equations in the two domains as:
%+1 %+1 %+1 %
oil voil,i1 + (1 + 2oil )voil,i oil voil,i+1 = voil,i (7)
%+1 %+1 %+1 %
water vwater,i1 + (1 + 2water )vwater,i water vwater,i+1 = vwater,i (8)
Note that the quantities on the LHS are all for time vf%+1 while the RHS is in terms of
value(s) at vf% .

(b) (i) Initially, both plates and the fluid are still. The initial conditions (t = 0) required
to solve the equations 7 and 8 are:

vwater (x, 0) = 0; voil (x, 0) = 0; for all x

That is,
0 0
vwater,i = 0; voil,i = 0; for all i = 1, 2..., 6 (9)

(ii) The respective boundary conditions are:


%+1
At the bottom plate (i=1): vwater (0, t) = 0 for all t, i.e, vwater,1 =0
%+1
At the top plate (i=6): voil (10, t) = V cm/s for all t > 0, i.e, voil,6 =V
%+1 %+1
Velocity continuity at the interface (i=4): voil,4 = vwater,4
(10)
Using forward and backward differences for oil and water respectively, we have at
the interface, from the tangential stress balance,
oil & %+1 %+1 %+1
' water & %+1 %+1 %+1
'
3voil,4 + 4voil,5 voil,6 = 3vwater,4 4vwater,3 + vwater,2 (11)
2x 2x
(iii) The corresponding finite difference equations at the internal node i = 5 in the oil
domain is,
%+1 %+1 %+1 %
oil voil,i1 + (1 + 2oil )voil,i oil voil,i+1 = voil,i (12)
and for the internal nodes i = 2, 3 in the water domain,
%+1 %+1 %+1 %
water vwater,i1 + (1 + 2water )vwater,i water vwater,i+1 = vwater,i (13)

(c) At each time step there are seven unknowns. In the water domain, the unknowns are
%+1 %+1 %+1 %+1 %+1 %+1
vwater,1 , vwater,2 , vwater,3 and vwater,4 , while in the oil domain they are, voil,4 , voil,5 and
%+1
voil,6 . We have an equal number of equations, namely, three equations for interior
nodes, and four boundary conditions. Thus, we have one equation for the oil at the
interior node i = 5 (Eq. 12), two equations for the water at the interior nodes i = 2
and 3 (Eqs. 13), and four boundary conditions (Eqs. 10 and Eqs. 11). Finally, we have
the initial condition ((Eqs. 9) which enables the time integration to proceed.

15

Вам также может понравиться