You are on page 1of 213

PROBLEM SOLUTIONS

MASS & HEAT TRANSFER


ANALYSIS OF MASS AND HEAT EXCHANGERS

Cambridge University Press

CHAPTER 2 SOLUTIONS

Problem 2.1

2A + B ⎯
⎯→ nD

Use a mole balance on each component.

dC A 2 dC B 1 1 2 dC D n n 2
A: = −rA = −kC A C B B: = − rA = − kC A C B D: = rA = kC A C B
dt dt 2 2 dt 2 2

From equilibrium, CA – CAi = 2(CB – CBi)

C A − C Ai
CB = + C Bi
2

-For the condition when CAi = CBi:

C A − C Ai C + C Ai
CB = + C Ai = A
2 2

dC A 2 ⎛ C + C Ai ⎞
Therefore, = − kC A ⎜ A ⎟
dt ⎝ 2 ⎠

-For the condition when CAi ≠ CBi:

C A − C Ai
CB = + C Bi
2

dC A 2 ⎛ C − C Ai ⎞
Therefore, = −kC A ⎜ A + C Bi ⎟
dt ⎝ 2 ⎠

Experiments can be performed to find the concentration of component A as a function of time.


The above equations can be solved, and the data plotted to obtain the value of the rate constant k
for both cases. Note that if the CB(t) or CD(t) are more easily obtainable, the differential

1
equations for the other components can be used and the equilibrium relationship can be
manipulated to fit these equations.

2
Problem 2.2

Use the data given to check if r=k’CA is a good model.

dC A
= r = − k ' C A (negative since losing component A)
dt

Integration of the above equation gives:

C A (t )
ln = −k ' t
C Ai

which equals:

ln C A (t ) = −k ' t + ln C Ai

Plot ln CA vs. t. Slope will give k’ if the data are linear.

1.8

1.7

1.6
y = -0.0026x + 1.6872
R2 = 0.992
1.5
ln CA

1.4

1.3

1.2

1.1

1
0 50 100 150 200 250
Tim e (m in)

Therefore, k’=0.0026 min-1. The plot is linear, so this is a good model.

3
Problem 2.3

CB t
dCB
i.e., ∫
CB 0 K eCB + { K e (C A0 − CB 0 ) + 1} CB − CB 0
2
= − k2 ∫ dt
0
....(2)

4
a2
dx
Left hand side of (2) is of the form ∫ ax
a1
2
+ bx + c
which is equal to

1 ⎛ 2ax + b − b 2 − 4ac ⎞
ln ⎜ ⎟ if b 2 − 4ac > 0 (which is certainly the case here).
b − 4ac ⎝ 2ax + b + b − 4ac ⎟⎠
2 ⎜ 2

Integration of (2) gives,

CB
⎡ 1 ⎛ 2 K eCB + K e (C A0 − CB 0 ) + 1 − q ⎞ ⎤
⎢ ln ⎜ ⎟ ⎥ = − k2t
⎢⎣ q ⎜⎝ 2 K eCB + K e (C A0 − CB 0 ) + 1 − q ⎟⎠ ⎥⎦ CB 0

( where q = { K e (C A0 − CB 0 ) + 1} + 4 K eCB 0 )
2


1
ln ⎜
{
⎛ 2 K eCB + K e (C A0 − CB 0 ) + 1 − q } {2 K Ce B0 }
+ K e (C A0 − CB 0 ) + 1 + q ⎞
⎟ = −k t
⎝ {
q ⎜ 2 K eCB + K e (C A0 − CB 0 ) + 1 + q } {2 K Ce B0 + K e (C A0 − C B0 ) + 1 −}q ⎟

2

....(3)

Since Ke, CA0 and CB0 are known, q can be calculated and comes out to be 6.6487.
Substituting the known values of Ke, CA0, CB0 and q in (3), we get

⎛ 27.74CB − 1.0154 ⎞
0.3878 ln ⎜ 5.7429 × ⎟ = −k2t where CB should be in moles/liter.
⎝ 27.74CB + 4.1416 ⎠

⎛ 27.74CB − 1.0154 ⎞
Plotting 0.3878 ln ⎜ 5.7429 × ⎟ versus t and fitting the data to a straight line will
⎝ 27.74CB + 4.1416 ⎠
give us k2 as the negative of the slope of the fitted line. Figure below shows the plotted data and
corresponding linear fit from which the slope comes out to be
–0.0005687 min-1.

Thus, k2 = 0.0005687 min-1

liter
k1 = K e k 2 = 0.0078879
mole. min

Thus, the constitutive equation for the rate of (CH3)2CCNOH production is

5
dCD
= k1C ACB − k2CD
dt
with k1 = 0.0078879 liter/(mole.min) and k2 = 0.0005687 min-1.

Figure: Plot to obtain the reverse rate constant, k2

6
Problem 2.4

First, we write the species mole balances for A and R, respectively, for an isothermal batch
system:

dC A dCR
= − rA = −k1C A and = rA − rs = k1C A − k2CR
dt dt

The species A balance is a separable first order ODE, and can thus be solved independently for
CA(t) with the initial condition CA(t=0)= CA0:

CA
dC A'
t

∫ CA' = −k1 ∫0 dt
CA 0

⎛C ⎞
⇒ ln ⎜ A ⎟ = −k1t
⎝ C A0 ⎠
⇒ C A (t ) = C A0 e − k1t

Substituting this result into the species R balance yields

dCR
= k1C A0 e − k1t − k2CR .
dt

This first order ODE is not separable, and thus requires an integrating factor in order to solve.
We have the model equation in the general form

dx
= q (t ) − p(t ) x
dt

whose general solution is

− p ( t ) dt ⎡ ∫ p (t ) dt q(t )dt ⎤ .
x(t ) = e ∫ ⎢⎣ ∫ e
⎦⎥

Substituting the appropriate terms from the R balance yields

CR (t ) = e ∫ ⎡⎢ ∫ e ∫ k1C A0 e − k1t dt ⎤⎥
− k2 dt k2 dt

⎣ ⎦
⇒ CR (t ) = k1C A0 e ⎡⎣ ∫ e
k2t ( k2 − k1 ) t
dt ⎤⎦
k1
⇒ CR (t ) = C A0 e − k2t ⎡⎣ A − e( k2 − k1 )t ⎤⎦
k2 − k1

7
k1
⇒ CR (t ) = C A0 ⎡⎣ Ae − k2t − e − k1t ⎤⎦ .
k2 − k1

Finally, using the initial condition CR(t=0)= 0, we obtain

k1
CR (t ) = C A0 ⎡⎣ e − k2t − e − k1t ⎤⎦ .
k2 − k1

dCR
The maximum value of CR is obtained when = 0 , such that
dt t =tmax

dCR (t ) k1 d
=0= C A0 ⎡⎣e− k2t − e− k1t ⎤⎦
dt k2 − k1 dt

⇒ 0 = −k2 e − k2tmax + k1e− k1tmax


⎛k ⎞
ln ⎜ 2 ⎟
= ⎝ 1⎠ .
k
⇒ tmax
(k2 − k1 )

Now plug this value into the equation for CR to find CR(tmax).

k1 ⎡ ⎛ − k ln (k 2 / k1 ) ⎞ ⎛ − k ln (k 2 / k1 ) ⎞⎤
C R (t max ) = C A0 ⎢exp⎜⎜ 2 ⎟⎟ − exp⎜⎜ 1 ⎟⎟⎥
k 2 − k1 ⎣ ⎝ (k 2 − k1 ) ⎠ ⎝ (k 2 − k 1 ) ⎠⎦

8
Problem 2.5

Extension of Problem 2.4…

First, we write the species mole balances for A and R, respectively, for an isothermal batch
system:

dC A dCR
= rA = −k1C A and = −rA − rs = k1C A − k2CR
dt dt

The species A balance is a separable first order ODE, and can thus be solved independently for
CA(t) with the initial condition CA(t=0)= CA0:

CA
dC A'
t

∫ CA' = −k1 ∫0 dt
CA 0

⎛C ⎞
⇒ ln ⎜ A ⎟ = −k1t
⎝ C A0 ⎠
⇒ C A (t ) = C A0 e− k1t

Substituting this result into the species R balance yields

dCR
= k1C A0 e − k1t − k2CR .
dt

With the solution as outlined in 2.4,

k1
CR (t ) = C A0 ⎡⎣ e − k2t − e− k1t ⎤⎦ .
k2 − k1

a) We are given reference values of the rate constants at 25°C. Therefore, we can cast the rate
constants in terms of the reference values by dividing the Arrhenius expression at any
temperature by that for the reference temperature,

Ea

ki (T ) k e RT ⎡ E ⎛1 1 ⎞⎤
= i ,0 Ea ⇒ ki (T ) = ki (Tref ) exp ⎢ − a ⎜ −
ki (Tref ) −
⎢ R ⎜ T T ⎟⎟ ⎥⎥
RT ⎣ ⎝ ref ⎠ ⎦
ki ,0 e ref

where here the given reference temperature is 298.15K. Now, CR(t), tmax, and CR,max can be
computed using the rate constants calculated from the above expression (given here in SI units):

9
b)
T (°C) k1 (s-1) k2 (s-1) tmax (s) CR,max (M)
0 3.28x10-4 1.05x10-4 5103 0.585
25 3.33x10-3 5.00x10-3 243 0.296
50 2.36x10-2 1.31x10-1 16.0 0.124
100 5.41x10-1 2.41x101 0.161 0.0205
200 3.88x101 2.99x104 2.22 x10-4 0.0012

Alternatively, we could have solved the species balances by numerical integration, yielding the
same results. However, numerical integration does not yield exact values for tmax and CR,max.
For completeness, the numerical solution to CR(t) at 25°C is plotted below, where the line
represents the analytical solution and the X’s represent the solution obtained using Matlab’s
ode15s numerical integration function. The solutions at other temperatures are qualitatively
similar, just with different time scales.

10
Problem 2.6

(a) Batch Reactor:

dCDBS
Rate of DBS decomposition, = −kCDBS which on integration gives
dt

1 CDBS (0)
CDBS (t ) = CDBS (0) e − kt ⇒ t= ln ....(1)
k CDBS (t )

Rate constant k is given in the problem to be 0.031 min-1 and the initial DBS concentration
CDBS(0) is 5 g/liter.

For CDBS(t) = 1.5 g/liter, we can find t from (1).

1 5
t= ln = 38.8 minutes
0.031 1.5

(b) CFSTR:

A = DBS

For a CFSTR, overall mass balance is

dV
= q AF − q A
dt

dV
At steady state, = 0 , so q AF = q A
dt

Species mass balance for A is


qAF, CAF qA, CA
d (VC A )
= q AF C AF − qC A − rV
dt

d (VC A )
Again, assuming steady state, = 0 . So
dt

q A (C AF − C A ) q A (C AF − C A )
V= =
r kC A

Using k = 0.031 min-1, CAF = 5 g/liter, CA = 1.5 g/liter and qA = 50 liter/min, we get

11
V = 3763 liters

This should be rounded to 3770 L.

(c) Residence time, τ = V/q = V/qA = 3770/50 = 75.4 minutes

12
Problem 2.7

For the first case, we have two tanks in a series where the effluent from Tank 1 is the input for
Tank 2. We wish to reduce the concentration of DBS from 5g/L to 1.5 g/L by the reaction given
in Problem 2.6.

qA, CAF qA, CA1 qA, CA2

Species mass balance for A in Tank 1 is

d (V1C A1 )
= q A (C AF − C A1 ) − kC A1V1
dt

Assuming steady-state, we can solve for CA1.

q A C AF
C A1 =
q A + kV1

Species mass balance for A in Tank 2 is

d (V2 C A 2 )
= q A (C A1 − C A 2 ) − kC A 2V2
dt

Assume steady-state and plug in the equation for CA1. Let V1=V2=V.

⎛ q C ⎞
0 = q A ⎜ A AF − C A 2 ⎟ − kC A 2V
⎜ q A + kV ⎟
⎝ ⎠

Solve for V.

V =
qA ( C AF − C A 2 ) = 50L / min( 5 g / L − 1.5 g / L )
k C A2 .031 min −1 1.5 g / L

V ~ 1340 L each

13
We want to find the exit concentration of DBS that can be achieved if we operate these two tanks
in parallel (as shown below).

qA/2, CA1

qA, CAF

qA/2, CA2

Since V1=V2, and the flow rates and feed concentrations are the same in each reactor, the exit
concentrations in each reactor will also be the same. Therefore, we need to do a mass balance
around only one of the reactors to find the exit concentration.

Species mass balance for A in Tank 1 is

d (V1C A1 )
= q A (C AF − C A1 ) − kC A1V1
dt

Assuming steady-state, we can solve for CA1.

q A C AF
C A1 =
q A + kV1

Remember that the flow rate in each reactor is half of the total available (qA=25 L/min).

25L / min(5 g / L)
C A1 = C A 2 = = 1.9 g/L
25L / min + 0.031 min −1 (1332 L)

14
Problem 2.8

q (51, 000 ft 3 /s)


First, let us assume plug flow for the river, such that v = = = 17.83 ft/s .
A (220 ft) × (13 ft)

We write the usual balance expression for plug flow kinetics, which reads

dC X − kC X
v = −rX − =
dz K + CX

which upon rearrangement yields

( K + C X ) dC k
= − dz .
X
CX v

We now integrate the first order separable ODE with the initial condition CX(z=0)=CX0 such that

CX CX z
dC X k
K ∫C C X C∫ dC X = − v ∫0 dz .
+
X0 X0

⎛C ⎞ kz
⇒ K ln ⎜ X ⎟ + C X − C X 0 = −
⎝ CX 0 ⎠ v

Kv ⎛ C X 0 ⎞ v
⇒ z= ln ⎜ ⎟ + (CX 0 − CX ) .
k ⎝ CX ⎠ k

We are given the following: K = 8.3 x 10-2 M, k = 7.8 x 10-5 M.s-1, CX0 = 0.01 M and CX =
0.0001 M. Given v = 17.83 ft/s from above, and substituting these values into the preceding
result gives z = 89636.7 ft = 16.98 miles. So, the fish should be safe to eat at a distance of
more than 17 miles downstream from the landfill.

15
Problem 2.9

A→ D

We need a continuous reactor here leaving us two choices, CFSTR and tubular reactor.
1
A plot of versus CA can help us determine which reactor would be smaller in volume for the
rA
1
same conversion. If is a decreasing function of CA, a tubular reactor will have a lower
rA
volume than a CFSTR for a given flow rate (see figure 2.11 of the text). In this problem, rA =
1
kCA which means that is a decreasing function of CA. Thus, a tubular reactor will have a
rA
lower volume for a given conversion and flow rate. We will verify this by first designing a
tubular reactor and comparing its volume to that of a CFSTR.

Tubular reactor:

CA0, q CA,CD, q

dC A
ν = − kC A ....(1)
dz
dC D
ν = kC A ....(2)
dz

⎛ kL ⎞
Solving (1), we get C A ( L) = C A0 exp⎜ − ⎟ ....(3)
⎝ ν ⎠

where L is the length of the reactor and CA0 is the inlet concentration of A.

Let us assume a conversion of X = 80 %.

So, C A ( L) = (1 − X )C A0 = 0.2CA0

CD(L) = (CA0 – CA(L)) since 1 mole of A gives one mole of D.

As CA0 = 0.2 M, we get CD(L) = 0.16 M

Also, qCD(L) = 50 moles/min, which gives

q = 312.5 liters/min (the total volumetric flow rate)

16
V L
Now, residence time Θ = =
q v

From (3), we get Θ = 321.9 min.

V = q Θ ≅ 100,590 liters

CFSTR:

CA0, q CA,CD, q

Assuming steady state, the equations for a CFSTR would be

q(C A0 − C A ) − kC AV = 0 ....(4)
q(C D 0 − C D ) + kC AV = 0 ....(5)

Taking CD0 = 0 and solving (5), we get

qC D = kC AV ....(6)

Again we assume a conversion of 80 %.

So, CA = 0.2CA0. Also, CA0 = 0.2 M, k = 0.005 min-1 and qCD = 50 moles/minute. Using these
values in (6), we get

V = 250,000 liters

CD = (CA0 – CA) = 0.16 M

So, q = 50/CD = 312.5 liters/min

Residence time, θ = V/q = 800 min.

Thus we observe that the volume and residence time for a CFSTR are much larger than that for a
tubular reactor for a fixed conversion. Here we assumed a conversion of 80 % but for any value
of conversion, the volume of the tubular reactor would be smaller due to the nature of the rate
expression. A plot of reactor volume versus conversion can be constructed to study the variation
of the relative sizes of the CFSTR and the tubular reactor.

17
Problem 2.10

2A → D

We need a continuous reactor here leaving us two choices, CFSTR and tubular reactor.
1
A plot of versus CA can help us determine which reactor would be smaller in volume for the
rA
1
same conversion. If is a decreasing function of CA, a tubular reactor will have a lower
rA
volume than a CFSTR for a given flow rate (see figure 2.11 of the text). In this problem, rA =
1
kCA2 which means that is a decreasing function of CA. Thus, a tubular reactor will have a
rA
lower volume for a given conversion and flow rate. We will verify this by first designing a
tubular reactor and comparing its volume to that of a CFSTR.

Tubular reactor:

CA0, q CA,CD, q

dC A 2
v = −2kC A ....(1)
dz
dC 2
v D = kC A ....(2)
dz

−1
⎛ L −1 ⎞
Solving (1), we get C A ( L) = ⎜ 2k + C A0 ⎟ ....(3)
⎝ v ⎠

where L is the length of the reactor and CA0 is the inlet concentration of A.

Let us assume a conversion of X = 80 %.

So, C A ( L) = (1 − X )C A0 = 0.2CA0

CD(L) = (CA0 – CA(L))/2 since 2 moles of A give one mole of D.

As CA0 = 0.3 M, we get CD(L) = 0.12 M

Also, qCD(L) = 50 moles/min, which gives

q = 416.67 liters/min (the total volumetric flow rate)

18
V L
Now, residence time Θ = =
q v

From (3), we get Θ = 2222.22 min.

V = q Θ ≅ 9.26 x 105 liters

CFSTR:

CA0, q CA,CD, q

Assuming steady state, the equations for a CFSTR would be

q ( C A0 − C A ) − 2kC A V = 0
2
....(4)
q ( CD 0 − CD ) + kC A V = 0
2
....(5)

Taking CD0 = 0 and solving (5), we get

2
qCD = kC A V ....(6)

Again we assume a conversion of 80 %.

So, CA = 0.2CA0. Also, CA0 = 0.3 M, k = 0.003 (M.min)-1 and qCD = 50 moles/minute. Using
these values in (6), we get

V = 4.63 x 106 liters

CD = (CA0 – CA)/2 = 0.12 M

So, q = 50/CD = 416.67 liters/min

Residence time, θ = V/q = 11112 min.

Thus we observe that the volume and residence time for a CFSTR are much larger than that for a
tubular reactor for a fixed conversion. Here we assumed a conversion of 80 % but for any value
of conversion, the volume of the tubular reactor would be smaller due to the nature of the rate
expression. A plot of reactor volume versus conversion can be constructed to study the variation
of the relative sizes of the CFSTR and the tubular reactor.

19
Problem 2.11

20
21
Problem 2.12

22
k = 0.075 min-1 was obtained from 2.11

23
CHAPTER 3 SOLUTIONS

Problem 3.1

Let us use subscript 1 for the process fluid (water) and 2 for the utility fluid (DOWTHERMTM).

So,
V1 = 1.4 m3, ρ1 = 1000 kg/m3, Cˆ p1 = 4.186 kJ/(kg.K), T1i = 20 ˚C, T1 = 60˚C
V2 = 1.4 m3, ρ2 = 1000 kg/m3, Cˆ = 3.6 kJ/(kg.K), T2i = 120 ˚C
p2

(a)

QLoad = ρ1V1Cp1 |T1-T1i| = 1000 kg/m3 x 1.4 m3 x 4.186 kJ/kg-K x |60-20| K = 2.34 x 105 kJ

(b) At equilibrium, the two fluids will have the same temperature (which we denote by T∞).

So T1,eq = T2,eq = T∞

ρ1V1Cˆ p1T1i + ρ 2V2Cˆ p 2T2i


T∞ = (equation 3.1.5 from text)
ρ1V1Cˆ p1 + ρ 2V2Cˆ p 2

Substituting the values, we get T∞ = 66.2 ˚C.


(Note that we have performed a level II analysis to obtain the result for T∞.)

Since the equilibrium temperature is greater than T1i ( = 60 ˚C), it is possible to carry out
this operation (of heating the process fluid from 20 to 60 ˚C). If the heat transfer
coefficient and the heat transfer area are known, the time it would take to carry out this
operation can be calculated (see problem 3.3).

(c) We need to find T2, final temperature of the DOWTHERM. Using level I analysis
(conservation of energy), we get

ρ1V1Cˆ p1 (T1 − T1i ) = − ρ 2V2 Cˆ p 2 (T2 − T2i ) (equation 3.1.3 from text).
ρ1V1Cˆ p1
⇒ T2 = (T1i − T1 ) + T2i
ρ 2V2 Cˆ p 2

Substituting the values, we obtain T2 = 73.5 ˚C.


Problem 3.2

Level III analysis of the system gives

dT1
ρ1V1Cˆ p1 = Qcoil
dt

Integrating this ODE from t=0 where T1=T1i, we obtain

ρ1V1Cˆ p1 (T1 − T1i ) = Qcoil t

We are given the following:

V1 = 1.4 m3, ρ1 = 1000 kg/m3, Cp1 = 4.186 kJ/(kg.K), T1i = 20 ˚C, T1 = 60˚C and Qcoil =
30 kW.

ρ1V1Cˆ p1 (T1 − T1i )


So, t = = 7813.867 sec. = 130.2 min.
Qcoil
Problem 3.3

By level I analysis, we obtain

ρ1V1Cˆ p1 (T1 − T1i ) = − ρ 2V2 Cˆ p 2 (T2 − T2i ) (equation 3.1.3)


ρ1V1Cˆ p1 (T1i − T1 )
⇒ T2 = T2i +
ρ 2V2Cˆ p 2

By level III analysis, assuming the usual constitutive relation for Q,


dT
ρ1V1Cˆ p1 1 = −Ua (T1 − T2 ) (equation 3.2.9)
dt

Substituting the expression for T2 above into 3.2.9 and rearranging, we get

dT1 Ua
=− ⎡⎣(1 + Θ ) T1 − (T1i + ΘT2i ) ⎤⎦
dt ρ 2V2 Cˆ p 2
ρ 2V2Cˆ p 2
where Θ = .
ρ1V1Cˆ p1

Note that our treatment here is the same as that in section 3.2.1 and that the model equation
above for T2 is the same as equation 3.2.12. Solving this equation, we write the expression for T1
as (equation 3.2.14)

⎛ T + ΘT2i ⎞ ⎛ Θ ⎞
⎟ [T1i − T2i ] e
−βt
T1 (t ) = ⎜ 1i ⎟+⎜
⎝ 1+ Θ ⎠ ⎝ 1+ Θ ⎠

⎛ Ua Ua ⎞
where β = ⎜ + ⎟
⎜ ρ V Cˆ ˆ ⎟
⎝ 1 1 p1 ρ 2V2 C p 2 ⎠


(1 + Θ ) T1 (t ) − (T1i + ΘT2i ) = e− β t
Θ [T1i − T2i ]

1 ⎛ (1 + Θ ) T1 (t ) − (T1i + ΘT2i ) ⎞
⇒ t=− ln ⎜⎜ ⎟⎟
β ⎝ Θ [T1i − T2i ] ⎠

We are given the following:


V1 = 1.4 m3, ρ1 = 1000 kg/m3, Cˆ p1 = 4.186 kJ/(kg.K), T1i = 20 ˚C, T1(t) = 60˚C
V2 = 1.4 m3, ρ2 = 1000 kg/m3, Cˆ = 3.6 kJ/(kg.K), T2i = 120 ˚C,
p2
2
a = 4.55 m
ρ 2V2Cˆ p 2
So, Θ = = 0.86
ρ1V1Cˆ p1

Using this information, we obtain the following results for various values of the heat transfer
coefficient:

U (W/m2K) β (s-1) t (min)


-4
100 1.679 x 10 198.8
500 8.396 x 10-4 39.8
1000 1.679 x 10-3 19.9
Problem 3.4

Here we have a semi-batch heat exchanger with mixed-mixed fluid motion.

By level III analysis (equations 3.3.6 and 3.3.7) we obtain for the process and utility fluids,
respectively:
dT
ρ1V1Cˆ p1 1 = −Ua [T1 − T2 ]
dt
dT
ρ 2V2 Cˆ p 2 2 = ρ 2 q2 Cˆ p 2 (T2 F − T2 ) + Ua [T1 − T2 ] .
dt

The equations above represent a set of coupled ODE’s in T1 and T2. We will solve them
here by numerical integration in a numerical software package (Matlab for example). Since we
do not have an explicit expression that can be solved for q2, we will solve the problem by picking
values of q2, then evaluating the equations, and analyzing the result to see if the required heat
transfer is obtained. This is iterated until appropriate q2 is found.
Semi-Batch Reactor
120
T1
110 T2

100

90

80
Temp (C)

70

60

50

40

30
q= 0.0017 m3/s

20
0 100 200 300 400 500 600 700 800 900
Time (s)

When U = 1000 W/m2K, we can get the required process heating (T1=60oC) for high
values of q2. By guessing values of q2, we obtain a value of q2=1.7x10-3 m3/s=1.7 L/s
required to heat the process stream to 60˚C in 15 minutes.

The T2 value at the exit corresponding to this flow value is T2 = 94.3oC.


Problem 3.5

We are given the following:


q1 = 100 L/min, ρ1 = 1000 kg/m3, Cˆ p1 = 4.184 kJ/(kg-˚C), T1F = 20 ˚C, T1 = 60˚C
ρ2 = 1000 kg/m3, Cˆ = 3.6 kJ/(kg-˚C), T2F = 120 ˚C
p2

^
(a) QLoad = ρ1 q1 C p1 (T1 − T1F )

QLoad = 1000 kg/m3(100 L/min)(0.001 m3/L)(1 min/60s)(4.184 kJ/kg-K)(60-20 K)

QLoad = 278.9 kJ/s

(b) The minimum utility fluid flow rate is found where T1 = T2=T∞. Using a Level I
analysis, we obtain:

^ ^
ρ1 q1 C p1 (T1F − T1 ) = − ρ 2 q 2 C p 2 (T2 F − T2 )

Therefore,

^
ρ1 q1 C p1 (T1F − T∞ )
q 2,min = ^
ρ 2 C p 2 (T∞ − T2 F )

1000kg / m 3 (100 L / min)(4.184kJ / kg − o C )(20 o C − 60 o C )


q 2,min = = 77.5 L/min
1000kg / m 3 (3.6kJ / kg − o C )(60 o C − 120 o C )

(c) This operation cannot be attained in practice as it will take an infinitely long residence time in
each compartment to reach thermal equilibrium between the two streams. A higher utility fluid
flow rate and lower T2 is needed.

A technically feasible analysis is a utility flow of 100 L/min.

From Level I analysis,

^
ρ1 q1 C p1 (T1F − T1 )
T2 = T2 F + ^
ρ 2 q2 C p 2

At a flow rate of 100 L/min,


1000kg / m 3 (100 L / min)(4.184kJ / kg − o C )(20 o C − 60 o C ))
T2 = 120 o C + = 73.5°C
1000kg / m 3 (100 L / min)(3.6kJ / kg − o C )

We see that this value seems to work by testing against the 10° rule (T2 > T1 + 10°).

(d) By Level III analysis,

d ⎛ ^
⎞ ^
⎜ ρ1V1 C p1 T1 ⎟ = ρ1 q1 C p1 (T1F − T1 ) − Ua(T1 − T2 )
dt ⎝ ⎠

We can assume steady-state operation for this process. Thus,

^
ρ1 q1 C p1 (T1F − T1 )
U=
a(T1 − T2 )

From Figure 3.7, we know the dimensions of the exchanger (D = 1m; h = 2m).

a = πDh =π(1m)(2m) = 6.28 m2

Therefore,

⎛ 0.1m 3 ⎞
1000kg / m 3 ⎜⎜ ⎟⎟(4.184kJ / kg − K )(20 o C − 60 o C )
U= ⎝ 60 s ⎠ = 3.29 kJ/(m2K)
6.28m (60 C − 73.51 C )
2 o o

*Note that there are a number of utility flow rate/exit temperature/heat transfer coefficients that
could work.
Problem 3.6
CV1

Q
T2 = constant = 35°C

T1, Cp1, V1 = 0.8 L, ρ1

Assumptions:
• At time, t = 0, the temperature in the well-mixed batch system is T1i.
• The water bath is held constant at a temperature of 35˚C.
• The properties of the bottle fluid can be considered those of water:
ρ1 = 1000 kg/m3, Cˆ p1 = 4.186 kJ/(kg.K)
• Heat transfer only occurs between the bath and the bottle.
• Area = 3.8 x 10-2 m2

Level I analysis

Expressing the word statement for control volume defined as the batch system:

H1 t +Δt = H1 t − QΔt
H1 = H1i − QΔt

The enthalpy can be expressed as H1 = ρ1Cˆ p1V1T1 so the word statement (Level I analysis)
becomes

ρ1Cˆ p1V1 (T1 − T1i ) = −QΔt

Dividing by Δt and taking the limit as Δt approaches zero yields the differential energy balance

dT1
ρ1Cˆ p1V1 = −Q .
dt

Level II analysis

At this level we consider the equilibrium limitations (i.e., what the final temperature of the well-
mixed batch system will be at large time, t). Here, realizing that the batch system exchanges heat
only with the constant-temperature bath, it is obvious that at tinf, T1=T2.
Level III analysis

At Level III, we would like to solve for T1, but this requires a constitutive relation to link Q to
the system temperatures. Namely, we propose that

Q = Ua (T1 − T2 )
such that
dT1
ρ1Cˆ p1V1 = −Ua (T1 − T2 ) .
dt

Since T2 is constant, this relation is separable and can be solved as


T1 t
dT Ua
∫T (T1 − 1T2 ) = −∫0 ρ Cˆ V dτ
1i 1 p1 1

⎛ T −T ⎞ Uat
ln ⎜ 1 2 ⎟ = −
⎝ T1i − T2 ⎠ ρ1Cˆ p1V1
Uat
or ln (T2 − T1 ) = − + ln (T2 − T1i ) .
ρ1Cˆ p1V1

We see from the above relation that the temperature of the fluid in the well-mixed batch system
is dependent only upon the fluid properties of fluid 1 when T2 is constant. Also, we note that the
solution is of the linear form y=mx+b (NOTE: we have intentionally left the last term on the
right hand side because T1i is not known and must be fit by the data). Thus a plot of the
experimental data in the form ln (T2 − T1 ) vs. t should yield a straight line with a slope, m =
Ua
, and a y-intercept, b = ln (T2 − T1i ) .
ρ1Cˆ p1V1

40 3.5
35 3
30 2.5
ln (T2 - T1)

25
2
T1 (C)

20
1.5 y = -0.0051x + 3.269
15 R2 = 0.9882
1
10
5 0.5

0 0
0 100 200 300 400 500 0 100 200 300 400 500
t (s) t (s)

By fitting the data, we can obtain values for the unknowns U and T1i. The appropriate plot of the
data with the best fit linear relationship is shown (right graph), yielding a slope of m=-0.0051 s-1
and a y-intercept of b=3.269 (NOTE: A plot of T1 vs. t is also shown on the left). Using these
values to compute the unknown parameters we obtain U=450 W/m2K and T1i=8.7˚C.

Problem 3.7

CV1

Q
T2 = constant

T1, Cp1, V1, ρ1


At time, t = 0, the temperature in the well-mixed batch system is T1i.

Here, we assume heat transfer only between the constant temperature water bath and the batch
system.

(a)
Level I analysis
Expressing the word statement for control volume defined as the batch system:
H1 t +Δt = H1 t − QΔt
H1 = H1i − QΔt
The enthalpy can be expressed as H1 = ρ1Cˆ p1V1T1 so the word statement (Level I analysis)
becomes
ρ1Cˆ p1V1 (T1 − T1i ) = −QΔt
Level II analysis
At this level we consider the equilibrium limitations (i.e., what the final temperature of the well-
mixed batch system will be at large time, t). Here, realizing that the batch system exchanges heat
only with the constant-temperature bath, it is obvious that at tinf, T1=T2.

Level III analysis


From the Level I analysis, we can divide by Δt and take the limit as Δt approaches zero to arrive
at the differential energy balance
dT
ρ1Cˆ p1V1 1 = −Q
dt
At Level III, we would like to solve for T1, but this requires a constitutive relation to link Q to
the system temperatures. Namely, we propose that
Q = Ua (T1 − T2 )
such that
dT1
ρ1Cˆ p1V1 = −Ua (T1 − T2 )
dt
Since T2 is constant, this relation is separable and can be solved as
T1 t
dT1 Ua
∫T (T1 − T2 ) ∫0 ρ Cˆ V dτ
= −
1i 1 p1 1

⎛ T −T ⎞ Uat
ln ⎜ 1 2 ⎟=− ˆ
⎝ T1i − T2 ⎠ ρ1C p1V1
⎧⎪ Uat ⎫⎪
T1 = T2 + (T1i − T2 ) exp ⎨− ⎬
ˆ
⎪⎩ ρ1C p1V1 ⎭⎪

We see from this equation that the temperature of the fluid in the well-mixed batch system is
dependent only upon the fluid properties of fluid 1 when T2 is constant.

(b) The average rate of heat transfer (heat load) from time t=0 to t is given as
t
1
Q = ∫ Q (τ )dτ
t0
From the equation derived from Level III analysis (boxed above), we can write Q as

⎧⎪ ⎧⎪ Uat ⎫⎪ ⎪⎫
Q = Ua (T1 − T2 ) = Ua ⎨T2 + (T1i − T2 ) exp ⎨− −
⎬ 2⎬T
ˆ
⎪⎩ ⎩⎪ ρ1C p1V1 ⎭⎪ ⎪⎭
Therefore,
1
t ⎧⎪ ⎧⎪ Uaτ ⎫⎪⎪⎫
⎨( 1i 2)
t ∫0
Q= Ua T − T exp ⎨− ˆ ⎬⎬dτ
⎪⎩ ρ
⎩⎪ 1 p1 1 ⎭⎪⎪⎭
C V
⎧ Uaτ ⎫
(T1i − T2 ) ∫ exp ⎪⎨− ˆ ⎪⎬dτ
t
Ua
Q=
t 0 ⎪⎩ ρ1C p1V1 ⎭⎪
ρ1Cˆ p1V1 ⎡ − Uat ⎤
(T1i − T2 ) ⎢e ρ1C p1V1 − 1⎥
ˆ
Q=
t ⎢ ⎥
⎣ ⎦
Problem 3.8

T1, Cp1, V1, ρ1

Qout
T2 = constant
Qheat

Level I analysis

Choosing the house as our control volume, we express the word statement for the total energy
balance:

H1 t +Δt = H1 t − Qout Δt + Qheat Δt

Substituting H1 = ρ1Cˆ p1V1T1 , dividing by Δt and taking the limit as Δt approaches zero yields the
differential energy balance
dT1
ρ1Cˆ p1V1 = −Qout + Qheat .
dt

In order to keep the house comfortable, we wish to operate the heating system in the house such
that T1 is constant, and thus

=0
dT
ρ1Cˆ p1V1 1 = −Qout + Qheat = 0
dt
⇒ Qheat = Qout

Level III analysis

In order to quantify the heat duty, and thus the cost, required to keep the house at the desired
temperature, we assume the usual constitutive relation for the heat transferred from the house to
the outside air,

Qheat = Qout = Ua (T1 − T2 ) .


12 windows -> each window area = 1.1 m x 1.3 m

Thus, the heat duty required to heat the house is

Qheat= 10 [W/m2K] x (12 x 1.1 m x 1.3 m) x (20˚C-4˚C) = 2.75 kW


and the cost to heat the house for 90 days (taking into account the 42% efficiency of the heating
system) will be

⎛ 1 kW supplied ⎞ ⎛ 24 hr ⎞ ⎛ $0.036 ⎞
2.745 kW × ⎜ ⎟ × ( 90 days ) × ⎜ ⎟×⎜ ⎟ = $508
⎝ 0.42 kW heat ⎠ ⎝ day ⎠ ⎝ kW ⋅ hr ⎠
Problem 3.9
Problem 3.10
⎛Q⎞
⎜ ⎟ = 3 kW/m
2

⎝ ⎠
a

m& 1 = 0.01 kg/s

T1F = 20 °C

⎛Q⎞
⎜ ⎟ = 3 kW/m
2

⎝a⎠
We are given the following:

ρ1 = 1000 kg/m3 , m& 1 = ρ1q1 = 0.01 kg/s ⇒ q1 = 10−5 m3 /s, Cˆ p1 = 4184 J/(kgK), T1F = 20 °C
Q
Flux = = 3000 W/m 2 , Tube diameter, Dtube = 0.075 m, U = 48 W/(m 2 K )
a

a.) Level I Analysis

We first derive the differential equation for T1 by applying conservation of energy on a


differential element of the tube assuming steady state. [NOTE: We have also implicitly
assumed a uniform temperature along the radial position of the tube.]
z z + Δz

Δz

Q
0 = ρ1q1Cˆ p1T ( z ) − ρ1q1Cˆ p1T ( z + Δz ) +
Δa
a
where Δa is the differential area through which heat is being transferred to the fluid in the
differential volume element and is given by π Dtube Δz .
ρ1q1Cˆ p1T ( z + Δz ) − ρ1q1Cˆ p1T ( z ) Q
So, = π Dtube
Δz a

Taking the limit as Δz → 0 , we obtain

dT Q
ρ1q1Cˆ p1 = π Dtube (assuming ρ1 , q1 and Cˆ p1 are independent of axial position)
dz a

Solving the above ODE, we obtain

π Dtube ⎛ Q ⎞
T1 ( z ) = T1F + ⎜ ⎟z
ρ1q1Cˆ p1 ⎝ a ⎠

ρ1q1Cˆ p1 ⎛ Q ⎞−1
z= (T1 ( z ) − T1F )
π Dtube ⎜⎝ a ⎟⎠

For T1(z) = 80 ˚C, z = 3.55 m.

b.) We can simply plot the expression for T1(z) derived above using the information given:
200

150
T (°C)

100
80°C

50 (z)
T1

0
0 2 4 6 8 10

z (m)
Problem 3.11

T2F = 120 ˚C,


q2 = 800 liters/hr.

T1F = 25 ˚C,
500 kg/hr
m& 1 = 1000 kg/hr.
2 - Utility fluid T1 =70°C
25 ˚C
1 - Styrene (process fluid)

T2

Data given:

T1F = 25 ˚C, T1 = 70 ˚C, m& 1 = ρ1q1 = 500 kg/hr = 0.1389 kg/sec., ρ1 = 906 kg/m3,
Cˆ = 1750 J/(kg.K)
p1

T2F = 120 ˚C, q2 = 800 liters/hr = 2.22x10-4 m3/sec.,


ρ = 1000 kg/m3, Cˆ = 3600 J/(kg.K), U = 230 W/(m2K), DTube = 1” = 0.0254 m.
2 p2

a.) We begin by computing the heat load for the process stream:

QLoad = ρ1q1Cˆ p1 T1 − T1F = 10937.5 W = 10.937 kW

From a Level III balance on the exchanger, we also have (equation 3.4.10):

QLoad
QLoad = UaΔTlm ⇒ a =
U ΔTlm

To find the log-mean temperature difference, ΔTlm , we need to know T2 (the exit
temperature of the utility stream). From level I analysis, we have

1 ρ 2 q2Cˆ p 2
T2 = T2 F − [T1 − T1F ] where Θ = = 3.29
Θ ρ1q1Cˆ p1

So, T2 = 106.32 ˚C

⎡(T2 − T1 ) − (T2 F − T1F ) ⎤⎦


⇒ ΔTlm = ⎣ = 61.03 ˚C
⎡ (T2 − T1 ) ⎤
ln ⎢ ⎥
⎣ (T2 F − T1F ) ⎦
QLoad
Thus, a = = 0.779 m 2
U ΔTlm
Since a = π DTube L , we obtain
L = a/πDtube = 9.8 m

b.) We now consider the case where the flow rate is doubled, so m& 1 = ρ1q1 = 1000 kg/hr =
0.278 kg/sec.

new
QLoad = ρ1q1Cˆ p1 T1 − T1F = 21875 W = 21.875 kW

ρ 2 q2Cˆ p 2
Θ new = = 1.643
ρ1q1Cˆ p1

1
T2new = T2 F − [T1 − T1F ] = 92.61 ˚C
Θ

⎡(T2new − T1 ) − (T2 F − T1F ) ⎤


ΔT new
=⎣ ⎦ = 50.43 ˚C
⎡ (T2new − T1 ) ⎤
lm

ln ⎢ ⎥
⎢⎣ (T2 F − T1F ) ⎥⎦

Since the heat transfer coefficient is a function of the flow in the exchanger, we must
consider the effects of doubling the flow rate on U. As U ∝ v 0.8 , we have (assuming plug flow
of course)

0.8
⎛ π Dtube 2 ⎞
⎜⎜ qnew ⎟⎟
4
=⎝ ⎠ = 20.8
U new
0.8
U old ⎛ π Dtube 2 ⎞
⎜⎜ qold ⎟⎟
⎝ 4 ⎠

⇒ U new = 400.5 W/(m 2 K)

Recalculating the required heat transfer area, we obtain:

new
QLoad 21875
anew = = m 2 = 1.083 m 2
U new ΔTlmnew
400.45 × 50.43

=> Lnew = anew/πDtube = 13.6 m


c.) The greatest attainable heat transfer occurs in the limit as q2 → ∞ . In this limit,
T2 → T2F . In this limiting case, we get

⎡(T2 F − T1 ) − (T2 F − T1F ) ⎤⎦


ΔTlmmax = ⎣ = 70.11 ˚C.
⎡ (T2 F − T1 ) ⎤
ln ⎢ ⎥
⎣ (T2 F − T1F ) ⎦

Q max = U new aΔTlmmax = 21871.017 W = 21.873 kW

But before, we calculated QLoad = 21.875 kW. Therefore, since Q max ≈ QLoad as q2 → ∞ ,
this operation most likely can not be performed at any utility flow rate, since it is
right on the limit (and as engineers, we never want to operate on the limit).
Problem 3.12

a.) First we determine our mass and energy balance equations for this system:

dC M
Mass balance: = rM = −2k p C Ii (1 − exp(− k d t ))C M (t )
dt

^ dT
Energy balance: ρV C p = (− ΔH Rxn )rM V + Q
dt

Since we are working with an isothermal system, the left side of the energy balance equation is
0.

Combination of the mass and energy balance equations gives:

Q(t ) = ΔH Rxn rM V = ΔH Rxn [−2k p C Ii (1 − exp(−k d t ))C M (t )]V

We next want to find the change in the concentration of monomer with time. We do this by
integrating the mass balance equation.

Solving the ODE with the initial condition CM(t=0) = CMi, we obtain:

⎛ C (t ) ⎞ kp
ln ⎜ M ⎟ = −2 CIi ⎡⎣ kd t + e − kd t − 1⎤⎦
⎝ CMi ⎠ kd

Rearranging this expression allows us to determine CM(t).

⎛ kp ⎞
CM (t ) = CMi exp ⎜ −2 CIi ⎡⎣ kd t + e − kd t − 1⎤⎦ ⎟
⎝ kd ⎠

1.00
0.90
0.80
0.70
0.60
CM (M)

0.50
0.40
0.30
0.20
0.10
0.00
0 200 400 600 800 1000
Time (s)
By plugging this into the rate expression, we obtain:

⎛ kp ⎞
− rM = 2k p C1i CMi ⎡⎣1 − exp ( −kd t ) ⎤⎦ exp ⎜ −2 CIi ⎡⎣ kd t + e − kd t − 1⎤⎦ ⎟
⎝ kd ⎠

and thus

⎛ kp ⎞
Qload = −2ΔH rxn k p C1i CMi ⎡⎣1 − exp ( − kd t ) ⎤⎦ exp ⎜ −2 CIi ⎡⎣ kd t + e − kd t − 1⎤⎦ ⎟ V
⎝ kd ⎠

We are given the following


ΔH rxn = −73 kJ/mol, k p = 1.87 ×102 (M s)-1 , kd = 8.45 ×10−6 s -1 , CIi = 0.01M, CMi = 1M ,
V = 1L

The heat load is plotted below using the above parameters.


Q (W)

t (s)

b.) To design the reactor jacket, we write the energy balance on the cooling jacket, assuming
steady state operation. [NOTE: since Qload changes with time in this case, the rate of
change of enthalpy of the cooling water must be negligible compared to the rate of heat
generation].
=0

= ρ 2 q2 (t )Cˆ p 2 (T2 f − T2 (t ) ) + Qload (t )


dT2
ρ 2V2Cˆ p 2
dt
Here, we have recognized that since Q varies with time, we must adjust the flow rate of
cooling water with time. We also know that the heat load must also equal the heat
transferred from the cooling water stream to the reactor fluid, and thus

Qload (t ) = Ua (T1 − T2 (t ) )

which upon rearrangement gives

Qload (t )
T2 (t ) = T1 − .
Ua

Substituting this constraint into the energy balance yields

⎛ Q (t ) ⎞
0 = ρ 2 q2 (t )Cˆ p 2 ⎜ T2 f − T1 + load ⎟ + Qload (t )
⎝ Ua ⎠

Solving this equation for q2(t) gives the design equation for the cooling jacket

Qload (t )
q2 (t ) = .
⎛ Q (t ) ⎞
ρ 2Cˆ p 2 ⎜ T1 − T2 f − load ⎟
⎝ Ua ⎠

All of the parameters in the design equation have been specified with the exception of the
heat transfer area. For a tank volume of 1 L, a jacket width on the order of centimeters is
reasonable. If we assume the worst surface area to volume ratio possible (a cube), this
corresponds to a heat transfer area on the order of 0.1 m2. Using this value for the area and the
thermal properties of the cooling water [ ρ 2 = 1000 kg/m3 , Cˆ p1 = 4184 J/(kgK) ], the flow rate of
cooling water is as plotted below (red curve).

We could also compute the minimum flow rate of cooling water required by taking the
limit as
a → ∞:

Qload (t )
q2,min (t ) = .
ρ 2Cˆ p 2 (T1 − T2 f )

For comparison, q2,min(t) is plotted as well (blue curve). It is important to note that
although we assumed pseudo-steady state operation for the cooling jacket, this is clearly not
the case as q2 is small and varies greatly with time. Thus, the simplifying assumption made
in the energy balance may introduce significant error in the analysis. However, it allows for
drastic simplification of the model equations in order to give an order of magnitude estimate and
limiting behavior for the required design.
q2 (m3/s)

t (s)
Problem 3.13
Problem 3.14
glass wall

Chamber 1 Q Chamber 2
1

V1 = 4 L V2 = 10 L
Q2
T1i = 20°C T2i = 70°C

Area for the glass wall = 2.4 cm x 0.25 m2

(a) System equilibrium temperature


From the word statement, a Level I analysis can be performed accounting for the enthalpy of the
system (heat is exchanged only between the adjacent chambers):
( H1 + H 2 ) t +Δt = ( H1 + H 2 ) t
Since H = ρ Cˆ pVT , then
ρ1Cˆ p1V1 (T1 − T1i ) = ρ 2Cˆ p 2V2 (T2i − T2 )
At equilibrium, T1 = T2 = T∞ so that
ρ1Cˆ p1V1 (T∞ − T1i ) = ρ 2Cˆ p 2V2 (T2i − T∞ )
and
ρ1Cˆ p1V1T1i + ρ 2Cˆ p 2V2T2i
T∞ =
ρ1Cˆ p1V1 + ρ 2Cˆ p 2V2
For liquid water in both control volumes ρ1 = ρ 2 ; Cˆ p1 = Cˆ p 2 . Therefore, the equilibrium
temperature is just the volume weighted average temperature
V1T1i + V2T2i ( 4 L ) ( 20 C ) + (10 L ) ( 70 C )
o o

T∞ = = = 55.7o C
V1 + V2 ( 4 L ) + (10 L )
(b) The energy balances for this system are

dH1 dH
= Q1 ; 2 = Q2
dt dt
V1 is heated so Q1 is positive.
Since Q1=-Q2
dH1
=Q
dt
dH 2
= −Q
dt

Then
dT1
ρ1V1Cˆ p1 = −Ua (T1 − T2 )
dt
dT
ρ 2V2 Cˆ p 2 2 = Ua (T1 − T2 )
dt

(c) From the Level I analysis, we can solve for T2 in terms of T∞ and T1 as
⎛ 1⎞ 1
T2 = ⎜1 + ⎟ T∞ − T1 ,
⎝ Θ⎠ Θ
where
ρ 2V2Cˆ p 2
Θ=
ρ V Cˆ 1 1 p1

and use this relation to decouple the energy balance equations determined in part b.
dT1 Ua Ua ⎛ ⎛ 1⎞ 1 ⎞
=− ( T1 − T2 ) = − ⎜ T1 − ⎜1 + ⎟ T∞ + T1 ⎟
dt ρ1V1Cˆ p1 ρ1V1Cˆ p1 ⎝ ⎝ Θ⎠ Θ ⎠
dT1 Ua ⎛ 1⎞
=−
ˆ ⎜1 + ⎟ (T1 − T∞ )
dt ρ1V1C p1 ⎝ Θ ⎠
Which is separable
T1
Ua ⎛ 1⎞
t
dT1
∫T (T1 − T∞ ) ∫0 ρ V Cˆ ⎜⎝1 + Θ ⎟⎠ dt
= −
1i 1 1 p1

⎛ T −T ⎞ ⎡ Ua ⎛ 1 ⎞⎤
ln ⎜ 1 ∞ ⎟ = − ⎢ ⎜ 1 + ⎟⎥ t
⎝ T1i − T∞ ⎠ ⎢⎣ ρ1V1Cˆ p1 ⎝ Θ ⎠ ⎥⎦

⎛ T −T ⎞
Therefore, plotting ln ⎜ 1 ∞ ⎟ vs. t will result in a linear plot with a slope of
⎝ T1i − T∞ ⎠
⎡ Ua ⎛ 1 ⎞⎤
slope = − ⎢ ⎜ 1 + ⎟⎥
⎢⎣ ρ1V1Cˆ p1 ⎝ Θ ⎠ ⎥⎦

The slope can be used to extract the heat transfer coefficient.


0 100 200 300 400 500 600
0

-0.05

-0.1
y = -6.57E-04x
2
-0.15 R = 1.00E+00

-0.2

-0.25

-0.3

-0.35

The slope is –6.57 x 10-4 s-1.

For the case of liquid water in both reservoirs,


ρ 2V2Cˆ p 2 V2 10 L
Θ= = = = 2.5
ρ1V1Cˆ p1 V1 4 L
⎡ ⎤

−6.57 x10−4 s −1 =−

⎢ (
1.4U 0.25m 2 ⎥
⎥)
⎢ ⎛ ⎞ ⎥
( 3
)( 3
⎢ 1000kg / m 0.004m ⎜ 4190
J
)
⎟⎥
⎣⎢ ⎝ kgK ⎠ ⎦⎥
W
U = 31.4 2
m K

Note that this is an ESTIMATED value since we only have two data points, and there exists an
uncertainty in this data that accompanies all results of this type.
Problem 3.15
41.3 °C

10 °C
Problem 3.16

a.) The sketch of the fermenter should resemble the diagram shown in Figure 15.9-1 of the
provided chapter of Sandler, 4th ed., but with a flow-through cooling jacket placed outside of the
tank. The diagram should contain a more physically accurate depiction of the tank, with a
stirring mechanism (impeller or otherwise) and a gas bubbler, as well as an inlet for the glucose
feed.

b.) The heat of combustion can be calculated in exactly the same fashion as in Illustration
15.9-6, of Sandler, 4th ed. Thus, the heat of combustion can be computed as mol product x heat
of combustion/product – mol reactant x heat of combustion/reactant:
0.235 mol biomass ×519.0 kJ/C-mol + 0.451 mol ethanol×684.5 kJ/C-mol − 1 mol glucose x
467.8 kJ/C-mol − 0.0399 mol ammonia×348.1 kJ/C-mol = −51.0 kJ for each mol of glucose
consumed.

c.) In order to be used to calculate the heat load for the fermenter, the glucose feed
concentration must be converted to moles. We thus obtain

50 g/L glucose = 50/(6*12+6)*6 =3.8 C-moles/L glucose.

The rate of ethanol production is calculated from the residence time (4 hours) and feed
glucose concentration is then given by:

1400[ L]
r = 0.451× × 3.8 C-mole/L = 0.17 C-mol/s = 10C-mol/min ethanol.
4 × 3600[ s ]

A Level II analysis is required to determine the minimum flow rate of cooling water. The
Qload on the reactor is given by

Qload = ΔH c ( N& ethanol ) = −51 [kJ] ×10 [C-moles/min] × (1 [min] / 60[sec]) = −8.5 kW

The required cooling water is given by a Level 1 balance as:

Qload = q2 ρ 2Cˆ P 2 (T2 F − T2 )


−8500 W = q2 × 1000[kg / m3 ] × 4184 [J/kg 0 C] × (5 − 25) [ 0 C]
∴ q2 = 1.02 ×10−4 m3 /s = 0.1 L/s = 6 L/min

d.) A level III analysis, assuming a 5 oC temperature difference as a driving force yields:
0 = q2 ρ 2Cˆ P 2 (T2 F − T2 ) + Ua (T1 − T2 )
−Ua (T1 − T2 )
∴ q2 =
ρ 2Cˆ P 2 (T2 F − T2 )
−309 [W/m 2 K] × 4.55 [m 2 ] × 5 [K]
=
1000 [kg/m3 ] × 4184 [kJ/kg K] × (5 − 20) [K]
= 1.12 x10−4 m3 /s = 6.72 L/min

(NOTE: A technically feasible design using temperature differences of 5-10 oC is reasonable.)


Problem 3.17

a.) For a continuous-flow coiled heat exchanger, we consider mixed-plug fluid


motions.
T2F = 5°C, q2

T1F=
80°C,
q1= 100
L/min
T1 = 50°C, q1

T2, q2

First, we perform a Level I analysis on the system that gives the same analysis as all continuous
flow systems:

ρ1Cˆ p1q1 (T1F − T1 ) = ρ 2Cˆ p 2 q2 (T2 − T2 F )

which we can simplify for the water-water system to

q1 (T1F − T1 ) = q2 (T2 − T2 F )

In this configuration, although we are able to return utility water at a temperature less than 60oC,
we must consider the design. In doing so, we recognize that the maximum temperature for the
utility return is governed by the outlet temperature of the tank, T1=50oC. Therefore, the cooling
water cannot be any greater than T1=50oC.

The heat load for the system can be calculated from the process stream properties

Qload = ρ1q1Cˆ p1 [T1 − T1F ]


⎛ kg ⎞ ⎛ L ⎞ ⎛ m3 min ⎞ ⎛ kJ ⎞
Qload = ⎜1000 3 ⎟ ⎜100 ⎟ ⎜ 4.184 ⎟ ⎡⎣50 C − 80 C ⎤⎦
o o
⎟⎜
⎝ m ⎠⎝ min ⎠⎝ 1000 L *60 s ⎠⎝ kgK ⎠
Qload = −209.2kW

We can calculate the minimum utility flow rate required to meet the cooling objectives by
considering the limit when the outlet temperatures approach one another, T1=T2=Tinf

q1 (T1F − Tinf ) = q2 (Tinf − T2 F )


q1 (T1F − Tinf ) (100 L / min ) (80o C − 50o C )
q2 = =
(Tinf − T2 F ) ( 50 C − 5 C )
o o

q2 = 66.7 L / min

We said previously that T2 must be less than 50oC, so we can use the 10oC rule of thumb and say
that a feasible design would be for T2=40oC. Then

q1 (T1F − T1 ) (100 L / min ) (80o C − 50o C )


q2 = =
(T2 − T2 F ) ( 40 C − 5 C )
o o

q2 = 85.7 L / min

We can determine the area required for heat transfer by considering the log mean temperature
difference form of the Level III analysis, where

Qload = Ua
(T2 − T1 ) − (T2 F − T1 )
⎡ (T − T ) ⎤
ln ⎢ 2 1 ⎥
⎣ (T2 F − T1 ) ⎦

−209.2kW = (1kW / m K ) a
( 40 C − 50 C ) − ( 5 C − 50 C )
2
o o
1
o o

⎡ ( 40 C − 50 C ) ⎤ o o
1
ln ⎢ ⎥
⎢⎣ ( 5 C − 50 C ) ⎥⎦
o o

a = 8.99 m 2

In order for this design to be feasible, we must size the pipe in order to ensure turbulent (plug)
flow.

4ρ q
d max = .
N π Rec μmax

For the water flow through the process piping,

⎛ m3 min ⎞
( 3
)
4 1000kg / m ( 85.7 L / min ) ⎜ ⎟
1000 L *60s ⎠
d max = ⎝
⎛ kg ⎞
π (10, 000 ) ⎜1422 x10−6 ⎟
⎝ ms ⎠
d max = 0.13 m = 5.0"
Therefore, we could use 5” nominal diameter, Schedule 80S steel pipe having an I.D. of 4.813”
and an O.D. of 5.563”. With this O.D., we can calculate the length of 5” nominal diameter pipe
required for the area of 8.99m2 for heat transfer, where a = π DL . Therefore,

a 8.99m 2
L= =
πD ⎛ 1 ft ⎞ ⎛ 1m ⎞
π ( 5.563in ) ⎜ ⎟⎜ ⎟
⎝ 12in ⎠ ⎝ 3.28 ft ⎠
L = 20.25 m = 66.4 ft
Given the length of pipe required, we must determine the tank size. If we assume a liquid height
of 2 m (leaving some head space in a tank of 7 feet in height), we can determine the tank
diameter based upon the pipe being coiled vertically around the circumference of the tank. For
piping of 5” nominal diameter, a reasonable turn radius is 7.5” (Perry’s, Table 10-27).
π D π ( 7.5in )
Therefore, the length of pipe within each of the turns is L = = = 11.78" and each
2 2
straight section of pipe is 1.62 m = 5.31 ft (based on the 2m total liquid height).

T2F, q2

T1F, q1

2m

T2, q2

T1, q1 15”

The length of pipe in the pipe section shown above is

⎛ 11.78in ⎞
Lsec tion = 2 ⎜ ⎟ + 2 ( 5.31 ft ) = 12.58 ft
⎝ 12in / ft ⎠

Therefore, the number of pipe turns (defined as in the diagram as two turns and two straight
sections) is given by

66.4 ft
No.Turns = = 5.3 ~ 6turns
12.58 ft

With the given turn radius, the width of each turn (defined above) is 30”. Therefore, we require
6*(30”) = 180” circumference for the coil, which translates into
(πD=180”) a coil diameter of 57.3” = 4.77ft = 1.46m. If we add 2” of clearance between the
coil and the inside tank wall, the diameter of the tank is given 61.3”=5.1 ft=1.56m.
Therefore, to summarize
T2F, q2 Process feed T1F=80oC q1=100L/min
o
Process out T1=50 C q1=100L/min
T1F, q1 Utility feed T2F=5oC q2=85.7L/min
Utility out T2F=40oC q2=85.7L/min
DTank=1.56 m
HTank=7 ft = 2.13 m
No. of coil turns = 6
T2, q2
DCoil=1.46 m
Dpipe=5” (nominal) (4.813” I.D. and 5.563” O.D.).

T1, q1

b) With the utility flow rate above, we can calculate the outlet temperature of the utility
stream for each of the four continuous flow heat exchangers. For any continuous
flow exchanger, the Level I balance is

q1 (T1F − T1 ) = q2 (T2 − T2 F )

Therefore, the outlet utility temperature must be

100 L / min
T2 = T2 F +
q1
q2
(T1F − T1 ) = 5o C + (
85.7 L / min
)
80o C − 50o C = 40o C

This appears to be fine for all exchanger designs in that it is less than both the process inlet
and outlet temperatures. Thus, we can calculate the area required for heat exchange in each
of the cases:

Case 1: Tank mixed-mixed:

Qload = ha (T2 − T1 )
Qload 1 −209.2kW 1
⇒ amixmix = = = 20.9m 2
(
U (T2 − T1 ) 1kW / m K 40 C − 50o C1
2 o
)
Case 2: Tank mixed-plug was determined in part (a) to be amixed-plug = 8.99m2.

Case 3: Tubular (co-current)


Qload = ha
(T2 − T1 ) − (T2 F − T1F )
⎡ (T − T ) ⎤
ln ⎢ 2 1 ⎥
⎣ (T2 F − T1F ) ⎦
⎡ (T2 − T1 ) ⎤
ln ⎢ o
(
⎡ 40o C − 50o C1 ⎤

)
ln ⎢
acocurrent
Q
= load
( − ) ⎥
⎣ 2 F 1F ⎦ = −209.2kW
T T ⎢
⎣ 5 C − 80 o
(C ⎥⎦ )
= 6.5m 2
2 o
( o o
) (
h (T2 − T1 ) − (T2 F − T1F ) 1kW / m K 40 C − 50 C1 − 5 C − 80 C o
)
Case 4: Plug-plug (counter-current)

Qload = ha
(T2 − T1F ) − (T2 F − T1 )
⎡ (T − T ) ⎤
ln ⎢ 2 1F ⎥
⎣ (T2 F − T1 ) ⎦
⎡ (T2 − T1F ) ⎤
ln ⎢ o
(
⎡ 40o C − 80o C ⎤

)
ln ⎢
a=
Qload ( T −
⎣ 2F 1 ⎦ T ) ⎥
=
− 209.2 kW ⎢
⎣ 5 C − 50 o
C (
⎥⎦
= 4.9m 2
)
2 o o
(
h (T2 − T1F ) − (T2 F − T1 ) 1kW / m K 40 C − 80 C − 5 C − 50 C
o o
) ( )
Exchanger type T2 [oC] area [m2]
Mixed-mixed 40 20.9
Mixed-plug 40 8.99
Plug-plug (co-current) 40 6.5
Plug-plug (counter-current) 40 4.9

From this summary, we can see that the efficiency is greatest for the counter-current plug-plug
case, requiring the least area for heat transfer. This results from the higher average temperature
difference along the length of the heat exchanger in this geometry. The least efficient design is
the mixed-mixed case where, the driving force remains lower and constant between the two
mixed vessels.
Problem 3.18

Utility in T2F q2
Process Inlet TpF
80 oC, 100 liters/min

Recycle to Exchanger
T1 q1 T1F q1

Well Mixed Tank


1000 liters
Heat Exchanger
Tp Tank Outlet Pump PROCESS Stream Outlet Tp
qp 50 oC, 100 liters/min

valve
Utility out T2 q2

We first construct a stream table to summarize the known stream properties:

Stream Vol. flow (L/min) Temperature (oC)


Process inlet qpF = 100 L/min TpF = 80oC
Tank outlet qp = ? Tp = 50oC
Recycle feed to
q1 = ? T1F = 50oC
exch.
Recycle exch.
q1 = ? T1 = ?
effluent
Utility in q2= ? T2F = 5oC
Utility out q2= ? T2 < 60oC
Process outlet qpo = 100 L/min Tp = 50oC

Performing a Level I analysis on the complete process (including the tank and recycle loop), we
obtain the familiar equation:

ρ p Cˆ p1q p (TpF − Tp ) = ρ 2Cˆ p 2 q2 (T2 − T2 F )

Under counter-current operation of the heat exchanger, we can afford to return the utility stream
at a temperature of no more than 60oC as shown in the table above. If we run a counter-current
heat exchanger, however, with an inlet process stream of 50oC, then the utility fluid in this case
can only be returned at a temperature less than 50oC. We apply the rule of thumb of maintaining
a 10oC temperature difference, such that the outlet utility stream will be at a temperature of 40oC.
Furthermore, because of the water-water system, the Level I analysis to determine the utility
flow rate, q2, simplifies to

q2 =
q p (TpF − Tp )
= 100 L / min
(80 C − 50 C )
o o

(T2 − T2 F ) ( 40 C − 5 C )
o o
q2 = 85.7 L / min

Performing a Level I analysis around the tank, and simplifying for the water-water system, we
obtain:

ρ pF Cˆ pp q pTpF + ρ1Cˆ p1q1T1 = ρ p Cˆ pp q pTp

⇒ q pF TpF + q1T1 = q pTp

Similarly, we can perform a Level I analysis around the recycle split, and its simplification for
the water-water system:

ρ p Cˆ pp q pTp = ρ1Cˆ p1q1T1F + ρ p Cˆ pp q poTp

⇒ q pTp = q1T1F + q poTp

The Level I analysis around the heat exchanger, and its simplification for the water-water system
is:

ρ1Cˆ p1q1 (T1F − T1 ) = ρ 2Cˆ p 2 q2 (T2 − T2 F )

⇒ q1 (T1F − T1 ) = q2 (T2 − T2 F )

Note that these balances are not all independent.

For a technically feasible design, we can assume that the heat exchanger allows for the cooling
of the recycle stream to a temperature of 10oC above the inlet utility temperature. Therefore, we
assume that T1 = 15oC. Therefore, recognizing that the volumetric flow balance on the tank is
given by

q pF + q1 = q p

and that the Level I energy balance on the tank is given as

q pTpF + q1T1 = q pTp ,

we solve explicitly for the flow rate of the recycle stream:

q pF TpF + q1T1 = ( q pF + q1 ) Tp
q pF TpF − q pF Tp
⇒ q1 =
(T p − T1 )
(100 L / min ) (80o C − 50o C )
q1 =
( 50 C − 15 C )
o o

q1 = 85.7 L / min
The tank outlet process flow rate is then given by

q p = q pF + q1 = 100 L / min + 85.7 L / min


q p = 185.7 L / min

This is the is the recycle flow rate required in order to cool the tank contents continuously.
Knowing the flow rate of the recycle stream, we can calculate the area for heat transfer required
in the double-pipe heat exchanger. The total load on the heat exchanger is given as

Qload = ρ1q1Cˆ p1 [T1 − T1F ]


⎛ kg ⎞ ⎛ L ⎞ ⎛ m3 min ⎞ ⎛ kJ ⎞ o
Qload = ⎜1000 3 ⎟ ⎜ 85.7 ⎟ ⎜ 4.184 ⎟ ⎡⎣15 C − 50 C ⎤⎦
o
⎟⎜
⎝ m ⎠⎝ min ⎠ ⎝ 1000 L *60 s ⎠ ⎝ kgK ⎠
Qload = −209.2kW

With this heat load, we can then calculate the area for heat transfer according to the log-mean
temperature difference formalism for the counter-current exchanger with the ratio of thermal
capacities, Θ, of unity.

Qload = ha (T2 F − T1 )
−209.2kW = ( 0.5kW / m 2 K ) a ( 5o C − 15o C )
a = 41.84m2

Therefore, we must calculate the maximum diameter of the tube side able to enforce turbulent
flow under the flow conditions in order to determine the pipe used.

4ρ q
d max =
N π Rec μmax

For the water flow through the process piping,

⎛ m3 min ⎞
( )
4 1000kg / m3 ( 85.7 L / min ) ⎜ ⎟
1000 L *60s ⎠
d max = ⎝
⎛ kg ⎞
π (10, 000 ) ⎜1080 x10−6 ⎟
⎝ ms ⎠
d max = 0.168m = 6.63inches

Therefore, we could use a 6” nominal Schedule 10S pipe (6.357” I.D. and 6.625” O.D.).
With this O.D., we can calculate the length of 6” nominal diameter pipe required for the area of
41.84m2 for heat transfer, where a = π DL .

a 41.84m2
L= =
πD ⎛ 1 ft ⎞ ⎛ 1m ⎞
π ( 6.625in ) ⎜ ⎟⎜ ⎟
⎝ 12in ⎠ ⎝ 3.28 ft ⎠
L = 79.12 m = 259.5ft

To deal with this large length (~13x20ft sections of pipe), we would want to consider an
exchanger geometry with a multiple-pass tube side.

We now construct a stream table with all of the flow and temperature conditions summarized for
each stream of the process flow diagram:

Stream Vol. flow (L/min) Temperature (oC)


Process inlet qpF = 100 TpF = 80
Tank outlet qp = 185.7 Tp = 50
Recycle feed to
q1 = 85.7 T1F = 50
exch.
Recycle exch.
q1 = 85.7 T1 = 15
effluent
Utility in q2= 85.7 T2F = 5
Utility out q2= 85.7 T2 = 40
Process outlet qpo = 100 Tp = 50

In general, the advantage of this design comes from the more efficient heat transfer achievable
through the external counter-current heat exchanger than could be obtained with a coil within the
tank or a jacket on the tank (i.e., the latter two options would require larger areas for heat transfer
because of the lower average driving force for heat transfer). The pumping required for the
recycle, however, would have to be addressed in assessing the operating costs and could be a
limiting factor in addition to the expense of the external counter-current heat exchanger in
comparison to a jacket or cooling coil.
Problem 3.19

a) The counter-current heat exchanger is the best for this situation because a higher outlet
temperature is permitted than in a co-current heat exchanger. To accomplish this cooling
in a co-current system, a very high contact area and utility flow rate would be required as
T2 couldn’t go any higher than 20°C even at infinite area.

b)
5°C

T1F = 120 °C 20°C


q1 = 100 L/min
U = 1000 W/m2K
ρ = 800 kg/m3
Cp1 = 2 kJ/kgK 100°C

We can calculate the necessary flow rate simply by performing a Level I analysis and
recognizing that for a countercurrent heat exchanger, the utility stream outlet temperature can
cross over the process stream outlet temperature.

^ ^
ρ1 q1 C p1 (T1 − T1F ) = ρ 2 q 2 C p 2 (T2 F − T2 )
m3 kg L kJ m3 kg kJ
( )(800 3 )(100 )(2 )(20 − 120C ) = ( )(1000 3 )q 2 (4.186 )(100 − 5C )
1000 L m min kgK 1000 L m kgK
L
q 2 = 40.2
min

To calculate the area and size of the exchanger, we can look at the technically feasible design
procedure found in pages 94-99 of the text. We need to start by calculating Qload:
^
Qload = ρ1 q1 C p1 T1 − T1F
min m3 kg L kJ
Qload = ( )( )(800 3 )(100 )(2 ) 20 − 120C
60 s 1000 L m min kgK
Qload = 266.7kW

Next we can use the log mean temperature difference to calculate the contact area needed.
(T2 F −T 1) − (T2 − T1F )
ΔT21 =
T −T 1
ln( 2 F )
T2 − T1F
(5 − 20) − (100 − 120)
ΔT21 =
5 − 20
ln( )
100 − 120
ΔT21 = 17.4C
Qload = ha ΔT21
1000W
266.7 kW ( )
Qload kW
a= =
h ΔT21 W
(1000 2 )(17.4 K )
mK
a = 15.3m 2

We must now size the exchanger. To ensure turbulent flow, we need to calculate dmax, the
maximum allowable diameter that maintains turbulent flow. To do this we can use the following
equation for the process stream tube size. We can use a maximum viscosity of 5 centipoise =
0.005 kg/ms for oil at 20°C.

4 ρ1 q1
d max =
π Re c μ max
kg L min m3
4(800 )(100 )( )( )
d max = m3 min 60 s 1000 L
kg
π (5000)(0.005 )
ms

d max = 0.067m = 2.67inches

So from the pipe sizing table we choose a pipe. We choose 2.5 inch 10S pipe with an inner
diameter of 2.635 inches=0.067m. Finally we can use the following equation to calculate the
length of pipe required to achieve our necessary contact area.

a 15.3m 2
L= =
πd max π (0.067m)

L=72.7m of pipe

We now need to calculate the diameter of the outer pipe to complete the equipment sizing. Since
we are referring to an annular section of outer pipe, we need to use the following equations to
calculate the maximum hydraulic diameter. We can use a maximum viscosity of 0.98 centipoise
= 0.00098 kg/ms for water at 5°C.
ρ 2 q2 (4 Ac ) 4 ρ 2 q2
Re = =
Pwet Ac μ max μ max (πd o + πd i )
kg L min m3
4(10003
)( 40 . 2 )( )( )
5000 = m min 60 s 1000 L
kg
(0.00098 )(πd o + π (.067 + 2(.003))m)
ms

d 0 = 0.101m = 3.97inches

So we size it to 5S pipe, 3.5 inches with an inner diameter of 3.834 inches.

So the single heat exchanger will have length 72.7m=238.5 feet, inner pipe of 2.5 inch 10S
pipe, outer pipe 3.5 inch 5S pipe.

c.)

To solve for the area required for two heat exchangers of equal size, we start by performing a
level I analysis on the whole system.

^ ^
ρ 1 q1 C p1 (T1b − T1F ) = ρ 2 q 2 C p 2 (T2 F − T2 )
m3 kg L kJ m3 kg kJ
( )(800 3 )(100 )(2 )(20 − 120C ) = ( )(1000 3 )q 2 (4.186 )(100 − 5C )
1000 L m min kgK 1000 L m kgK
L
q 2 = 40.2
min
It is clear that we will obtain the same value for q2 from the overall Level I analysis. Therefore,
we will perform a Level III analysis around the stream splitter, the stream mixer, and the two
heat exchangers.

Balance around stream splitter:

Since T2 aF = T2bF = T2 F ,

q 2T2 F = q 2 a T2 aF + q 2bT2 aF
q 2 = q2 a + q 2 b (1)

Balance around stream mixer:

q 2T2 = q 2 a T2 a + q 2bT2b (2)

Balance around heat exchanger A:


^ ^
ρ1 q1 C p1 (T1a − T1F ) = ρ 2 q 2 a C p 2 (T2 F − T2 a ) (3)

Balance around heat exchanger B:

^ ^
ρ1 q1 C p1 (T1b − T1a ) = ρ 2 q 2b C p 2 (T2 F − T2b ) (4)

Unknowns: q2a, q2b, T2a, T2b, T1a

At this point, we have 4 equations and 5 unknowns. The only other information we have is that
the areas of the two exchangers must be the same:

Qload − a = Ua ΔT21− a
Qload −b = Ua ΔT21−b
Qload −a Ua ΔT21− a ΔT21− a
= =
Qload −b Ua ΔT21−b ΔT21−b

Similar to our definitions above, we can define the log mean temperature differences for the two
exchangers as:
(T2 F −T 1a ) − (T2 a − T1F )
ΔT21− a =
T −T 1a
ln( 2 F )
T2 a − T1F
(T2 F −T 1b) − (T2b − T1a )
ΔT21−b =
T −T 1b
ln( 2 F )
T2b − T1a

So the ratio becomes:


T2 F −T 1b
ln( )
Qload − a (T2 F −T 1a ) − (T2 a − T1F ) T2b − T1a
=
Qload −b T −T 1a (T2 F −T 1b) − (T2b − T1a )
ln( 2 F )
T2 a − T1F
T2 F −T 1b
ln( )
T1a − T1F (T2 F −T 1a ) − (T2 a − T1F ) T2 a − T1a
= (5)
T1b − T1a T −T 1a (T2 F −T 1b) − (T2 a − T1a )
ln( 2 F )
T2 a − T1F

We can use Mathcad or another program to solve these 5 equations for the 5 unknowns.
However, the problem is not solvable for the conditions given. We would need to decrease T2 to
50°C for the solution to converge. This is because the temperature gradient is so much greater in
heat exchanger A than in heat exchanger B that the majority of the cooling and therefore the
majority of the q2 stream must go to the first exchanger. It must return at a temperature higher
than 100°C in order to have a final temperature of 100°C after mixing with stream q2b.

d.) To plot the temperature profile for part b as a function of distance down the tube, we need
use the model equations for counter-current flow. We can now easily calculate Θ and β and use
these in the model equations (3.4.13 and 3.4.14)

kg L kJ
^ (1000 )(40.2 )(4.186 )
ρ 2 q2 C p 2 m 3
min kgK
Θ= ^
= = 1.05
kg L kJ
ρ1 q1 C p1 (800 3 )(100 )(2 )
m min kgK
ha 1 1
β= ( − )
L ρ1C p1q1 ρ 2 C p 2 q 2
kJ 60 s
(1 )(15.3m 2 )( )
2
m sK min ( 1 1
β= 3
− )
m kg L kJ kg L kJ
(72.7 m)( ) (800 3 )(100 )(2 ) (1000 3 )(40.2 )(4.186 )
1000 L m min kgK m min kgK
β = 0.0039m −1
Now we can use the model equations, plugging in our values:

ΘT2 F − e − βL T1F Θ
T1 ( z ) = ( − βL
)+ − βL
(T1F − T2 F )e − βz
Θ−e Θ−e
− βL
ΘT − e T1F 1
T2 ( z ) = ( 2 F − βL
)+ − βL
(T1F − T2 F )e − βz
Θ−e Θ−e
140
120
Temperature (degC)

100
80 T1
60 T2

40
20
0
0 20 40 60
length (m)

Since we cannot solve for the temperature profiles for the heat exchangers in series, we cannot
include another graph. We also cannot conclude whether more or less contact area would be
required in using the two heat exchangers over just using a single heat exchanger.
Problem 3.20

a.) T2<40oC

t1=T1F=80oC t2=T1=50oC
q1 = 100 L/min q1 = 100 L/min

T1=T2F=5oC

In counter-current operation, the minimum flow rate, q2, will correspond to the maximum
possible outlet temperature on the shell side. With the current specifications for the return
cooling water, this temperature is T2 = 40oC. Therefore, the minimum utility flow rate is given
by the Level I analysis for the water-water system:

ρ1q1Cˆ p1 (T1F − T1 ) = ρ 2 q2Cˆ p 2 (T2 − T2 F )

q2 = q1
( T1F − T1 )
= 100 L / min
( 80o C − 50o C )
(T2 − T2 F ) ( 40o C − 5o C )
q2,min = 85.7 L / min

Calculating the heat load from the properties of stream one, we obtain
⎛ kg ⎞ ⎛ L ⎞ ⎛ m3 ⎞ ⎛ min ⎞ ⎛ kJ ⎞
Qload = ρ1q1Cˆ p1 [T1 − T1F ] = ⎜1000 3 ⎟ ⎜ 100 ⎟⎜ ⎟⎜ ⎟ ⎜ 4.184 ⎟ ⎡⎣50 C − 80 C ⎤⎦
o o

⎝ m ⎠⎝ min ⎠⎝ 1000 L ⎠⎝ 60 s ⎠⎝ kgK ⎠


Qload = −209.3kW

Therefore, since

Q = Ua
(T2 F − T1 ) − (T2 − T1F )
⎡ (T − T ) ⎤
ln ⎢ 2 F 1 ⎥
⎣ (T2 − T1F ) ⎦

for a counter-current exchanger, the minimum area required is


⎡ (T2 F − T1 ) ⎤
ln ⎢
(
⎡ 5o C − 50o C ⎤

)
ln ⎢
amin =
Q ( T −
⎣ 2 1F ⎦ T ) ⎥
=
−209.3 × 10 W
3 ⎢
⎣ 40 o
C (
− 80 o
C ⎥
⎦ )
2 o
( o o
) (
U (T2 F − T1 ) − (T2 − T1F ) 1000 W/m K 5 C − 50 C − 40 C − 80o C )
amin = 4.93m 2

b.) Since we are still considering counter-current flow in this case, the minimum utility flow rate
for this case is the same as in part (a) since it is calculated based on the Level I analysis.
Therefore,
q2,min = 85.7 L / min

Now we must consider


⎛ ⎞
⎜ ⎟
⎜ (T2 F − T1 ) − (T2 − T1F ) ⎟
Q = ha ⎜ ⎟F .
⎡ (T2 F − T1 ) ⎤
⎜ ln ⎢ ⎥ ⎟

⎝ ⎣ ( T2 − T1 F ) ⎦

We see that with the same inlet and outlet temperatures, and the same heat load, Q=-209.3kW,
the area in this case can be related to the area calculated in part (a) as

a (a)
a correlation = ,
F

where F is the correction factor for the multiple tube heat exchanger. To determine F, one must
calculate the dimensionless temperature differences

T1 − T2 t −t
R= and S = 2 1 .
t2 − t1 T1 − t1

In terms of the standard nomenclature used in the text, these relations are

T2 F − T2 5o C − 40o C
R= = = 1.17
T1 − T1F 50o C − 80o C
50o C − 80o C
S= = 0.4
5o C − 80o C

From the FT vs. S curves in the provided figure, we determine that F = 0.95 . Therefore, the
minimum area for the multiple tube (single pass) design is

4.93m 2
a correlation = = 5.19m 2
0.95

c.) T1=T2F=5oC

t1=T1F=80oC
q1 = 100 L/min

t2=T1=50oC
q1 = 100 L/min

T2=40oC
The Level I analysis is the same as for the cases in part (a) and (b). Therefore, the minimum
flow rate in this case is still q2,min = 85.7 L / min , and Q = −209.3kW .

The minimum area for heat transfer, however, will be different. For the coiled-mixed system,

Q = ha
(T2 − T1 ) − (T2 F − T1 )
⎡ (T − T ) ⎤
ln ⎢ 2 1 ⎥
⎣ (T2 F − T1 ) ⎦

Therefore, the minimum area for the mixed-plug case is

⎡ (T2 − T1 ) ⎤ (
ln ⎢ o
)
⎡ 40o C − 50o C ⎤

ln ⎢
a=
Q ( T − ) ⎥
⎣ 2 F 1 ⎦ = −209.3x10 W
T 3 ⎢
⎣ (
5 C − 50 o
C)⎥⎦
( ) (
h (T2 − T1 ) − (T2 F − T1 ) 1000W / m 2 K 40o C − 50o C − 5o C − 50o C )
amin = 8.99m 2

We can see that the minimum area required in the mixed-plug case is greater than both of the
areas calculated for the single-pass, double pipe exchanger (part (a)) and the multiple-tube,
single pass exchanger in part (b). Moreover, it is clear that the single-pass, double pipe
exchanger area is the minimum possible area for the counter-current design as a result of the
large driving force for heat transfer maintained along the length of the heat exchanger. If we
then consider the case in part (c) where we consider mixed instead of plug flow on the shell side,
the resulting area represents a maximum, with the case of multiple tubes (single pass) falling
somewhere in between as a result of the fact that the flow in the shell side is not necessarily plug
flow, but also not completely well-mixed.
CHAPTER 4 SOLUTIONS

Problem 4.1

I – TCE phase
II – Aqueous phase

I
Initial acetone concentration in phase I, C Ai  0 g/liter.
II
Initial acetone concentration in phase II, C Ai  33 g/liter.
II
Volume of phase II, V = 700 liters

We have two vessels (1500 liters and 2200 liters in volume) and a total of 1400 liters of TCE
available to us. We can either use the entire amount of TCE in a single vessel to contact it with
the aqueous phase or we can use the two vessels in series.

 Single Vessel (2200liters):

Volume of phase I, VI = 1400 liters

Level I balance gives,

V I C AI  V II C AII  V I C Ai
I
 V II C Ai
II
(1)

Lowest concentration of acetone in water can be achieved when equilibrium is established


between the two phases (no net transfer of acetone will take place after equilibrium is reached).
The equilibrium relationship between the concentrations in the two phases is

C AI ,eq  MC AII,eq (2)

Using (1) and (2), we get

V I C Ai
I
 V II C Ai
II
C AII,eq 
V I M  V II

I
We have, C Ai  0, C Ai
II
 33 g/liter, V I  1400 liters, V II  700 liters, M  2

C AII,eq  6.6 g/liter

 Vessels in series:

If we want to use the two vessels in series, we need to find out how the 1400 liters of
TCE should be distributed between the two vessels so as to effect maximum acetone

1
extraction. Lets assume that the volume of TCE in the first vessel is x liters; the amount
of TCE in the second vessel would then be (1400 – x) liters.

For vessel 1, VI = x liters.

V I C Ai
I
 V II C Ai
II
700  33 23100
C AII,eq   
V M V
I II
2 x  700 2 x  700

For vessel 2, VI = (1400 – x) liters and the initial concentration of acetone in the aqueous
23100
phase would be C AiII  (the exit concentration of acetone in the aqueous phase
2 x  700
from the first vessel). So,

23100
700 
V I C AiI  V II C AiII 2 x  700  1.617 107
C AII,eq  
V I M  V II 2(1400  x)  700 (2 x  700)(3500  2 x)

We wish to find the value of x for which C AII,eq is minimum. Setting the derivative of
C AII,eq with respect to x zero, we get

dC AII,eq 2 2
  0
dx (3500  2 x)(2 x  700) (2 x  700)(3500  2 x) 2
2

  (3500  2 x)  (2 x  700)  0
 x  700

So, in the case where we use the two vessels in series, the lowest concentration of acetone
in water can be achieved when we use the two vessels in series and divide the TCE
equally between the two vessels (700 liters each).

The final acetone concentration will be

1.617 107 1.617 107


C AII,eq    3.67 g/liter
(2 x  700)(3500  2 x) 2100  2100

We see that using the two vessels in series gives us a lower final acetone concentration in
the aqueous phase.

Note: The difference in density between phase I and II can be used to separate them in
the first vessel and transfer the aqueous phase to vessel 2.

2
Problem 4.2

Phase I – TCE + Acetone


Phase II – Water + Acetone

Using level III analysis, we get

d (C AI V I )
dt

  K m a C AI  MC AII 
d (C AII V II )
dt

 K m a C AI  MC AII 
If we assume that the volumes of the two phases do not change substantially during the
operation (i.e., VI and VII remain constant), the above equations become

VI
dC AI
dt

  K m a C AI  MC AII  (1)

V II
d (C AII )
dt

 K m a C AI  MC AII 
Using level I balance on acetone, we have

V I C AI  V II C AII  V I C Ai
I
 V II C Ai
II

V I C Ai
I
 V II C Ai
II
 V I C AI
 C AII 
V II

Plugging this expression for C AII in equation (1), we get

V
dC AI  M
  K m a C AI  II V I C Ai
I I
 V II C Ai
II
 
 V I C AI  
dt  V 

V V I II dC AI
dt
  
  K m a C AI V II  MV I  M V I C Ai
I

 V II C Ai
II

C AI t
dC AI Kma
 C AI V II  MV I   M V I C AiI  V II C AiII  

V IV II 0
dt
 
I
C Ai

3
1   
 C AI V II  MV I  M V I C AiI

II 
 V II C Ai Kma
   I II t
V II
 MV I

ln 
  I II

C Ai  MC Ai V II
 V V

C AiI , C AiII , V I , V II and M are known. So, to determine the time required to carry out the
operation, we will require the knowledge of the mass transfer coefficient, Km and the total
interfacial area between the two phases, a.

4
Problem 4.3

Phase I: TCE + Acetone


Phase II: Water + Acetone

VI = VII = 700 liters


qF = q = 10 liters/min

Level I balance for this semi-batch contactor gives

d (V I C AI ) d (V II C AII )
  q FI C AF
I
 q I C AI
dt dt

Level III balance gives,

V II
dC AII
dt

  K m a C AI  MC AII  (1)

VI
dC AI
dt

 q FI C AF
I
 
 C AI  K m a C AI  MC AII  (2)

Equations (1) and (2) are coupled but can easily be solved numerically. If we assume that
changes in C AII with time are negligible, then the above equations get de-coupled and can be
solved analytically.

This liquid-liquid semi-batch operation will not work because the two phases cannot be
separated. Semi-batch contactors are generally useful for gas-liquid or solid-liquid mass
contactors.

5
Problem 4.4

To determine the saturated concentration of NaCl in water ( C SI ), the following steps need to be
carried out:

 Measure the weight of the salt available using a balance (say wi grams).
 Fill the 1 liter beaker with some fixed volume of water (say V I = 0.5 liters) and add the
available salt to it while constantly stirring/agitating the solution.
 After the solution is saturated, no more salt will dissolve in the water-salt solution and
addition of any more salt would result in the salt precipitating out of the solution
(becoming visable).
 Measure the weight of the total undissolved salt (say wf grams).
 Moles of salt in the saturated solution will be given by (wi – wf)/MNaCl
(where MNaCl is the molecular weight of NaCl in grams/mole).
(w  w f )
 CSI  Ii moles/liter.
V M NaCl

6
Problem 4.5

(a) The development of the model equations is as follows. First, we define:

I – Aqueous phase (water + octanoic acid)


II – Xylene + octanoic acid

VI = 2.25 liters, VII = 0.2 liter

C AiI = 2.75x10-4 gram-moles/liter, C AI ,eq = 7.8x10-5 gram-moles/liter


II
C Ai  0 gram-moles/liter.

By a level I balance,

V I C AI ,eq  V II C AII,eq  V I C AiI  V II C AiII .

V I (C Ai
I
 C AI ,eq )
As C II
Ai = 0, we have C AII,eq  II
= 221.625x10-5 gram-moles/liter
V

C AI ,eq
and M  = 0.0352.
C AII,eq

By level IV balance,

d (C AI V I )
dt

  K m a C AI  MC AII 
Assuming constant volume,

V
dC AI
I
dt

  K m a C AI  MC AII 
By level I balance,

V I C AI  V II C AII  V I C Ai
I
 V II C Ai
II

V I (C Ai
I
 C AI )
As C AiII = 0, we have C AII 
V II

7
dC AI  MV I 
So, V I   K m a C AI  II (C Ai
I
 C AI )
dt  V 

C AI t
dC AI Kma
 C I II
 MV I )  MV I C Ai
I
 
V I V II
 dt
C I
Ai
A (V 0

Integrating, we obtain

1  C AI (V II  MV I )  MV I C Ai
I 
Kma
ln     I II t .
(V II  MV ) 
I I
C AiV II
 V V

(b) We must now linearize the model equation in order to fit the data and compute Kma.
From part a, we have

 C AI (V II  MV I )  MV I C Ai
I 
K m a (V II  MV I )
ln   t
 
I
C Ai V II V I V II

 C I (V II  MV I )  MV I C Ai
I 
(V II  MV I )
So, plotting ln  A  against t should yield a

I
C Ai V II  V I V II
straight line with a slope equal to  K m a .

Time C AI x104 (gram- (V II  MV I )  C AI (V II  MV I )  MV I C Ai


I 
(seconds), t t ln  
moles/liter) V I V II 
I
C Ai V II 
0 2.75 0 0
10 2.13 6.204444444 -0.377948991
20 1.72 12.40888889 -0.739956762
30 1.45 18.61333333 -1.078595781
40 1.23 24.81777778 -1.476688644
60 1.03 37.22666667 -2.064628167
80 0.94 49.63555556 -2.511108765
120 0.83 74.45333333 -3.675442858
 0.78

8
0
-0.5
-1

y -1.5
-2
-2.5
y = -0.0515x
-3
-3.5
R2 = 0.9887
-4
0 20 40 60 80
x

(c) From the slope of the above plot (where x is linearized time and y is the linearized
concentration), we obtain Kma = 0.0515 liters/sec. = 51.5 cm3/sec. This value is of the
same order as that reported, i.e., 75 cm3/sec. NOTE: the fitting of the data requires that
y(x=0) = 0, as we know that C AiII (t=0) = 0.

9
Problem 4.6

(a) CO2 bubbler reaction pathway

CO2  H 2O  H 2CO3 K1  1.0


H 2CO3  H 2O  HCO3  H 3O  K 2  4.45 x107
HCO3  H 2O  CO32  H 3O  K 3  4.69 x1011

where concentrations are in molal and the water concentration is included in the equilibrium
constant by convention. In addition to the above pathway, we recognize another critical
equilibrium constant is given by

H 2O  H 3O   OH  KW  1x1014 M

We are also given the partition coefficient for CO2 as

CCO2  MPCO2 M  3.34 x102 M atm

Then, assuming that the PCO2 is 1.0 atm,

CO2   CCO   3.34 x102 M


2

atm 1atm 

Then, we can write the equilibrium expression for each of the above relations:

 HCO3   H 3O  
K1    
 1.0
CO2 
 HCO3   H 3O  
K2    
 4.45 x107
 H 2CO3 
CO32   H 3O  
K3    
 4.69 x1011
 HCO3  
 
KW   H 3O   OH    1x1014

We see that we have 5 equations and six unknowns. Therefore, we need to use the final relation,
the principle of electroneutrality

 H 3O    OH     HCO3   2 CO32 


   
This, gives us enough relations to solve for [H3O+] = 1.22x10-4 M. Therefore, the expected pH
at equilibrium is
pH = -log ([H3O+] = 1.22x10-4 M) = 3.91

10
This equilibrium analysis is a Level II analysis.

(b) Recognizing that this system can be evaluated by considering the bubbling CO 2 as plug flow
fluid motion through a well-mixed bulk fluid phase (I), the Level III analysis gives

dC AI K a
  mI C AI  M C AII  (1)
dt V
II
dC A K aA
  m IIv c C AI  MC AII  (2)
dz q

Let’s think about relation (1). The pressure of CO 2 within the dispersed phase (bubbles) is
regulated by the balance with the water pressure. Despite transfer of CO2 to the water, the
pressure will remain relatively constant over the span of our system (i.e., for a relatively small
height, h). We can also assume a negligible change in the bubble volume in rising from the
sparger to the water surface. Therefore, the average concentration of the bubble phase for this
small system is approximately the CO2 saturation pressure. Therefore, Eq. (1) can be simplified
to
dC AI K a
  mI C AI  MPCO2  ,
dt V

which can be separated and solved

 0.9C AI ,eq  MPCO2  Kma


ln   I t
 C A,o  MPCO2
I
 V

Since the concentration of CO2 in the continuous water phase is initially zero,

 MPCO2  0.9C AI ,eq  Kma


ln   I t. (3)
 MPCO2  V

But the left-hand side of (3) is the same between the 20- and 1-hole sparger systems. Therefore,
if we write (3) to describe each system, and divide the expressions, all of the common terms
cancel, leaving
 MPCO2  0.9C AI ,eq 
ln   K m a20
 MPCO2  20 hole  V I t20

 MPCO2  0.9C AI ,eq  K m a1
t1
ln   V I
 MPCO2 1hole
a20 hole t1
 (4)
a1hole t20

11
Since we know the time required for the 20-hole sparger system to reach 90% of equilibrium, we
must determine the ratio of areas for mass transfer between the system in order to calculate the
time required for the 1-hole sparger system to reach 90% of equilibrium.

To do this, we now consider the balance between the Stoke’s drag, FD, and the buoyancy force,
FB ,
6 I vb Rb   Rb3 g   I   II 
4
3

Assuming the density of the dispersed phase is negligible, and solving for the rise velocity

2  Rb3 g  I
vb  (5)
9  I Rb

Ultimately, we wish to calculate the residence time of the bubbles of a given radius within the
system.
h

vb

where the height of the water in each system, h, is constant. Therefore, the residence time is
given by
9h I
 .
2 Rb2 g  I

This translates to a volume of the dispersed phase (VII) of

9q II h I
VII  q II 
2 Rb2 g  I

Ultimately, we would like to find an expression for the area available for mass transfer as a
function of the bubble radius. The total area for mass transfer is given by

a  N b  4 Rb2 ,

where Nb is the number of bubbles within the system. The number of bubbles is also given by

VII
Nb  .
Vb

Therefore,

12
9q II h I 9q II h I 27 q II h I
Nb   
2Vb Rb2 g  I 4  8 Rb5 g  I
2   Rb3  Rb2 g  I
3 

such that the total area for mass transfer as a function of the bubble radius is given by

27 q II h I 27 q II h I
a  4 R 2
 (6)
8 Rb5 g  I 2 Rb3 g  I
b

Then, since the volumetric flow rate (qII), the liquid height (h), the viscosity (I), and the density
(I) are constant between the 20-hole and 1-hole sparger systems, we see that the ratio of the
areas for mass transfer between the systems is given by

3
a20 hole  Rb ,1 
  (7)
a1hole  Rb ,20 

Since the volumetric flow rate of the CO2 (Phase II) is the same between systems, we recognize
that
20  Vb ,20 hole  Vb ,1hole 
Therefore,
4  4 
20    Rb3,20 hole     Rb3,1hole 
3  3 
3
Rb ,1hole
 20 (8)
Rb3,20 hole

Therefore, from (7) and (8),


a20 hole
 20 .
a1hole

From the Level III analysis (4), we learned that

a20 hole t1

a1hole t20
Since t20 = 30 seconds
a20 hole
t1  t20  20  30sec  600sec  10 min .
a1hole

Since we have a smaller area (by a factor of 20) available for mass transfer in the 20-hole
sparger case, the longer time (10 minutes) required to reach 90% of equilibrium when the
1-hole sparger is used, seems to make intuitive sense.

13
Problem 4.7

For each of the cases, we define:


A = MeCl
Phase I = Aqueous phase
Phase II = Gaseous phase

CASE I: Well-stirred batch tank

The level I and IV balances for the batch system are derived in a similar manner to those in the
text. Writing the word statement in terms of mathematical variables for the conservation of mass
yields the level I balance for the closed batch system:

V CI I
A  V II C AII  t t

 V I C AI  V II C AII  t

lim 

d V C V CI I
A
II II
A  0
t 0 dt
I I
dV C dV II C AII
  A
dt dt

If we assume that the N2 bubbles have negligible volume change upon mass transfer, we can
integrate the above equation and obtain the familiar algebraic form of the level I balance:

 
V I C AI  C AiI  V II C AII  C AiII .  
Similarly, we can write the level IV balances for species conservation of MeCl in the aqueous
phase phase

V C 
I I
A t t  
 V I C AI  K m a C AI  MC AII t
t
 

d V I C AI   K .
lim 
t 0 dt
m 
a C  MC
I
A
II
A 
Similarly for the gaseous phase we obtain


d V II C AII K
dt
m 
a C AI  MC AII . 
The Level IV balances cannot be integrated as was the case previously derived in the text, since
we explicitly assume that Km is some unknown function of the MeCl concentration.

14
CASE II: Well-mixed semi-batch system

Following a similar procedure as previously, we derive the level I mass balance for the semi-
batch system as follows:

V C I I
A  V II C AII  t t

 V I C AI  V II C AII   q C
t
II II
Af  C AII t

d V I C AI  V II C AII q
lim 
t 0 dt
II
C II
Af  C AII 
The level IV balance remains the same for the aqueous phase, and thus


d V I C AI   K
dt
m 
(C A )a C AI  MC AII 
Whereas the level IV balance for the gaseous phase now becomes

V II
C AII  t t

 V II C AII   q C
t
II II
Af 
 C AII t  K m a C AI  MC AII t  

d V II C AII q
lim 
t 0 dt
II
C II
Af 
 C AII t  K m a C AI  MC AII  
NOTE: We can assume that we are dealing with relatively large concentrations of MeCl in each
phase, since this is usually the reason for non-constant values of Km. Thus, we cannot make the
usual pseudo-steady state assumption and would need to integrate the coupled level IV balances
in order to use them for analysis and design of mass contactor equipment.

CASE III: Well-mixed continuous tank system

The level I balance for the continuous mixed-mixed system is derived the as follows

V C I I
A  V II C AII  t t

 V I C AI  V II C AII   q C
t
I I
Af   
 C AI t  q II C AfII  C AII t


d V I C AI  V II C AII q
lim 
t 0 dt
I
C I
Af  
 C AI  q II C AfII  C AII 
Assuming steady state operation, we arrive at the familiar level I balance for a continuous flow
system

0  q I C AfI  C AI  q II C AfII  C AII   
or

15
q II

C AI  C AfI   C AfII  C AII ,    qI
.

The level IV species balance in the gaseous phase proceeds exactly the same as in the previous
case, yielding


d V II C AII q
dt
II
C II
Af 
 C AII t  K m a C AI  MC AII  
which upon assuming steady state operation reduces to

 
0  q II C AfII  C AII  K m a C AI  MC AII .  
Similarly, for the aqueous phase,

 
0  q I C AfI  C AI  K m a C AI  MC AII  
CASE IV: Well-mixed continuous reacting tank system

Since, for the reacting system, MeCl (species A) is not conserved, we must write an overall mass
balance for the combined mass of the MeCl as well as the product species N in the liquid phase.
Assuming that the reaction product N is insoluble in the gaseous phase, the level I balance
proceeds as

V C
I I
A 
 C NI  V II C AII  t t
 
 V I C AI  C NI  V II C AII    q C
t
I I
Af     
 C AI  C NfI  C NI  t  q II C AfII  C AII t

d V  C I I

 C NI  V II C 
II

   
 C AfI  C AI  C NfI  C NI   q II C AfII  C AII  
A A
lim  q I
 
t 0 dt

which for steady state operation reduces to

  
0  q I  C AfI  C AI  CNfI  CNI   q II C AfII  C AII .   
Note that the level I balance for the reacting system contains no reaction rate term. This is
because the change in species concentrations of MeCl and N are taken care of by the time
derivative term. Turning to the level IV balances, we first recognize that, if N is insoluble in the
gas phase, the level IV gas phase balance is the same as in the previous case

 
0  q II C AfII  C AII  K m a C AI  MC AII .  

16
In order to fully specify the system, a level IV balance is required for both MeCl and N in the
aqueous phase. Assuming the given reaction rate for the liquid-phase reaction, the level IV
aqueous phase balance on MeCl becomes
 V I C AI  t t
  t
   
 V I C AI  q I C AfI  C AI t  K m a C AI  MC AII t  krV I C AC N t


d V I C AI q
dt
I
C I
Af   
 C AI  K m a C AI  MC AII  krV I C AC N

which for steady state operation reduces to

   
0  q I C AfI  C AI  K m a C AI  MC AII  krV I C AC N .

Similarly, for the reaction product N (again assuming no mass transfer of N to the gas), we
obtain

 
0  q I C NfI  C NI  krV I C AC N .

17
Problem 4.8

A key concept is that the rate of mass transfer from the drug is fixed by the constant
concentration and fixed diffusion rate through the capsule. This problem is worded somewhat
ambiguously and some guidance may be necessary.

We are given the following information about Joe’s drug tablet (II= stomach, I = tablet):

I
CAI/VI= 1 x 10-7 mol/cc
CAII~ 0 (bulk concentration in stomach)

If the active drug (A) in the capsule is in a separate crystalline phase, then we can write a balance
on the total amount of A in the system, comprised of: crystalline A, the active A in the capsule
(CAI) and active A in the stomach (CAII) as:

(cVc+VI CAI+VII CAII)0 = (cVc+VI CAI+VII CAII)t

We first need to calculate the interfacial area for mass transfer.


Performing a level III balance on the system yields the following, where we have assumed
constant volume of the tablet:

VI dCAI/dt = VI rA – kma (CAI - M CAII)

Where rA is such that CAI ~ constant and CAII~ 0. This suggests that the rate of “reaction”, or
dissolution of the drug, equals the rate of mass transfer from the capsule.

Thus, rA = km (a/VI ) CAI

From the problem statement, rA = 1 x 10-11 mol/sec, and CAI/VI= 1 x 10-7 mol/cc

Thus kma = 1 x 10-11 mol/sec/(1 x 10-7 mol/cc) = 1 x 10-4 cc/sec = 1 x 10-10 m3/sec

If we had a value for km we could calculate the required area. Alternatively, given that most
capsules ~ 1 cm in diameter, and assuming that we have a spherical tablet, we can use equation
4.2.20 and table 4.6 to estimate the. For a sphere, α=4.836.

a =  [V]2/3 = 4.836 ( d3/6)2/3 = 4.836 x (0.52 cm3) 2/3 = 3.13 cm3

Then, km = 1 x 10-10 m3/sec / (3.13 x 10-4m2) = 3.2 x 10-5 m/sec.

18
This value seems reasonable, and could be checked against membrane transport models (see
Chapter 5). Finally we can compute the amount of active A required. Since the relative rate of
transfer is constant, application of the total mass load yields

mload = rA dt = 1 x 10-11 mol/sec x 8 hours x 3600 sec/hr = 2.9 x 10-7 mol

Given CAI/VI= 1 x 10-7 mol/cc, and the volume of the capsule is 0.52 cm3, the active A in the
capsule in the dissolved phase is 0.52 x 10-13 mol, which is negligible. So, we require 2.9 x 10-7
mol to be in the capsule (as crystalline material). Typical small molecule drugs have molecular
weight of 500 g/mol or less. Then, 2.9 x 10-7 mol x 500 g/mol = 1.5 x 10-4 g, or 0.15 mg.

For a density of 1 g/cc, this would occupy a volume of 1.5 x 10-4 cc. Given the capsule volume
of 0.52 cm3 (cc), this is less than 1% of the volume.

NOTE: There are many considerations for controlled release design, but this is one strategy. We
need to engineer a constant release rate by exploiting dissolution (fast) limited by a capsule
membrane with a specified km.

19
Problem 4.9

Phase I = Aqueous phase


Phase II = Gaseous phase

The Level IV balances for a batch system are:

I
dC A
  K m a [ C A  MC A ]
I I II
V
dt
II
dC A
 K m a [ C A  MC A ]
II I II
V
dt

Since we are using a gas-liquid system, we can recognize that we no longer have a concentration
but a partial pressure for oxygen, and focus on the equation for C AI, since that is the
concentration we are asked to plot over time. We first need to determine the relationship between
the Henry’s law constant and partial pressure and MCAII:

MC A  K H PA
II II

II
P
 A
II
Using the ideal gas law, we know that C A
RT

II
MPA
 K H PA
II

RT
M
 KH
RT
M  K H RT

So the differential equation becomes:

I
dC A
  K m a[C A  K H RTC A ]   K m a[C A  K H PA ]
I I II I II
V
dt

This equation is separable if we can assume PAII remains constant and evaluate the validity of
that assumption later on. Therefore, we can integrate using separation of variables.

CAI
 Kma
I t
dC A
 [C A  K H PA ]
I II

VI
dt
C I Ai 0

C A  K H PA  Kma
I II

ln[ ] t
C Ai  K H PA
I II
VI

20
We know that CAiI is zero (system is held in vacuum before beginning). We can easily calculate
the contact area of phase I from the surface of the liquid.
a  r 2   (10cm) 2
a  314.1cm 2

And then the volume is found to be using the formula for volume of a cylinder:

V I  (r 2 )h
V I  ( (10cm) 2 )(15cm)
V I  4712cm 3

We are also given that:

cm
K m  2.1  10 3
s
mol
K H  1.26  10 3
Latm
PA  PO2  0.21PAIR

Now we can go back to the result of our integration and manipulate the equation until we can
obtain a plot as a function of time for PAIR values of 1, 5 and 10 atm.

 Kma
C A (t )   K H PA exp( t )  K H PA
I II II
I
V

Plugging in, we obtain

3 cm 2
 ( 2.1  10 )(314.1cm )
I 3 mol s 3 mol
C A (t )   (1.26  10 )(0.21PAIR atm ) exp[ t ]  (1.26  10 )(0.21PAIR atm )
Latm 3 Latm
( 4712cm )

This equation is plotted at the end of this solution (dotted lines).

But first we need to go back and perform the solution more rigorously to determine the validity
of the assumption of constant partial pressure, PO2. Let’s look again at the now coupled
differential equations.

I
dC A
  K m a[C A  K H PA ]
I II
VI
dt
II
dC A
 K m a[C A  K H PA ]
I II
V II
dt

21
We recognize that we can use the model equation for CAI(t) as solved for in equation 4.3.4, being
sure to correct for the difference between the pressure/concentration units.
M 
C A (t )  (C Ai  C Ai )  (C Ai  MC Ai )e  t
I I II I II

M  M 

Here again,

M  K H RT
mol Latm
M  (1.26 10 3 )(0.0821 )(298 K )
Latm molK
M  3.08 10  2

Also,
V II

VI
5L

1mL 1L
(4712cm 3 )( 3
)( )
1cm 1000mL
  1.061

Next we can calculate β.

1 1
  MK m a ( I
 II )
MV V
cm 1 1
  (3.08  10  2 )(2.1  10 3 )(314.1cm 2 )( 2
 )
s (3.08  10 )(4712cm 3 ) 3
1cm 1000mL
(5 L)( )( )
1mL 1L
  1.44  10  4 s 1

Finally, we need to convert pressure to concentration.

II
PAi 0.21PAIR atm
C Ai    8.58  10 3 PAIR
II

RT Latm
(0.0821 )(298 K )
molK

Plugging these into the model equation and plotting, we obtain:

3.08 10 2 1.061 4


C A I (t )  2
[0  (1.061)(8.58 10 3 PAIR )]  2
[0  3.08 10  2 (8.58 10 3 PAIR )]e  (1.4410 )t
3.08 10  1.061 3.08 10  1.061

22
Problem 4.5

0.0035

0.003
P=10 atm

0.0025

0.002
Ca(mol/L)

P=5 atm
0.0015

0.001

0.0005
P=1 atm

0
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000
time (s)

In the figure above, the dotted lines correspond to the concentration of A (O 2) in the solution
assuming that the chamber pressure (concentration of O2 in the chamber) changes negligibly
with time. The solid lines correspond to the rigorous solution of the coupled model. It is obvious
that the assumption of constant partial pressure of O2 leads to fairly significant deviations from
the rigorous solution, and is therefore not an accurate assumption.

23
Problem 4.10
CO2II, qII

CO2FI, qI CO2I, qI

Given:
dpore= 0.1 m CO2FII, qII
CO2FI = 40 mmHg
CO2I = 95 mmHg
M’ = 1.53 x 10-6 mol/cm3/mmHg
Km = 2.91 x 10-3 cm/sec
qI = 200 mL/min

In terms of the notation in the text, we recognize that we can express C AI as

MC AII  M ' PAII

where phase II is the oxygen gas phase. Using the Ideal Gas law, we can express the
concentration in the gas phase as
P II
C AII  A
RT

such that the relation between the partition coefficient given and the dimensionless one used
throughout the text is

 mol  l atm   1000cm3   760mmHg 


  298 K  
6
M  M ' RT  1.53 x10   0.082  
 cm3 mmHg   mol K   l   1atm  .
M  28.41

We will then be able to use this dimensionless partition coefficient in the model equations. The
partition coefficient provided assumes that at equilibrium there is a linear relation between the
oxygen gas pressure and the oxygen dissolved in the blood. In the medical profession,
concentrations of gas absorbed in the liquid phase are often expressed in terms corresponding
equilibrium gas phase pressure (mm Hg). This is simply due to the fact that the units of ‘mmHg’
tend to be more easily understood than gmol/m3, for instance. We can, therefore, calculate the
concentration of oxygen in the blood by using the partition coefficient:

CO2FI = CAFI = 40 mmHg (1.53 x 10-6 mol/cm3/mmHg)


= 6.12 x 10-5 mol/cm3 = 6.12x10-2 mol/L

CO2I = CAI = 95 mmHg (1.53 x 10-6 mol/cm3/mmHg)


= 1.45 x 10-4 mol/cm3 = 1.45x10-1 mol/L

24
A Level I analysis of the system yields:

qFI C AF
I
 qFII C AF
II
 q I C AI  q II C AII

q  q
II I C I
AF  C AI 
CII
A  C AF
II

If we assume that we have oxygen available at 760 mmHg, we can calculate the corresponding
concentration of oxygen in the gas phase:

n P 760mmHg
II
C AF   
V RT  l atm   760mmHg 
 0.082    298 K 
 mol K   1atm 
mol
II
C AF  4.1x102
l

Then we can solve for the minimum oxygen flow rate. This occurs for the counter-current flow
case when

,eq  MC A,eq .
I II
C AF

As discussed above, the inlet oxygen concentration in the blood was given in terms of the
corresponding equilibrium pressure. Therefore, to solve for the minimum flow rate, the outlet
equilibrium pressure of oxygen in II is simply 40 mmHg. We can convert this to a concentration
via the Ideal Gas law

n P 40mmHg
C AII   
V RT  l atm   760mmHg 
 0.082    298 K 
 mol K   1atm 
mol
C AII  1.45 x101
l

Therefore, the minimum oxygen flow rate can be calculated as

q II
q I C I
AF  C AI   200 mL  6.12 x10 2
 1.45 x101   432 ml / min
min
C II
A  C AF
II
 min  2.2 x10 3
 4.1x102 
If we consider an oxygen flow rate of 500 ml/min, we can calculate the outlet concentration of
the oxygen stream (not asked for) (II) from the Level I analysis.

qI
C AII 
q II
I
C AF  C AI  C AF
II

25
 200  mol
C AII  
 0.5
 
6.12 x102  1.45 x10 1  4.1x10 2 
 l
mol
C AII  7.5 x103
l

(NOTE: If we chose a higher flow rate we might find that there would be minimal change in the
concentration of the oxygen in phase II, but we would have to justify this if we were to simply
solve the model equation assuming constant oxygen pressure.)

Performing a level III analysis gives the model equations in the text (4.4.7). We can solve the
model equation describing the concentration of oxygen in the blood (I)

  C AII, F  e   L C AI , F   C AI , F  MC AII, F   z
C AI  M       e
   Me   L     Me
 L

for  in order to calculate the area required for a feasible design.

  C AI , L  M  C AII, F 
  L  ln  
 MC A, L  M  C A, F  C A, F    M  
I II I

Here,
q II 500
   2.5
q I 200
M  28.41
C AI , L  1.45 x101 mol / m3
C AI , F  6.12 x102 mol / m3
C AII, F  4.1x102 mol / m3

Therefore,
Kma
  L  1.91    M 
q II
Solving for the area, where Km=2.91x10-5 m/sec and qII=500mL/min=8.3x10-6 m3/sec,

a
1.91q II


1.91 8.3 x10 6 m3 / s 
Km   M  
2.91x105 m / s  2.16  28.41 
a  2.08 x102 m 2

This a is the total area of the pores that is required since mass transfer through the polypropylene
is negligible. We can assume a reasonable porosity of =0.5, which means that half of the
surface area of the tube is available (through the pores) for mass transfer. Then, the total tubing

26
surface area is given as twice (1/) the area calculated above (a) such that amicroporous=4.2x10-2
m2.

In order for the model equations to hold, we would want to calculate the diameter required for
turbulent flow of the blood.

d max 
4 q

4 
1060kg / m3 8.3 x106 m3 / s 
 747 x106 m  0.7 mm 
3 2
N  Rec  max  5000 3 x10 Nsm
Here, it is important to note that we have used a low critical Reynolds number. One would have
to actually try to minimize the shear stress on the blood cells to avoid blood cell rupture, and
should also try to minimize the residence time in the unit.

If we then assume an inner diameter of the tube of 600x10-6 m = 0.7mm, we can calculate the
total length of tubing required.

L
amicroporous

 4.2 x10 m   22m 
2 2

 dtube , ID   600 x10 m  6


Since this unit will have to be used in a medical setting, it must be reasonably sized. We choose
a length, L, of 0.3m, which means that we must have a total of 74 microporous tubes in the unit.

The corresponding low residence time in this unit is given by


74  0.3m  600 x106 m 
2

 1.8s .

4 3.33 x106 m3 / s 
If we assume a thickness of the polypropylene tubes of a fraction of the diameter, t=10m, then
we can calculate the maximum diameter of the shell housing this tube bank based on the
hydraulic diameter:

The wetted perimeter of the tubes is given by


Pwet ,tubes  N  d outer  74  600 x10 6 m  20 x10 6 m   0.144m
Then

Pshell 
4 q
 Ptube 
 
4 1.164kg / m3 500m3 / s 
 0.144
Re 10000
Pshell  8.89 x102 m

Therefore, the diameter of the shell casing (inner diameter) is 2.8x10-2 m = 2.8 cm.

27
Problem 4.11

We have two mass contactors:

(1) Phase I = supercritical CO2; Phase II = coffee beans


(2) Phase I = supercritical CO2; Phase II = water

(a) CA = concentration of caffeine


CAI = MCAII
Phase I = supercritical CO2; Phase II = coffee beans

CASE I:
Batch system of coffee beans and supercritical CO2 at t=0.
0
Level I:
(CAIVI – CAiIVI) = - (CAIIVII – CAiIIVII)

and assuming volume change is negligible.

At equilibrium,

CA,eqIVI = - (CA,eqII – CAiII)VII

MCA,eqIIVI = - (CA,eqII – CAiII)VII

II
C Ai V II

II
C A,eq
MV I  V II

Level IV:
VI
dC AI
dt

  K m a C AI  MC AII 

V II
dC AII
dt

 K m a C AI  MC AII 
Substitute in CAII obtained from Level I balance and solve for C AI.

M M
C A (t )  C AiII  C AiII e  t
I

M  M 

M
C A (t )  C AiII (1  e  t )
I

M 

28
V II K a M 
where   I and   mI 1   .
V V  

CAII(t) can be found from a Level I balance since we now know C AI(t).

CASE II:
Semi-batch system of coffee beans and supercritical CO2 at t=0.

ASSUME we have a well-mixed system, so we will use mixed-mixed fluid motion.

Level IV balances:

V II
dC AII
dt

 K m a C AI  MC AII 

VI
dC AI
dt

 q I C AF
I
 
 C AI  K m a C AI  MC AII 
The above equations can be solved rigorously.

NOTE: If we had a tank exchanger that we did not mix, but instead loaded with coffee beans and
passed supercritical CO2 through, we would expect the concentration of both phases to change
with tank depth. We would have to then model the fluid phase using plug flow, but the dispersed
phase would be difficult to model since there would be changes in caffeine concentration within
the beans with space and time.

(b) CA = concentration of caffeine


CAI = MCAII
Phase I = supercritical CO2; Phase II = water

CASE I:
Continuous system of supercritical CO2 and water.

Let qFI = qI and qFII = qII. Let CAFII = 0 (pure water).

Level I:
(CAIqI – CAFIqI) = - (CAIIqII – CAFIIqII)

CAFIqI = CAIqI + CAIIqII

Level IV:

29
V IdC AI
dt
 q I C AFI
  
 C AI  K m a C AI  MC AII 
V II
dC AII
dt

  q II C AII  K m a C AI  MC AII 
Assume steady state

0  q I C AF
I
  
 C AI  K m a C AI  MC AII 

0   q II C AII  K m a C AI  MC AII 

From Level I C A 
II qI

I
C AF  C AI 
q II
Therefore,
 I 
0q C I
 I
AF
I
A  qI
 C  K m a C A  M II C AF
I
 C AI  
 q 

Solve for CAI:

 I qI  I  I qI 
C q  K m a  MK m a II   C AF q  MK m a II 
I
A
 q   q 

q II
Multiply through by I .
q Kma

 q II q II  I  q
II

C AI   I  M   C AF  M
 Kma q   Kma 

q II q II
Let   and   .
qI Kma

C AI     M   C AF
I
  M 

C AI 
I
C AF   M 
    M 

CAII can be found from a Level I balance since we now know CAI.

30
CASE II:
Tubular reactor with co-current flow.

Assume negligible changes in q. (Let qFI = qI and qFII = qII.)


Level I is the same as Case I above:

(CAIqI – CAFIqI) = - (CAIIqII – CAFIIqII)

CAFIqI = CAIqI + CAIIqII

Level IV:

qI
dC AI
dz
K a

  m C AI  MC AII
L

q II
dC AII K m a I
dt

L

C A  MC AII 
Solving this gives:

I
MC AF C AF
I
C AI ( z )   e  z
M M

I
C AF CI
C AII ( z )   AF e  z
M M

with parameters as given above.

31
CHAPTER 5 SOLUTIONS

Problem 5.1

a) We can start by using the word statement to perform an energy balance on a differential slice
of the tank wall

 4r 2 C p Tr t  t   4r 2 C p Tr t  q 4r 2 t z  q 4r 2 t z  z

Dividing all terms by Δt and Δr and taking the limit as they approach zero, we get the
derivatives:

 
( r 2 C p T )   (qr 2 )
t r

Since we know radius is not a function of time and that density and Cp are constant,

 
r 2 C p (T )   (qr 2 )
t r

We also have Fourier’s constitutive equation of heat flow:

T
qr  k
r

We can plug this into the balance equation, assuming k is constant.

 k  T 2
C p (T )  2 ( r )
t r r r

Assuming steady state, we can set the time derivative to zero.

k d dT 2
0 ( r )
r 2 dr dr

We can integrate twice to get the general form of the solution.

dT
 d ( dr r )   0dr
2

dT 2
r A
dr

1
dT A
 2
dr r
A
 dT   r 2
dr

A
T (r )  B
r

The boundary conditions are:

T ( R1 )  T1
T ( R2 )  T2

So we can apply these to solve for A and B, our unknown constants

A
T1  B
R1
A
T2  B
R2

Subtracting to eliminate B, we get:

A A 1 1
T1  T2    A(  )
R1 R2 R1 R2
R2  R1
T1  T2  A( )
R1 R2
(T1  T2 ) R1 R2
A
R2  R1

Now we can go back to one of the equations to get B.

(T1  T2 ) R1 R2
T1  B
R1 ( R2  R1 )
(T1  T2 ) R1 R2
B  T1 
R1 ( R2  R1 )
T1 R1 R2  T1 R1 R1  T1 R1 R2  T2 R1 R2
B
R1 ( R2  R1 )
T2 R2  T1 R1
B
( R2  R1 )

Plugging back in, we obtain the final form of the temperature profile

2
(T1  T2 ) R1 R2 T2 R2  T1 R1
T (r )  
( R2  R1 )r ( R2  R1 )

Now that we have the temperature profile, it is trivial to obtain an expression for the mass boil
off rate, m . We can start by calculating the energy boil off rate, Q using Fourier’s constitutive
relation.

Q  qa
dT dT
Q   ka  4r 2 k
dr dr

dT A
From above,  2
dr r

(T  T ) R R
Q  4r 2 k 1 2 1 2 2
( R2  R1 )r
(T  T ) R R
Q  4k 1 2 1 2
( R2  R1 )

Q
m 
Hv
b) Plugging in the values we are given, we can now calculate Q
0.305m
(60  25)(49.5 ft )(50 ft )( )
 W 1 ft
Q  4 (16 )
mK (49.5 ft  50 ft )
Q  2.6  10 7 W

And then we can solve for m given Q and the enthalpy of vaporization.
J
 2.6  10 7
Q sec
m  
Hv J
2.14  10 5
kg
kg
m  121
sec

3
Problem 5.2

a.) We begin by performing an energy balance around a differential slice of the annulus

 2rLC p Tr t  t   2rLC p Tr t  q 2rLt r  q 2rLt r  r  r 2 LQ t

Canceling like terms, dividing all terms by Δt and Δr and taking the limit as they approach zero,
we get the derivatives:

 
(rC p T )   (qr )  Q r
t r

Since we know radius is not a function of time and that density and C p are constant,

T 
C p r   (qr )  Q r
t r

Assuming steady state, we can set the time derivative to zero.

d
0 (qr )  Q r
dr

Plugging in Fourier’s constitutive equation for the energy flux, we get

d dT
0 (k r )  Q r
dr dr
d dT
k ( r )  Q r
dr dr

We can integrate twice to get the general form of the solution.

dT Q r
 d( dr
r)   
k
dr

dT Q r 2
r A
dr 2k
dT Q r A
 
dr 2k r

4
Q r A
 dT   ( 2k r
 )dr

Q r 2
T (r )    A ln r  B
4k

The boundary conditions are as follows:

@ r  R1 , q  0
.
@ r  R2 , T  T2

We now apply the BC’s to solve for A and B, our unknown constants. For flux equal to zero, the
first derivative of temperature must equal zero at R1.

Q R1 A
  0
2k R1
A Q R1

R1 2k
Q R1
2

A
2k

Now we can use the second boundary condition to determine the value of B.

Q R2 Q R1
2 2
T2   ln R2  B
4k 2k
Q R2 Q R1
2 2
B  T2   ln R2
4k 2k

Plugging back in, we obtain the final form of the temperature profile

Q r 2 Q R1 Q R2 Q R1
2 2 2

T (r )   ln r  T2   ln R2
4k 2k 4k 2k
Q R2 Q R1
2 2
r r
T (r )  T2  (1  ( ) 2 )  ln( )
4k R2 2k R2
Q R2
2
r r
T (r )  T2  [(1  ( ) 2  2 ln( )]
4k R2 R2

A plot of this solution is similar to the following:


(Arbitrary values chosen for R2 = 10, T2 = 100)

5
b.) Now we need to evaluate q, energy flux, at the outer surface, R2.We can simply take the
derivative of the temperature with respect to r, and evaluate it at R2.

Q R2 Q R1
2
dT
q R2  k R2  
dr 2 2 R2
Q R2 R
q R2  [1  ( 1 ) 2 ]
2 R2

Next we need to confirm that this result is consistent with an overall energy balance. The total
heat generated (Q) is equal to:

Q  VQ
Q  L( R2  R1 )Q
2 2

So the heat flux through the outer surface is equal to

Q L( R2  R1 )Q
2 2

q  
2R2 L
R2
a

We can simplify to:

6
Q ( R2  R1 )Q
2 2

q R2  
a 2 R2
Q R2 R
q R2  [1 ( 1 ) 2 ]
2 R2

It is obvious that this is the same result obtained as from taking the derivative of the temperature
profile, so we can feel confident that our temperature profile from part a is correct and that
energy flux is consistent.

7
Problem 5.3

a.) We are presented with the same physical situation as the previous problem. Thus, the
general balance equation will remain the same:

T 
C p r   (qr )  Q r
t r

Assuming steady state and applying Fourier’s constitutive equation, the balance becomes:

k d dT
0 ( r )  Q
r dr dr

However, we now need to treat balances around the oil and around the shaft differently. In the
oil, we are given that

R1 2
Q   [ ]
R2  R1

But in the shaft,

Q  0

So the energy balance for the shaft becomes:

k d dT
0 ( r) .
r dr dr

We can integrate twice to solve for the general form of the temperature profile:

dT
 d ( dr r )   0dr
dT
rA
dr
dT A

dr r
A
 dT   r dr
T (r )  A ln r  B

Now we need the boundary conditions for the shaft. From symmetry, we know:

8
dT
@ r  0, 0
dr

and we also know

@ r  R1 , T  Ts ,out .

We can solve for the integration constants by plugging into the general form. A must be equal to
zero in order for the derivative (dT/dr) to equal zero, so

Ts ,out  B

Therefore,

Tshaft (r )  Ts ,out which is constant.

In the oil, recall the energy balance equation is:

k d dT
0 ( r )  Q
r dr dr

This is the exact same equation that we integrated in Problem 2, part a. The general solution is:

Q r 2
Toil (r )    A ln r  B
4k

But now we will have different boundary conditions as we have relaxed the assumption that the
shaft is an insulator:

dT
@ r  R1 ,  0 Since T is constant
dr
@ r  R2 , T  To ,out

Solving for the unknown constants of integration,

Q R1 A
  0
2k R1
A Q R1

R1 2k
Q R1
2

A
2k

9
Q R2 Q R1
2 2

To ,out   ln R2  B
4k 2k
Q R2 Q R1
2 2

B  To ,out   ln R2
4k 2k

So the temperature profile becomes:

Q r 2 Q R1 Q R2 Q R1
2 2 2

Toil (r )   ln r  To ,out   ln R2
4k 2k 4k 2k
Q R2 r 2 Q R1
2 2
r
Toil (r )  To ,out  (1  ( ) )  ln( )
4k R2 2k R2
Q R2
2
r r
Toil (r )  To,out  [(1  ( ) 2  2 ln( )]
4k R2 R2

But we don’t know what To,out is equal to. Performing a level III balance on the oil yields

Q  haT
Q  h 2R2 L(To ,out  T2 )

and in the oil, we know that

Q  Q V  Q L( R2  R1 ) .
2 2

Setting these equal gives

2hR2 (To ,out  T2 )  Q ( R2  R1 )


2 2

2hR2To ,out  2hR2T2  Q ( R2  R1 )


2 2

2hR2To ,out  Q ( R2  R1 )  2hR2T2


2 2

Q ( R2  R1 )
2 2

To ,out   T2
2hR2

Plugging this into the temperature profile equation, we get the final expression for T oil(r)

Q ( R2  R1 ) Q R2
2 2 2
r r
Toil (r )   T2  [(1  ( ) 2  2 ln( )]
2hR2 4k R2 R2

Recall that Tshaft=Toil evaluated at R1, so

10
Q ( R2  R1 ) Q R2
2 2 2
R R
Tshaft (r )   T2  [(1  ( 1 ) 2  2 ln( 1 )]
2hR2 4k R2 R2

b.) No, the temperature of the shaft is equal to the outer temperature of the oil, and is constant,
as determined in part a. The shaft temperature profile has no functional dependence on position,
so will not vary throughout. The temperature of the shaft also does not depend on the material of
construction of the shaft. The temperature of the shaft is determined solely by the outer
temperature of the oil, and is therefore a function only of the oil properties, the resistance to heat
transfer in the outside film (h), and the device geometry (a).

c.) Using the following parameters:

R1=1 cm, R2 = 1.2 cm, T2=25°C, μ=0.1 Pa-s, Ω=10-10,000 rpm


From Table 1: koil = 0.14 W/m-K, hoil = 100 W/m2K

The location-independent temperature of the shaft can be plotted as a function of rotation rate,
yielding the plot below.

301
Shaft temperature (K)

300

299

298
0 2000 4000 6000 8000 10000
Rotation rate (rpm)

d.) Again we can use the level III balance on the oil to obtain an expression for the overall heat
transfer coefficient.

Q  haT
Q  h 2R2 L(Tshaft  T2 )
Q  Q V  Q L( R  R )
2 2
2 1

h 2 R2 (Tshaft  T2 )  Q ( R  R1 )
2 2
2

Q ( R2  R1 )
2 2
h
2 R2 (Tshaft  T2 )

11
Problem 5.4

The tiles on the space shuttle can essentially be thought of as slabs of material that separate the
aluminum hull of the shuttle from the atmosphere. The melt temperature of aluminum is:

Tmelt Al  677 o C (http://www.omega.com/temperature/Z/pdf/z048.pdf).

Assuming that we would like the hull to stay intact, the maximum temperature of the hull should
be something significantly less than this. Let’s choose 600°C. Typical temperatures of the
atmosphere outside of the hull due to frictional heating around the shuttle can easily reach
around

Tout  1600o C .

We can assume a thermal conductivity of the shuttle tile equal to that of air, since the tile is
porous, such that nearly all of the material is air.

 W 
k  0.026 
 m  K 

We can calculate the flux into the aluminum based upon tile thickness of 10cm required to keep
the aluminum sufficiently below its melting temperature by Newton’s law of cooling:

k
q (Tout  600 C)
h
 W 
0.026 
q  m  K  (1600 C  600 C)
0.1 m 
W
q  260
m2

This is a small amount of heat per area, and can be easily dissipated.

12
Problem 5.5
(Note for reference, the forensic “rule of thumb” is time(hr)=98-T(oF)/1.5, which yields 42
hours)

R=15cm

Ta (bulk air) = 20°C


k = 1.5 x 10-3 cal/cm-sec-K
 = 1.1 g/cc
Cˆ p = 0.83 cal/g-K
h = 8 W/m2-K

We begin by converting the given values into SI units:


cal 4.19 J 100cm
k  1.5 x103 * * 
cm  sec K cal m
W
k  0.63
mK
g 106 cc 1kg kg
  1.1 * 3 * 3  1100 3
cc m 10 g m
cal 4.19 J J
Cˆ p  0.83 *  3.5
gK cal gK
W
0.63
k mK m2
   1.6 x107
 Cˆ p 1100 kg *3500 J s
3
m kgK
The next step is to evaluate the Biot number, which is defined in terms of the radius of the
cylinder:
W
8 2 *0.15m
hR
Bi   m K  1.9
k W
0.63
mK
This value suggests that conduction and convection are comparable, such that a lumped analysis
will not be accurate and the full temperature profile must be considered.

A simple and quick estimate can be obtained through the use of the Gurnie and Lurie charts,
which can be found in the Chemical Engineers’ Handbook, Figure 10-2, for an infinite cylinder.
To use this chart, we must first calculate the dimensionless centerline temperature:
Tcenter  T 29  20
  0.53
T0  T 37  20
Then, we need to calculate “m”, which is just m  1 , which is 0.525. From the chart with this
Bi
value as the y-axis, we estimate the Fourier number (x-axis)to be:

13
t
Fo   0.4
R2
 0.15m 
2
R2
 t  (0.4)  1*  5.8 x104 sec
 m 2
1.6 x107
s
 t  16h

Conclusion: The murder occurred approximately 16 hours ago, therefore Howard, who left town
the previous day (more than 16 hours ago) is innocent. Mickey is a possible culprit as he was
available to commit the crime for the past 30 hours. Note that the “forensic rule” cited above
would suggest a much longer time, which would implicate Howard, not Mickey.

Note, a more accurate estimate can be obtained by solving the energy balance for this problem.
To start this problem, we begin by performing a level III energy balance on a thin cylindrical
shell of thickness r ,

2 C p rLr T | t  t T | t   2Lrq r  | r rq r  | r  r 


T 1  ( rqr )
 C p 
t r r

T
Using q r  k (Fourier’s constitutive equation for heat conduction), we obtain
r

T    T 
  r 
t r r  r 

k
where   is the thermal diffusivity.
C p

T  2 T  T
  2 
t r r r
The initial and boundary conditions are:
t  0 : T  37 0C
dT
r  0 : qr  0  0
dr
dT
r  R : qr   k  h T ( R )  T 
dr

A direct numerical calculation is difficult as noted in the text due to the discontinuity at the start
of the process. To obtain a robust numerical solution, we can discretize above equation using

14
dT T (ri 1 )  T (ri ) d 2T T (ri 1 )  2T (ri )  T (ri 1 )
 and 
dr r dr 2 r 2
and

dT T j 1  T j

dt t
Further, we can define the dimensionless Fourier number as:
t
Fo  2
r
Then, we find
  r   r  
 T j 1 (ri )  T j (ri )  Fo * T j (ri 1 )   2   T j (ri )  1   T j (ri 1 )  i  1 to (n  1)
  ri   ri  

Our initial condition is T (ri )  37 o C at t = 0 (initially the temperature everywhere in the


cylinder is the normal human body temperature).

We can discretize the boundary conditions to get

T (r1 )  T (r0 )
(i) 0
r

T (rn )  T (rn1 )  h T
(ii)  k   rn   T 
r

This last boundary condition can be formulated to give an equation for the temperature at the
surface:
T (rn )  T (rn1 )  h T r  T 
k
r   n 
T j (rn 1 )  Bi * T
T j (rn ) 
1  Bi
Here r0  0 and rn  R  0.15 m

Note that the numerical solution is controlled by the dimensionless Fourier number, which for
stability must be small.
t
 0.5
r 2
This requirement sets a relationship between the time step and the spatial resolution. The
equations were programmed into an excel spreadsheet and the following temperature profiles
obtained (using a radius step of 0.015m and a Fourier time step of 0.4):

15
Forensic Analysis of Body
Cooling

38
36
34
32 S1

30

Time Series (550 sec)


S21

28 S41

26 S61
24
S81
22
20 S101
0.00
0.03
0.06
0.09
0.12
0.15

r (m)

The variation of the temperature with time at the center of the body is shown in the figure below.

16
Centerline Temperature

38

36

34

32
Temp (celsius)

30

28

26

24

22

20
0 4 8 12 16
time (hr)

The numerical solution shows some interesting trends, including a “lag-time” for which the core
temperature does not change substantially over the first 3-4 hours after death. The measured
temperature is achieved around 12.5 hours, which is less than the estimate, but within the
accuracy possible in reading the chart. Clearly, there are many approximations modeling the
cooling of a corpse post mortem, and the subject remains both controversial in the courts as well
as an active area of research.

17
Problem 5.6

We are given the following information:

Thickness of glass pane, L = 8 mm = 0.008 m


By industry standards, 1 R-value = 0.17611 m2K/W
Note that these units include the area. d
aRtotal = R-4 = 4  0.17611 m2.K/W
= 0.70444 m2 K/W

kGlass = 0.78 W/m2 K

Applying the resistances-in-series formula,

L d L
a Rtotal   
kGlass k Ar kGlass Argon gas
2L d
 
kGlass k Ar

Substituting values and rearranging, we obtain


d
0.70444  0.0205   d  0.68393k Ar .
k Ar

We will now determine kAr (thermal conductivity of Argon) using Chapman-Enskong theory,
which obtains the following formula for the thermal conductivity

25  mk BT ˆ
k CV
32  2  k
  12   6 
where  is a parameter in the Lennard-Jones potential, U LJ  4      
 r   r  

The collision integral,  k is given by

1.16145 0.52487 2.16178


k   
 k BT 
0.14874
 kT  kT
  exp  0.77320 B  exp  2.43787 B 
        

For Argon¶,   3.39 Å ,   117.5k B (kB is the Boltzmann constant).


Assume T = 0 ºC = 273.15 K


Panagiotopoulos et al., Fluid Phase Equil., 66, 57-75 (1991)

18
Substituting these values into the formula yields a collision integral of  k  1.12 . Also,
assuming argon to be an ideal gas at the temperature of interest,

 
Cˆ v  R  12.471 J/mol.K   
3 1
  311.775 J/kg.K
2  0.04 kg/mol 
25  mk BT ˆ
k CV =1.7x10-2 W/m.K
32  2  k
where m is the mass per atom = (0.04 kg/mol)/NA

d  0.68393k Ar  11.63 mm

19
Problem 5.7

A schematic of Fick’s funnel apparatus is shown to RL z=L


the right, with the positive z-axis in upward
direction. The model equation is derived by
performing a shell balance around a section of the
funnel with height Δz. The funnel has radius R 0 at a Δz
z=0 and RL at z=L. (Note: in Fick’s original
experiment, the funnel was arranged such that
RL>R0, which is the same as the diagram shown R0 z
here.) z=0

If we assume a linear profile for the funnel radius, then we obtain the z-dependent funnel radius
and cross-sectional area by the following expressions:

R0  RL
r ( z )  R0  z
L
R  RL 
2

a ( z )    R0  0 z
 L 

The species mass balance remains the same as in the cylindrical case, but we now have to
account for the change in cross-sectional area. Assuming steady state, the shell balance on the
funicular shell becomes

C A 1   J z a( z ) 
  0.
t a( z ) z

Substituting for a(z) using the above expression, and using Fick’s constitutive equation for the
molar flux, we obtain

d  R  RL  dC A 
2

  R0  0 z  0.
dz   L  dz 

Integrating this equation once yields

dC A A
 .
R0  RL 
2
dz 
 R0  z
 L 

Integrating again yields

AL2
C A ( z )  ...  B 
 R0  RL   R0  RL  z  R0 L 

20
We will use the same boundary conditions as in the original derivation, namely

@ z  0, C A  C Asat
.
@ z  L, C A  0

Substituting these boundary conditions into the integrated balance equation, we get

AL
C A ( L)  0  B 
R0  R0  RL 
AL
B
R0  RL  R0 

and

AL2
C A (0)  C Asat  B 
  RL L   R0  RL 
AL AL
 C Asat  
R0  RL  R0  RL  RL  R0 
R0 RL C Asat RL C Asat
 A , B
L  R0  RL 
Substituting the integration constants back into the general solution yields the concentration
profile in the funnel

R0C Asat  RL L 
CA ( z)  1  .
 R0  RL    R0  RL  z  R0 L  

b.) We can make the concentration profile general for any saturation concentration and any
funnel length by defining the following dimensionless variables, as well as the aspect ratio:

CA z R
C  sat , z  , R  L .
CA L R0

Rearranging the concentration profile, we get

 
1  R .
C ( z )  1
 
   
1  R   1  R z  1 


21
We have now non-dimensionalized the problem such that the only parameter is the aspect ratio
of the funnel. Thus, for any value of R , we can plot a solution that applies to any value of the
saturation concentration and any length of funnel. The plot below gives solutions for various
values of R . One limiting case of this result is RL→R0 (or R →1), which yields the straight line
first obtained by Fick, as seen in the figure. (Note: R =1 produces a singularity in the solution
derived here. However, as shown in the text, the obtained profile scales as C  1  z )
As the aspect ratio is increased, the profile assumes higher and higher curvature. Since the
product of the flux and the cross-sectional area must be constant, there must be a higher flux, and
thus a steeper concentration gradient, near the bottom of the funnel. This agrees with Fick’s
experiments, showing the same curved profile for diffusion in the funnel geometry.

Concentration profile for funnel geometries


1

0.9

0.8
R=1
0.7

0.6
C/Csat

R=2
0.5

0.4
R=5
0.3
R=10
0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
z/L

c.) Using Fick’s constitutive equation for the molar flux, we know that, for the funnel geometry:

dC A D C sat dC
J A, z   DAB   AB A
dz L dz

22
 
dC 1 d  R 
 1
dz 1  R dz   1  R z  1 
    
 
R  1  R 
 
 2 


1  R   1  R z  1 
    
R

 
2
 R  1 z  1
 
DAB C Asat R
 J A, z , funnel 
 
2
L  R  1 z  1

We similarly derived the molar flux for the cylindrical case, for which we derived

DAB C Asat
J A, z ,cylinder  .
L

We can compare the dissolution rate between the funnel and the cylinder by relating the total rate of
diffusion in each case:

rdissolve   aJ A, z    RL2 J A, z
z 1

 R 2 DAB C Asat
 rdissolve ,cylinder 
L
 RL DAB C Asat
2
 rdissolve , funnel 
LR

Looking at these two expressions, we relate the error in a diffusion coefficient measured in the funnel
geometry but assuming a cylindrical profile is related to the ratio of rdissolve in each case.

rfunnel 1
 .
rcylinder R

Thus, if we take measurements in a funnel but use the result derived for a cylinder, our obtained value
of DAB will be overpredicted by roughly a factor of R (i.e. crystals at the bottom of a funnel
will dissolve more slowly than crystals in a cylinder).

23
Problem 5.8

We are given the following information for the tube and the fluid:

L = 20 cm = 0.2 m
D = 3.4x10-6 cm2/s = 3.4x10-10 m2/s
k = 5.3x10-9 M/s (consumption rate of X)
a = cross-sectional area

a.) The 1-D shell balance on the tube in the z-direction is derived as follows:

az  C |t t C |t   a  J |z  J |z z  t  kazt


C J
  k
t z
z  z
C L
Use constitutive equation (Fick’s law), J   D , we obtain z
z

C  2C z
 D 2 k
t t

C  2C
Assuming steady state,  0 . So, 0 D 2 k
t t
At the bottom of the tube, we have the same situation as Fick’s original experiment, where the
crystals are in equilibrium with the liquid phase, and so

@ z  0, C  Csat .

Since the tube is sealed at both ends, there can be no diffusion of Agent X out of the top of the
tube. Setting the molar flux of Agent X at the top of the tube equal to zero yields the other
boundary conditions,

dC
@ z  L,  0 (since top is impermeable)
dz zL

24
b.) Using the model equation and the boundary conditions from the previous part, the solution
proceeds as follows:

 2C k

z 2 D
C k
  z A
z D
k 2
 C  z  z  Az  B
2D

 kL
Using boundary conditions, A  and B  Csat
D

k 2 kL
 C z  z  z  Csat
2D D

c.) Evaluating the above concentration profile at the top of the tube,

k 2 kL
C  L  L  L  Csat
2D D
.
kL2
 C  L   Csat 
2D

Plugging in the values, we obtain C(L) = 0.69 M.

d.) We can make the concentration profile derived in part b dimensionless by defining a
dimensionless coordinate and concentration:

C z
C  , z 
Csat L

 C  z   C  Csat and z  z  L

So,
k kL
C  Csat   z  L   z  L  Csat
2

2D D
2
kL kL2
 C  z 2  z  1
2 DCsat DCsat

25
We can see from this functional form that the non-uniformity in the concentration profile comes
from the value of the dimensionless quantity

kL2

DCsat

26
Problem 5.9

We are given:

Ceq eq
membrane =0.14Caqueous
DA = 6.7x10-8 cm2/s
nA,i = 10-3 mol
CA,in = 0.2 M
R = 5 mm
h = 0.2 mm

a.) Performing a level IV shell balance around the membrane, we obtain:

4 r 2 r  C |t t C |t   4  r 2 J |z  r 2 J |z z  t
C 1 J
  2
t r r

Applying Fick’s constitutive equation for the flux J, and assuming steady state, we obtain:

DA d  2 dC A 
 r 0 .
r 2 dr  dr 

Since the ODE is second order in CA, we need two boundary conditions involving CA. If we
assume that the capsule is very small compared to its surroundings, and that the medium
surrounding the capsule is well-mixed, then:

C A (r  R)  0 .

The second boundary condition is the relation we are given relating the membrane concentration
to that inside the capsule, which is

C A (r  R  h)  0.14Cin .

b.) Integrating the ODE once, we obtain

27

dC A 
 d  r
dr  
  r
2

dC A
 A 2
dr r

Integrating again, we obtain

dr
 dC r2
A  A

A
CA  B 
r

After substituting the boundary conditions and solving for the integration constants, we obtain

0.14Cin R ( R  h)
A
h
A 0.14Cin ( R  h)
B 
R h

and thus

0.14Cin ( R  h)  R 
CA    1 .
h r 

c.) Since our system is defined in radial coordinates, the flux will change with position in the
membrane. Using Fick’s law for the definition of the flux, we obtain:

dC A 0.14 DACin ( R  h) R
J A , r   DA 
dr h r2

It makes the most sense to evaluate the flux at the outside of the membrane (r=R), since this will
determine the release rate of the drug to the body.

0.14 DACin ( R  h) mol


J A, r rR
  9.0  107 2 .
Rh ms

d.) The drug should be re-administered when the contents of the capsule are depleted, or when
the capsule leaves the stomach. The time it takes to deplete the entire contents of the capsule can
be approximated using the flux calculated in part c:

28
ni ni
J A, r   t
rR
4 R 2t 4 R J A,r
2
rR

t  3.5  106 sec  41 days

(NOTE: this is not an exact calculation, since the concentration inside the capsule, and thus the
flux, changes with time)

Since this time is considerably longer than most physiological processes, it is safe to assume that
the drug should be administered after however long it takes to leave the stomach.

e.) Since the membrane of the capsule is very thin, depending on the material of construction it
could potentially burst inside the stomach (especially if it is broken down under acidic
conditions), causing a potentially toxic dose of the drug to be released to the stomach.

29
Problem 5.10

We are given:

Apool = 30m x 10m = 300m2


ρH2O = 1.0 kg/L
MH2O = 18 g/mol
Cgas = 4.0x10-2 mol/L
h = 10 cm
dz/dt = 20 μm/hr
yH2O(z=0) = 0.031
yH2O(z=h) = 0

DATA:
At STP, 1 mole of an ideal gas occupies 22.4 liters, so the concentration of air at STP is ~0.04
mol/L. 100% relative humidity at STP corresponds to a partial pressure of 3148 Pa, hence
yH2O(z=0) = 0.031

a.) The flux expression for the Arnold cell includes both the diffusive and convective fluxes:

dCH 2O
N H 2O  J H 2O  vCH 2O where J H 2O   DH 2O .
dz

The velocity v can be defined in terms of the total molar flux:

N
v where N  N H 2O  N air .
Cgas

However, since the air cannot penetrate the surface of the water, Nair(z=0) = 0, and thus if we
make the pseudo-steady state assumption, Nair=0 everywhere. Thus, substituting in for v as well
as CH2O=CgasyH2O, we obtain for the molar flux of water:

dyH 2O
N H 2O  Cgas DH 2O  Cgas yH 2O N H 2O
dz
.
Cgas DH 2O dyH 2O
 N H 2O 
1  y H 2O dz

The boundary conditions for this expression are given in the problem, such that we can integrate
the ODE directly:

30
h yH 2O ( z  h ) dyH 2O
N H 2O  dz  Cgas DH 2O 
0 yH 2O ( z  0) 1  y H 2O
.
Cgas DH 2O 1  y H 2O ( z  h ) 
 N H 2O  ln  
h 1  yH 2O ( z  0) 

We can also relate the molar flux of water to the mass loss resulting from the change in height of
the water over time:

 H 2O dz
N H 2O  .
M H 2O dt

Thus,

dz C gas M H 2O DH 2O 1  yH 2O ( z  h) 
 ln   .
dt  H 2O h 1  yH 2O ( z  0) 

b.) Using the model equation just derived to obtain the experimentally measured value of D H2O,
we obtain:

dz 1
 H 2O h  1  y H O ( z  h)  
dt  ln 
DH 2O  2

Cgas M H 2O  1  yH O ( z  0)  
  2 
m2
DH 2O  2.5  105
s

c.) The value of DH2O determined by the Arnold cell experiment is within a few percent of the
accepted literature value.

d.) Assuming that the only mass loss from the pool is by diffusion of water away from the
surface of the pool, then the change in height of the pool will be dz/dt evaluated under the
conditions specified for the pool (yH2O = 0.5*0.031 = 0.0155 when h=1m):

dz Cgas M H 2O DH 2O 1  0.0155 
 ln 
dt  H 2O h  1  0.031 
dz
 3.67  1010 m/s
dt

Thus, the flow rate of water that must be supplied to the pool to keep the water at a constant
height is

31
dz
qH 2O  Apool  1.1 107 m3 /s  9.4 L/day
dt

e.) The estimated flow rate of water required is far too low than what is actually required, for
two significant reasons. First, and most importantly, any convective mass transfer over the
surface of the pool is neglected (which is only accurate if Joe builds a very tall retaining wall
around his pool!). Also, since the surface of the pool is most likely not flat, the effective area for
mass transfer at the surface of the pool will be much greater than that used here.

32
Problem 5.11

This is an alternative method for measuring the thermal diffusivity (and hence, the thermal
conductivity) by measurement of the phase shift in a driven system. The geometry is an
insulated bar (we can assume it is infinitely long for this problem) in the y-direction. The
analysis follows from equation 5.15, Fourier’s “second law”:
T  2T
 2
t y
The initial and boundary conditions are:
t  0, y, T  T0
y  0, t  0, T  T0  A sin(t )
y  , t  0, T  T0
However, we are concerned with the steady, time periodic solution. This is found by the method
of separation of variables, where:
T  y, t   T0    B ( y )ieit 
Where () denotes the real part of the function. Substitution into the governing equation yields:
 d 2 B ( y ) it 

 B ( y )( )e it
    2
ie 
 dy 
d 2 B ( y ) i
  B( y )
dy 2 
i i
y  y
 B ( y )  C1e 
 C2 e 

Application of the first boundary condition insures that C1  0 . Application of the second
1
boundary condition shows that C2  A . Using the identity i   1  i  , the final solution
2
becomes:


y   
T ( y, t )  T0  Ae 2 Sin  t  y 
 2 
Now, using two thermistors, separated by a distance L, we can see that the phase shift between

the amplitude peaks in the sinusoidally varying temperature profile will be given by: L.
2
Thus, an accurate measurement of the phase shift in the temperature signal, the distance between
the thermistors, and the driving frequency, yields a direct measurement of the thermal diffusivity.

Note that this problem analogous to understanding temperature profiles in soils, for example,
where a periodic temperature is applied to the surface of the soil.

33
Problem 5.12

a.) A diagram of the release of Taxol from the cylindrical water


wormlike micelles is given to the right, where phase I is the
micelle core, phase II is the micelle shell, and phase III is the shell
bulk water phase. Since we are given the concentration of
Taxol in weight percent, we will express the concentration of core
r2
Taxol (A) in terms of weight percent, wA, in each phase,
which will vary both with the radial position and with time. r1
wAI
We are given r1=11 nm, r2=10 nm. Since the volume of
micelles is on the order of 1% of the volume of water, if we wAII JA,r
assume the water phase to be well-mixed, then it is safe to
assume that wA,  0 .
III

wAIII,
b.) If the Biot number for mass transfer between the core and its surroundings (shell and water)
is small, then the concentration inside the core will be uniform, and we can lump the mass
transfer from the shell to the water into an overall mass transfer coefficient, Km. Furthermore, if
we assume that the water is well-mixed, then resistance to mass transfer outside the micelle will
be negligible such that wAIII (r )  wAIII,  0 .

Writing a level IV balance on the micelle phase using the lumped overall rate of mass transfer,
we obtain


 IV I wAI
t t
 wAI   K a  w  Mw  t
t
I
m
I III
A, 
I
A


( r12 L) wAI
t t
 w    K (2 r L) Mw t
I
A t m 1
I
A

wAI 2MKm
   wAI , where  
t r1

If we assume that the Taxol is initially contained entirely within the micelle core at a uniform
concentration, then the initial condition for the above ODE is

 g micelle   gA  4
wAI (t  0)  wAI ,0   0.01    0.045   4.5 10 .
 g solution   g micelle 

Integrating the model equation, we obtain


 wI 
ln  IA     t
w 
 A,0 

34
where the bracketed term represents the fraction of Taxol remaining in the micelle. The data
provided is given in terms of the fraction of Taxol released from the micelles, f. However, we
wAI ,0  wAI
know that f  , and thus,
wAI ,0
ln 1  f     t .

c.) The above model equation for percent release of the Taxol versus time is in linear form,
where y=ln(1-f), x=t, and the slope=-β. Converting the experimental data to this form yields the
linearized data in the table below:

t (days) t (sec) x 104 f ln(1-f)


0.00 0 0.00 0
1.00 9 0.17 -0.186
2.00 17 0.32 -0.387
3.00 26 0.37 -0.462
6.00 52 0.51 -0.713
11.00 95 0.59 -0.891

The plot below shows the linearized experimental data along with the linearized fit, yielding a
value of β = 1.1x10-6 s-1. This corresponds to Km=1.22x10-12. If this resistance is solely due to
diffusion through the shell of the micelles (1/Km = L/DAB), the diffusivity corresponding to this
value of Km would be on the order of 10-20 m2/s! This would be a very low diffusivity even for a
dense solid, so is fairly unreasonable. Also, the fit of the model to the experimental data is very
poor, suggesting that a lumped analysis that considers only diffusion of Taxol through the
micelle shell is insufficient to explain the release behavior of the wormlike micelles.

-0.2
y = -1.112E-06x
2
-0.4 R = 7.744E-01
ln (1-f)

-0.6

-0.8

-1
0 200000 400000 600000 800000 1000000

time (s)

d.) Since we derived the model equation assuming an overall rate of mass transfer between the
micelle core and the water, the model equation will not change, since we lumped both the
resistance from the micelle shell and the resistance from the water into Km, such that:

35
1
Km 
(r2  r1 ) M

DAB km

where km is the mass transfer coefficient in the water. Thus, using an external resistance in
the water phase will not provide a better fit to the experimental data. However, it may give
us a diffusivity in the micelle shell that is more reasonable.

e.) If we now relax the assumption of small Biot number for mass transfer, we must derive a
new model equation using a shell balance around a differential shell control volume in the
micelle.

wA
 Vw A t t 
 VwA t   ajA,r r  ajA,r r r
 t, where jA,r    DAB
r

2 rLr   wA t t   wA t   2 L ajA,r r  ajA,r r r
 t
wA DAB   wA 
  r 
t r r  r 

Since we are interested in the time-dependent release of Taxol from the micelles, wA will depend
on both time and radial position. We must also specify boundary conditions for each phase
(core, shell). Since the model PDE is first order in time, we need an initial condition, which, if
we assume a uniform concentration of Taxol in the core of the micelle and nowhere else, yields

 wA,0 , 0  r  r1 
 
wA (t  0)   0 , r1  r  r2  .
 0,rr 
 2 

Since the model PDE is second order in radial position, we also need two boundary conditions
for each phase. At the center of the micelle, we have symmetry such that:

wA
(1) 0.
r r  0,t

At the core-shell boundary, we need two boundary conditions, one for each side of the boundary.
These are given by assuming local equilibrium at the core-shell boundary as well as equality of
flux across the boundary in each phase (NOTE: we must leave the densities in the flux
expressions because they are different for each phase):

(2) wAI (r  r1 , t )  M 1wAII (r  r1 , t )

36
wAI wAII
(3)  I DAB
I
  II DAB
II
.
r r  r1 ,t
r r  r1 ,t

where M1 is the partition coefficient between the core material and the shell material. We need
one remaining boundary condition for the shell-water interface, where we will assume some
resistance to mass transfer in the water, such that,

(4)  II DAB
II wAII
r r  r1 ,t

  III km wAIII,  M 2 wAII
r  r2 ,t .
Where M2 is the partition coefficient between the shell material and the water. These boundary
conditions are sufficient to solve the model equation for the concentration profile, as well as the
release rate, of Taxol in the micelle over time. The total amount of Taxol released would then be
the integral of the rate of mass transfer (e.g. from BC 4) from the micelle to the water over time.

37
Problem 5.13

a.) We are now asked to solve the murder given in Problem 5.5 using the short-time penetration
solution, which is valid when the Biot number is very high and external resistances are
negligible, such that the surface temperature of the body will be the same as the surface
temperature in the room. The penetration solution is given as (5.8.3)

T ( y, t )  T0  y 
 1  erf  .
T1  T0  2 t 

where y is the depth into the medium (in this case, y=(15 cm)-r), T0 is the initial body
temperature (37°C) and T1 is the temperature of the room (20°C). We want to find the time at
which T(15 cm,t) = 29°C. Rearranging the penetration solution, we get,

 y  29C  37C
erf    1  20C  37C  0.47 .
 2 t 

Using figure 5.16 in the text, the dimensionless time corresponding to this dimensionless
temperature is approximately 0.5. Thus,

y
 0.5
2 t
y2 (0.15 m) 2
t  
4(0.5) 2  4(0.5) 2  (1.56 107 m 2 / s)
 t  1.4  105 s  40 hours
We find that this estimate is nearly three times that obtained by the more accurate methods
discussed in problem 5.5. This is not surprisingly, as the temperature profile has penetrated
completely into the cylinder well before this time and hence, the penetration solution
overestimates the time required to cool.

b.) We will now use the lumped analysis method in order to solve the crime. If the Biot number
for the dead body is small, we can assume the temperature throughout the body is constant, and
can therefore solve for the body temperature over time by a level III balance on the dead body

 Cˆ p R 2 L T |t t T |t   2 RLh T  T  t


T 2h
  T  T 
t  Cˆ p R

T  T 2h
If we define   and   , we obtain
T0  T  Cˆ p R

38

  
t

whose solution with the initial condition  (0) = 1 is

 (t )  e   t .

We wish to compute the time such that  (t)=0.5, as in the previous part. We first need to
calculate β,

2h 2(8 [W/m 2 K])


 
 Cˆ p R (1100 [kg/m ])(3.5 [J/g K])(1000 [g/kg])(0.15 [m])
3

  2.77  105 s -1

Finally, solving for t, we obtain,

ln( ) ln(0.47)
t  5 -1
 2.7 x104 s  7.6 hours .
 2.77 10 s

c.) If we had used the penetration solution, we would have incorrectly deduced that Howard
committed the murder. The short-time penetration solution assumes that the body is much larger
than the depth of the temperature change - or in other words, sufficiently far from the surface, the
temperature is always at 37°C. This is obviously not the case, so the penetration solution cannot
be trusted. On the other hand, if we had used the result of the lumped analysis, which is valid
only for small Biot numbers, we would have correctly implicated, who could have committed the
crime in the past 30 hours. Clearly, the accurate numerical analysis lies between the two
extremes given by the short-time penetration solution and the lumped analysis.

39
CHAPTER 6 SOLUTIONS

Problem 6.1

(a) We begin by carrying out material balance on a thin spherical shell


(of thickness Δr ) inside the catalyst pellet (assuming steady state)

N Ar 4πr 2 − N Ar 4πr 2 + rA 4πr 2 Δr = 0


r r + Δr

where N Ar is the molar flux of A and r A is the rate of generation of A.


Dividing throughout by 4πΔr and taking the limit as Δr → 0 , we get

(
d N Ar r 2 )
− r 2 rA = 0
dr

Now,
dC A
N Ar = x A ( N Ar + N Br ) − De .
dr

We assume equimolar counter-diffusion, i.e. N Ar = − N Br . This implies that the number of moles
of A crossing the thin shell is equal to that of B crossing the shell but the two components flow
in opposite directions. We get

⎛ dC A ⎞
d⎜r 2 ⎟
⎝ dr ⎠
De − r 2 rA = 0
dr

Assuming the reaction to be first order, rA = −kC A

d 2C A dC A
r 2 De + 2rDe − kC A r 2 = 0
dr 2 dr

Dividing by r2D e throughout,

d 2C A 2 dC A kC A
+ − =0 …(1)
dr 2 r dr De

Boundary conditions:
dC A
(i) At r = 0, =0
dr
(ii) At r = R p , C A = C As
where R p is the radius of the pellet and C As is the concentration at the pellet surface.

We now make equation (1) dimensionless by defining dimensionless concentration and length.

CA r
C *A = , r* = ⇒ C A = C A* C As , r = r * R p
C As Rp

Plugging these in (1),

C As d 2C A* 2C As dC A* kC As C A*
+ − =0
R p2 dr *2 r * R p2 dr * De
2 *
d 2C * 2 dC * kR C
⇒ *2A + * *A − p A = 0
dr r dr De
d 2C A* 2 dC A*
⇒ + − ϕ 2C A* = 0 ...(2)
dr *2 r * dr *

kR p2
The dimensionless parameter ϕ = is the Thiele modulus.
De
Boundary conditions in dimensionless form are:

dC *A
(i) At r = 0, =0
dr *
(ii) At r = R p , C *A = 1

By making the substitution p = C A* r * in (2), we get

d2 p
*2
−ϕ2 p = 0
dr

which can be easily solved using the boundary conditions above to give

1 ⎛ sinh ϕ r * ⎞
C A* = ⎜ ⎟
r * ⎝ sinh ϕ ⎠

(b) Since the reactor is a CSTR, the usual model equation for a CSTR would apply. So, at steady
state,
q A0 ( C Ab 0 − C Ab ) + rAV = 0
rAV
⇒ C Ab = C Ab 0 +
q A0

where rA = −kC A and C Ab is the concentration in bulk (as opposed to the concentration inside the
catalyst pellet, C A ).

Let us now define two effectiveness factors (see “Elements of Chemical Reaction Engineering”,
H.S. Fogler, pp. 757 for more details)

Actual overall rate


η=
Rate if the entire catalyst surface were exposed
to the external surface concentration CAs

Actual overall rate


Ω=
Rate if the entire catalyst surface is exposed to the bulk concentration CAb

η is known as the internal effectiveness factor and Ω is known as the overall effectiveness
factor. The following steps are involved in the transport and reaction of A:

‰ Transport of A from the bulk solution to the catalyst surface. This is external mass
transfer.
‰ Transport of A inside the catalyst pores. This is internal mass transfer.
‰ Reaction of A on and inside the catalyst.

If we assume that the resistance to external mass transfer is negligible, we will have C As = C Ab .
3
So, Ω = η = 2 (ϕ coth ϕ − 1) (see “Elements of Chemical Reaction Engineering”, H.S. Fogler,
ϕ
pp. 749). Using the overall effectiveness factor, Ω , we can write the actual rate r A in terms of
the rate (r Ab ) that we would have if the entire catalyst surface were exposed to the bulk
concentration C Ab . So,

rA = ΩrAb = η rAb (assuming negligible external mass transfer resistance)


= −η kC Ab

η kC AbV
C Ab = C Ab 0 −
q AF
So, −1
⎡ η kV ⎤
⇒ C Ab = C Ab 0 ⎢1 + ⎥
⎣ q AF ⎦
−1
C − C Ab ⎡ η kV ⎤ η kV
Conversion, X A = Ab 0 = 1 − ⎢1 + ⎥ =
C Ab 0 ⎣ q AF ⎦ q AF + η kV

kR 2
(c) Thiele modulus, ϕ = . Plugging in the values, we get
De

ϕ = 3.101

(d) The expression for conversion (as calculated in part (b)) is

−1
C − C Ab ⎡ η kV ⎤ η kV
X A = Ab 0 = 1 − ⎢1 + ⎥ =
C Ab 0 ⎣ q AF ⎦ q AF + η kV

3 3
Also, η = (ϕ coth ϕ − 1) = ( 3.101coth(3.101) − 1) = 0.6594
ϕ 2
3.1012

So, X AL = 0.6128

(e)

‰ d p = 2 mm: So, d p = 0.002 m, R p = 0.001 m.

3
New Thiele modulus, ϕ = 1.24 ⇒ η = (1.24 coth(1.24) − 1) = 0.9102
1.242

New conversion, X AL = 0.6860

‰ d p = 10 mm: So, d p = 0.01 m, R p = 0.005 m.

3
∴ ϕ = 6.2017 ⇒ η = ( 6.2017 coth(6.2017) − 1) = 0.4057
6.2017 2

New conversion, X AL = 0.4933


The plot above shows the variation of conversion with pellet radius.
Thus, we conclude that increasing the pellet size leads to a decrease in the conversion achieved
and if we go below a certain pellet size, the conversion stays constant with any further decrease
in size.
Problem 6.2

a) We are given the heat flux and temperature difference for the absence of external coolant, so
we can calculate the heat transfer coefficient using Newton’s law of cooling.

Q = h(Tin − Tw )
W
10000
h=
Q
= m2
(Tin − Tw ) (1100 − 1000)C
W
h = 100
m2 K

For the new system (with cooling), the h will remain the same, but the temperature difference
will change. We can easily calculate the new value of Q, the heat duty of coolant required.

Q = h(Tin − Tw )
W
Q = 100 (1100 − 600)C
m2 K
kW
Q = 50 2
m

b) We need to look to dimensionless calculations to see the effect these changes will make.

VLρ
Re =
μ

If we increase the exhaust velocity, V, by 2, and decrease the length, L, by 2, our Reynolds
number will remain the same (aka, “rocket performance will be maintained”). To calculate the
heat duty, however, we need to see how the heat transfer coefficient (via the Nusselt number)
will be affected by these changes. Since Prandtl number is a function only of material properties
(so will remain constant) and we have shown that Reynolds number will remain constant through
these changes, we realize the Nusselt number must remain the same.

hL
Nu =
k

We double L, so the heat transfer coefficient must be cut in half for the Nusselt to remain the
same. So the new heat duty becomes:
Q = h(Tin − Tw )
W
Q = 50 (1100 − 600)C
m2K
kW
Q = 25 2
m
Problem 6.3

(a) The following information is provided (sim = non-radioactive simulant, rad = radioactive
material):

α sim = α rad
μ sim = 2μ rad
k sim = 2k rad
Ĉ p,sim = 2Ĉ p,rad
ρ sim = ρ rad

A shell-and-tube heat exchanger is used to cool the radioactive fluid. We are to determine the
heat transfer coefficient that the radioactive material (h rad ) would have (Note: typo in textbook) if
h sim = 100 W/m2-K, Re sim = 50,000, and flow is kept at the same rate.

ρVD
Re =
μ

Therefore,

ρ simVsim Dsim ρ V D
Re sim = and Re rad = rad rad rad .
μ sim μ rad

From the relationships given above and V sim = V rad (flow stays the same) and D sim = D rad (same
pipe), we get:

2 ρ simVsim Dsim
Re rad = = 2 Re sim = 100,000
μ sim

Cˆ p μ
We also know that Pr = .
k

Therefore,

⎛1 ˆ ⎞⎛ 1 ⎞
⎜ C p , sim ⎟⎜ μ sim ⎟
Cˆ p , sim μ sim Cˆ p ,rad μ rad 2 ⎠⎝ 2 ⎠ = 1 Pr .
Prsim = and Prrad = =⎝
⎛1 ⎞
sim
k sim k rad 2
⎜ sim ⎟
k
⎝2 ⎠
Using the Colburn analogy:

hD
Nu = 0.023 Re 0.8 Pr 0.333 =
k

Therefore,

0.8 0.333
Nu rad 0.023 Re rad Prrad
= = 2 0.8 (0.5) 0.333 = 1.382
Nu sim 0.023 Re sim 0.8 Prsim 0.333

Likewise,

⎛h ⎞
⎞ ⎜ rad 1 ⎟
Nu rad
⎛ hrad Drad



k rad ⎠ ⎝ ⎜
2
( )k ⎟ 2h
sim ⎠
= = = rad
Nu sim ⎛ hsim Dsim ⎞ ⎛ hsim ⎞ hsim
⎜ ⎟ ⎜ ⎟
⎝ k sim ⎠ ⎝ k sim ⎠

Putting them together gives

hrad =
(
1.382hsim 1.382 100W / m 2 − K
=
)
2 2

h rad = 69.1 W/m2-K

(b) We desire to scale up the prototype without changing the tube side pressure drop. Scale up
implies that the total flow rate is to be increased. The constraints are:
1) fixed pressure drop
2) same temperature change on the process fluid.
Mathematically, the problem becomes finding solutions of the following:

L 32
ΔP = 2 f ρV 2 = 2 5 f ρ Lq12
D π D
ˆ (T − T ) = Ua ΔT
Qload = ρ1q1Cp ( 3.52 )
1 1 1F 21

Nu = 0.023Re0.8 Pr 0.4 (6.33)


plus the relations for U (such as 5.4.12) and Table 3.5.

As a first trial, we might consider trying to hold the velocity fixed and then, increasing the total
flow rate will also require increasing the diameter. This also means that Re and Nu will change.
The increase in area required will also require a change in length to satisfy the above equations.
However, we quickly realize that changing the diameter and length of the tube independently is
not possible because of the pressure drop relationship. Thus, the problem is not likely to yield a
meaningful solution if we restrict ourselves to the use of a single pipe.

Therefore, a more reasonable strategy is to employ a multiple tube shell and tube exchanger. As
a first iteration, would be constructed of tubes of the same diameter, length, thickness and
material as in the prototype. Holding the same velocity in the tubes as in the prototypes means
that scale up is achieved by increasing the number of tubes to accommodate the increase in total
flow rate. Considerations such as those discussed in problem 3.20 would enable a technical
feasible design and specification of the shell side utility flow to maintain the required heat duty
on the process stream. This procedure has the additional advantage that the behavior of the
radioactive fluid in the process scale exchanger should be similar to that in the prototype as the
Re and Nu are held constant in the scale up procedure.
Problem 6.4

This problem is similar to that given in Example 6.7 except that we are asked to use penetrtation
theory. Assuming the boundary layer is thin relative to the size of the bubble, we can simplify
to account for velocity only along the surface of the bubble.

From the penetration theory, we know that (example 6.6)

DAVmax
km = 2
πL

We are given that D O2/water = 2.1 x 10-9 m2/s and D O2/air = 2 x 10-5 m2/s. For a calculation, we will
follow the example in 6.6 and take a 5mm bubble with a rise velocity of 0.235 m/s.
Therefore (by the penetration model),

2.1× 10−9 m 2 / s *0.235m / s


kmI = 2 = 3.5 × 10−3 m / s
π *0.005m
DO2 / N 2 2 × 10−5 m 2 / s
kmII kmI * = 3.5 × 10−3 m / s * = 0.34m / s
DO2 / water 2.1× 10−9 m 2 / s

(note, the approximate sign in the second equation is because this form of the penetration theory
is derived for mass transfer outside of the bubble, and therefore, this should only be considered
as an estimate).

We can see from these results that k m I << k m II. Thus, the dominant resistance to mass transfer is
clearly on the liquid side (phase I).
Problem 6.5

We are given the following information:

g
μ H O = 0.65cp = 0.0065
2
cm ⋅ s
PrH 2O = 4.3
W
k H 2O = 0.63
mK
D = 0.05m
L 1m 3 1 min
(10 )( )( )
q min 1000 L 60 sec = 0.085 m
v= =
Ac π (.025m) 2 s
J
C p H O = 4.18
2
gK
g
ρ H O = 990
2
L
−4 m2 K
R foul ,in = 1.7 ⋅ 10
W
m2 K
R foul ,out = 2.3 ⋅ 10 −4
W
Tin = 15C
Tout = 75C

(NOTE: values of constants for water obtained from


Deen, Analysis of Transport Phenomena 1998)

a.) The first order of business is to determine the “appropriate flow conditions” by calculating
whether we are in the laminar or turbulent flow regime.

g m 1000 L
(990 )(0.05m)(0.085 )( )
ρDv L s 1 m 3
Re = = = 6473
μ (0.0065
g
)(
100cm
)
cm ⋅ s 1m
Since Re is greater than 2100, we can assume turbulent flow, and the two correlations we can use
to calculate the heat transfer coefficient in the air are the Dittus-Boelter and Colburn
correlations.

The Dittus-Boelter correlation gives us


hH 2O D
Nu = = 0.023Re0.8 Pr 0.4
k H 2O
0.023Re0.8 Pr 0.4 k H 2O
hH 2O =
D
W
0.023(6473)0.8 (4.3)0.4 (0.63 )
hH 2O = mK
0.05m
W
hH 2O , DB = 581 2
m K

whereas the Colburn correlation yields

hH 2O D
Nu = = 0.023Re0.8 Pr1/ 3
k H 2O
0.023Re0.8 Pr1/ 3 k H 2O
hH 2O =
D
W
0.023(6473)0.8 (4.3)1/ 3 (0.63 )
hH 2O = mK
0.05m
W
hH 2O ,C = 527 2
m K

Now we can calculate the overall heat transfer coefficients using these two correlations by
summing the resistances in series. Again, for the Dittus-Boelter correlation, we obtain

1
U=
1
( + R foul ,in + R foul ,out )
hH 2O
1
U DB = 2 2
1 −4 m K −4 m K
( + 1.7 ⋅ 10 + 2.3 ⋅ 10 )
W W W
581 2
m K
W
U DB = 471 2
m K

and for the Colburn correlation we obtain


1
UC = 2 2
1 −4 m K −4 m K
( + 1.7 ⋅ 10 + 2.3 ⋅ 10 )
W W W
527
m2 K
W
U C = 435 2
m K

b.) Depending on the correlation one chooses to use in calculating a heat transfer coefficient,
there seems to be a variation of ~10%. Therefore, the uncertainty of the calculation should be
~10%.

c.) To calculate the pipe length, we need to perform an energy balance around a differential slice
(Δz) of the pipe.

ρ C pT (π R 2 Δz ) t +Δt = ρ C pT (π R 2 Δz ) t + q ρ C pT Δt z − q ρ C pT Δt z +Δz + U (2π RΔz )(Tsteam − T z )Δt


∂T ∂T
π R2 ρC p = −q ρ C p + U 2π R (Tsteam − T )
∂t ∂z
∂T ∂T 2U
ρC p = −v ρ C p + (Tsteam − T )
∂t ∂z R

Assuming steady state,

dT 2U
0 = −v ρ C p + (Tsteam − T )
dz R
dT 2U
vρC p = (Tsteam − T )
dz R
dT d (Tsteam − T ) 2U
=− = (Tsteam − T )
dz dz Rv ρ C p

Now we can separate and integrate:

d (Tsteam − T )
Tout L
2U
− ∫
Tin
=∫
(Tsteam − T ) 0 RvρC p
dz

(Tsteam − Tin ) 2UL


ln[ ]=
(Tsteam − Tout ) RvρC p

Solving for L, we obtain

RvρC p (Tsteam − Tin )


L= ln[ ]
2U (Tsteam − Tout )
m g J 1000 L
(0.025m)(0.085 )(990 )(4.18 )( )
s L gK 1m 3 (100 − 15)
L= ln[ ]
W (100 − 75)
2(435 2 )
m K
L = 12.4m

Note: This problem can be solved without resorting to shell balance. We can instead use the
concept of heat load and log-mean temperature difference to get

Qload = ρ qC p (Tout − Tin ) = UaΔTlm


⎡(Tsteam − Tin ) − (Tsteam − Tout ) ⎤⎦
⇒ ρ qC p ⎡⎣(Tsteam − Tin ) − (Tsteam − Tout ) ⎤⎦ = Ua ⎣
⎡ (T −T ) ⎤
ln ⎢ steam in ⎥
⎣ (Tsteam − Tout ) ⎦
ρπ R 2vC p 2π RL
⇒ = (Qa = 2π RL, q = π R 2v)
U ⎡ (T −T ) ⎤
ln ⎢ steam in ⎥
⎣ (Tsteam − Tout ) ⎦

ρ vC p R ⎡ (T −T ) ⎤
⇒ L= ln ⎢ steam in ⎥ = 12.4m
2U ⎣ (Tsteam − Tout ) ⎦

d.) Now we need to make pipe length, L, a function of pipe radius, R. We can also see that
velocity and overall heat transfer coefficient are a function of pipe radius.

Rv( R) ρ C p ⎡ (T −T ) ⎤
L( R) = ln ⎢ steam in ⎥
2U ( R) ⎣ (Tsteam − Tout ) ⎦
q
v( R) =
π R2
1
U ( R) =
⎛ 1 ⎞
⎜⎜ + R foul ,in + R foul ,out ⎟

⎝ hH 2O ( R) ⎠

We must calculate h(R) using the correlation, using one that applies for laminar flow when
Re ≤ 2100. Notice that for laminar flow, the equation to determine h is coupled with the equation
to determine L and we will need a non-linear solver to get L.

A plot of L against R is shown in figure 1. For the laminar region (Re ≤ 2100), the Bennett-
Myers correlation was used to get h. For the turbulent region (Re > 10000), Dittus-Boelter
correlation was used to obtain h. For the transition region (2100 < Re ≤ 10000), the linear
interpolation of the form given below was used to obtain h.

⎛ Retur − Re ⎞ ⎛ Re− Relam ⎞


htr = ⎜ ⎟ hlam + ⎜ ⎟ htur
⎝ Retur − Relam ⎠ ⎝ Retur − Relam ⎠
where
Retur = 10000, Relam = 2100, Re = Reynold's number for a given pipe diameter
hlam = Heat transfer coefficient calculated using Bennett-Myers correlation
htur = Heat transfer coefficient calculated using Dittus-Boelter correlation

Figure 1: Length of the tube required to carry out the desired operation.

e.) From the definition of friction factor, we have

2Lf ρ v 2
ΔP =
D

16
and for laminar flow, f = .
Re
0.0791
For turbulent flow, we can use the Blasius equation f = for “hydraulically smooth” pipe.
Re1/ 4
Since the pipe in this problem has a roughness associated with it, we need to take that into
account. Roughness, ε = 0.026 mm . Relative roughness is defined as ε / D . So, to get the exact
friction factor in the turbulent flow regime, its value has to be read from a plot of friction factor
against Reynolds number at the correct value of ε / D (for example, see Fig. 6.2-2, Transport
Phenomena, R.B. Bird, W.E. Stewart and E.N. Lightfoot, 1960). For simplicity, here we will just
use Blasius equation to obtain the value of friction factor (as it gives us a functional form for f ).

In the transition region (2100 < Re ≤ 10000), we will calculate the friction factor f using the
following formula:

⎛ Retur − Re ⎞ ⎛ Re− Relam ⎞


ftr = ⎜ ⎟ flam + ⎜ ⎟ ftur
⎝ Retur − Relam ⎠ ⎝ Retur − Relam ⎠
where
Retur = 10000, Relam = 2100, Re = Reynold's number for a given pipe diameter
flam = Friction factor calculated for laminar flow
ftur = Friction factor calculated using Blasius equation

Figure 2: Pressure drop across the tube.


Problem 6.6

a) We are given a pipe length and left to calculate the overall heat transfer coefficient, U, using
the pressure drop through the pipe.
g
μ H 2O = 0.65cp = 0.0065
cm ⋅ s
PrH 2O = 4.3
W
k H 2O = 0.63
mK
D = 0.05m
L 1m 3 1 min
(10 )( )( )
v=
q
= min 1000 L 60 sec = 0.085 m
Ac π (.025m) 2 s
J
C p H O = 4.18
2
gK
g
ρ H O = 990
2
L
m2 K
R foul ,in = 1.7 ⋅ 10 − 4
W
m2 K
R foul ,out = 2.3 ⋅ 10 −4
W
Tin = 15C
Tout = 75C
L = 5m

To calculate the friction factor, we can look in Perry’s Chemical Engineers’ Handbook. The 7th
Edition contains information in pages 6-9 and 6-10. We find that we should use the Blasius
equation, which applies for systems having a Reynolds number between 4000 and 100000. We
first confirm that the roughness is not significant in this calculation by calculating the ratio of
ε/D.
ε 0.026mm
= = 5.2 ⋅ 10 − 4
D 50mm

An ε/D ratio this low will only affect the friction factor at Reynolds numbers greater than
~80000 (see figure 6-9 in Perry’s) so we can use the straightforward equation (6-37 in Perry’s).

f = 0.079 Re −1 / 4
f = 0.079(6473) −1 / 4
f = 0.0088

Now we can use the Chilton-Colburn analogy (6.2.22) to calculate h.


f h
jH = = (Pr) 2 / 3
2 ρC pV
fρC pV
h=
2(Pr) 2 / 3
g J m 1000 L
(0.0088)(990 )(4.18 )(0.085 )( )
L gK s 1m 3
h=
2(4.3) 2 / 3
W
h = 585 2
m K

From here we can follow a similar calculation as in problem 6.5a to obtain the overall heat
transfer coefficient.

1
U=
1
( + R foul ,in + R foul ,out )
hin
1
U=
1 m2K m2 K
( + 1.7 ⋅ 10 − 4 + 2.3 ⋅ 10 − 4 )
W W W
585 2
m K
W
U = 474 2
m K

b) We can return to the equation we derived in problem 6.5c, since we now know the length and
use this to calculate the steam temperature.

RvρC p (Tsteam − Tin )


L= ln[ ]
2U (Tsteam − Tout )
m g J 1000 L
(0.025m)(0.085 )(990 )(4.18 )( )
s L gK 1m 3 (T − 15)
5m = ln[ steam ]
W (Tsteam − 75)
2(474 2 )
m K
Tsteam = 159C
Problem 6.7

The heat transfer correlations given in problem 6.5 can all be expressed as follows (including the Bennett-
Myers correlation upon rearrangement)

e
⎛D⎞ ⎛μ ⎞
d

Nu = a Re Pr ⎜ ⎟ ⎜ b ⎟
b c

⎝ L ⎠ ⎝ μs ⎠

where the coefficients and exponents are summarized in the table below:

Correlation Valid for a b c d e


Bennett-Myers Re < 2100 1.64 1/3 1/3 1/3 -
Sieder-Tate Re < 2100 1.86 1/3 1/3 1/3 0.14
Dittus-Boelter Re > 10,000 0.023 4/5 2/5 - -
Colburn Re > 10,000 0.023 4/5 1/3 - -

The Colburn j-factors for heat and mass transfer are given by

Nu
jH =
Re Pr1/ 3
Sh
jD =
ReSc1/ 3

If we use the Chilton-Colburn relation, j H =j D , and thus

Nu Sh
1/ 3
= 1/ 3
Pr Sc
NuSc1/ 3
⇒ Sh =
Pr1/ 3

Rearranging this expression and substituting the general correlation above for Nu, we obtain a general
correlation for the Sherwood number,

e' f'
⎛D⎞ ⎛μ ⎞
Sh = a ' Re Sc Pr ⎜ ⎟ ⎜ b ⎟
b' c' d'

⎝ L ⎠ ⎝ μs ⎠

where, when we substitute in for Nu in the Chilton-Colburn relation, we obtain the following coefficients
and exponents for the individual correlations

Correlation Valid for a' b' c' d' e' f'


Bennett-Myers Re < 2100 1.64 1/3 1/3 - 1/3 -
Sieder-Tate Re < 2100 1.86 1/3 1/3 - 1/3 0.14
Dittus-Boelter Re > 10,000 0.023 4/5 1/3 0.067 - -
Colburn Re > 10,000 0.023 4/5 1/3 - - -
The mass transfer correlation arising from the Colburn correlation is plotted below for Reynolds number
between 1000 and 100,000 (with Sc=1). Obviously, the correlation is not valid over this entire range.
However, extending the range gives an idea of what the mass transfer coefficient may be in the transition
region between laminar and turbulent flow. Note that this plot is the same whether we are considering
mass transfer or heat transfer, as long as Pr=Sc=1. Thus, we have reduced the Chilton-Colburn analogy
to the Reynolds analogy for this special case.

3
Colburn correlation for mass transfer using Chilton-Colburn analogy
10

2
10
Sh

0.8

1
10

0
10
3 4 5
10 10 10
Re
Problem 6.8

We are given the correlation:


2
−0.2
⎛ D ⎞ ⎛ ( D − Di ) ρ v ⎞
0.5 −
hi ⎛ Cpμ ⎞ 3
= 0.023 ⎜ o ⎟ ⎜ o ⎟ ⎜ ⎟
Cp ρv ⎝ Di ⎠ ⎝ μ ⎠ ⎝ k ⎠
2
−0.2
⎛ Do ⎞ ⎛ ( Do − Di ) ρ v ⎞
0.5
⎛ Cpμ ⎞3
hi
⇒ ⎜ ⎟ = 0.023 ⎜ ⎟ ⎜ ⎟
Cp ρv ⎝ k ⎠ ⎝ Di ⎠ ⎝ μ ⎠
−0.2
⎛ D ⎞ ⎛ ( D − D ) ρv ⎞
0.5
Nu 2
⇒ ( Pr ) 3 = 0.023 ⎜ o ⎟ ⎜ o i ⎟
RePr ⎝ Di ⎠ ⎝ μ ⎠

and the following information:

Di = 2 in. = 0.0508 m, Do = 3 in. = 0.0762 m


ρ = 0.0807 lb m /ft 3 = 1.294 kg/m3 , μ = 1.176 ×10−5 lb m /ft.s = 1.752 × 10−5 kg/m.s
v = 31 ft/s = 9.4488 m/s

Using Chilton-Colburn analogy,

−0.2
⎛ D ⎞ ⎛ ( D − D ) ρv ⎞
0.5
Sh 2
Nu 2
( Sc ) 3 = ( Pr ) 3 = 0.023 ⎜ o ⎟ ⎜ o i ⎟
ReSc RePr ⎝ Di ⎠ ⎝ μ ⎠
2
−0.2
⎛ Do ⎞ ⎛ ( Do − Di ) ρ v ⎞
0.5
K ⎛ μ ⎞3
⇒ m⎜ ⎟ = 0.023 ⎜ ⎟ ⎜ ⎟
v ⎝ ρ DAB ⎠ ⎝ Di ⎠ ⎝ μ ⎠
2
−0.2
⎛ D ⎞ ⎛ ( D − Di ) ρ v ⎞
0.5 −
⎛ μ ⎞ 3
⇒ K m = 0.023v ⎜ o ⎟ ⎜ o ⎟ ⎜ ⎟
⎝ Di ⎠ ⎝ μ ⎠ ⎝ ρ DAB ⎠

The diffusivity of naphthalene in air is¶, DAB = 8.2 × 10−6 m 2 /s .

Plugging in the values, we get K m = 0.0269 m/s.



Heath et al., Risk assessment of total petroleum hydrocarbons, in Hydrocarbon contaminated
soils, vol. III, 1993.
Problem 6.9

a.) Assuming steady state, no convection, and no volumetric heat generation, the energy balance
equation for the fluid in spherical coordinates is:

α ∂ ∂T
0= (r 2 )
r ∂r
2
∂r

We can easily integrate to obtain the general solution:

∂ 2 ∂T
(r )=0
∂r ∂r
∂T
r2 =A
∂r
∂T A
= 2
∂r r
A
T (r ) = + B
r

The boundary conditions we are given are:

T (r = R) = 50C
T (r = ∞) = 25C

Solving for the integration constants, we obtain:

25 = B
A
50 = + 25
R
A
25 =
R
A = 25R

Thus, the temperature profile outside the sphere is:

25 R
T (r ) = + 25
r
R
T (r ) = 25( + 1)
r

By setting the heat flux at the interface to Newton’s law of cooling for the rate of heat transfer,
we obtain
∂T
q = −k air r = R = − h(Ts − T∞ )
∂r
∂T
k air r=R
h= ∂r
(Ts − T∞ )
25R
k air r=R
h= r2
(Ts − T∞ )
25
k air
h= R
25
k
h = air
R

And we know the Nusselt number is equal to


hD
Nu =
k air
k air 2 R
Nu =
R k air
Nu = 2

Now we can compute the temperature as a function of time by beginning with a Level III energy
balance around the sphere.

4 4 3
ρC p ( πr 3 )T t + Δt = ρC p ( πr )T t + qΔt
3 3
4 dT
ρC p ( πr 3 ) =q
3 dt

And now we can use a constitutive equation for q.

4 dTs
ρC p ( πr 3 ) = −h(4πr 2 )(T − T∞ )
3 dt
1 dTs 4πr 2 h 3h
=− =−
(T − T∞ ) dt 4
ρC p ( πr 3 ) ρC p r
3

Let’s separate and integrate.


Ts t
1 3h
∫0 (T − T∞ )dTs = −∫0 ρC p r dt
T − T∞ 3h
ln( )=− t
Ts − T∞ ρC p r
3h
T − T∞ −
ρC r
t
=e p
Ts − T∞
3h
− t
ρC p r
T (t ) = (Ts − T∞ )e + T∞

We are given the following information:

J
C p air = 1.005
gK
g
ρ air = 1127
m3
W
k air = 0.0271
mK

We need to calculate the value of h from the Nusselt number.

hD
Nu =
kair
Nu (k air )
h=
D
W
2(0.0271)
h= mK
0.01m
W
h = 5.42
K

So the temperature function becomes:

W
3( 5.42 )
− K t
g J
(1127 3 )(1.005 )( 0.005 m )
T (t ) = 25e m gK
+ 25

T (t ) = 25e −2.87 t + 25
Temperature as a function of time

60

50

40
Temperature (deg C)

30

20

10

0
0 1 2 3 4 5 6
time (s)

b.) Now the problem changes because we no longer have a quiescent fluid, but a fluid with a
velocity of 1 cm/sec. We also are given a correlation for the Nusselt number that depends on the
Reynolds and Prandtl numbers. We need to start by calculating the Reynolds number for the
given flow rate.

m
(0.01m)(0.01 )
Dv
s = 5.9
Re = =
υ m2
(16.97 ⋅ 10 −6 )
s

This is in the laminar range, and from any table (Perry’s, etc) we can look up the value of the
Prandtl number for air.

Pr = 0.72

So the new Nusselt number becomes:

Nu = 2 + 0.60 Re1 / 2 Pr 1 / 3
Nu = 2 + 0.60(5.9)1 / 2 (0.72)1 / 3
Nu = 3.3

From this new value of the Nusselt number, we need to re-calculate the value of h.
hD
Nu =
k air
Nu (k air )
h=
D
W
3.3(0.0271 )
h= mK
0.01m
W
h = 8.94
K

We can use the same expression for the temperature of the lead sphere with the new value of h to
obtain:

W
3(8.94 )
− K t
g J
(1127 3 )(1.005 )( 0.005 m )
T (t ) = 25e m gK
+ 25
−4.75t
T (t ) = 25e + 25
Temperature as a function of time

60

50

40
red = no flow
Temperature (deg C)

blue = flow of 1 cm/sec

30

20

10

0
0 1 2 3 4 5 6
time (s)

Thus, adding a velocity flow will cool the sphere more quickly.

c.) Now we need to calculate the Biot number for the two cases in parts a and b. From Perry’s
handbook, we see that k lead =35W/mK, and the equation for Biot number is:

hair D
Bi =
k lead
For the “no flow” case (part a), the Biot number is:

W
(5.42 )(0.01m)
Bi = K
W
35
mK
−3
Bi = 1.5 ⋅ 10

For the flow case (part b), the Biot number is.

W
(8.94 )(0.01m)
Bi = K
W
35
mK
−3
Bi = 2.6 ⋅ 10

We can make the “lumped assumption” as long as the Biot number is much less than one (which
it is in both cases). If we assume that the lumped assumption becomes invalid at a Biot number
of 0.01, we can back-calculate the heat transfer coefficient and therefore, Reynolds number and
fluid velocity for this value of Biot.

hair D
Bi =
k lead
Bi ⋅ k lead
hair =
D
W
0.01 ⋅ 35
hair = mK
0.01m
W
hair = 35
K

Now we can calculate the new maximum Nusselt number.

hD
Nu =
k air
W
(35)(0.01m)
Nu = K
W
0.0271
mK
Nu = 12.9
Since the Prandtl number will remain the same (property of the fluid), we can calculate the new
maximum Reynolds number from the heat transfer correlation:

Nu = 2 + 0.60 Re1 / 2 Pr 1 / 3
12.9 = 2 + 0.60 Re1 / 2 (0.72)1 / 3
10.9
1/ 3
= Re1 / 2
0.6(0.72)
10.9
Re = ( )2
0.6(0.72)1 / 3
Re = 411

And finally, we can use the definition of Reynolds number to calculate the maximum fluid
velocity rate for which the lumped temperature analysis will apply.

Dv (0.01m)v
411 = =
υ m2
(16.97 ⋅ 10 −6 )
s
m cm
v = 0.70 = 70
s s
Problem 6.10

a.) We need to start by calculating the Nusselt number, which we can use to calculate the heat
transfer coefficient. Let’s start with the Reynolds number to determine which correlation we
should use for the Nusselt.

3miles 1m 1hr m
v= = 1.34
hr 0.0006214mile 3600s s
m
(1.34 )(0.3m)
vD s
Re = = = 2.5 ⋅104
ν m 2
(1.6 ⋅10−5 )
s

There are a number of correlations that can be used to determine the Nusselt number for a
circular cylinder in cross flow. One such, suggested by Zhukauskas1 is given below. For this
correlation at 103<Re<2x105, we use the values of constants as C=0.26, m=0.6. Prandtl for air is
0.72, so we can use n=0.37. Prandtl at the surface should be the same value.

Pr 1 / 4
Nu = C (Re) m (Pr) n ( )
Prs
Nu = 0.26(2.5 ⋅ 10 4 ) 0.6 (0.72) 0.37 (1)1 / 4
Nu = 100

And we can use this value of the Nusselt to calculate h.

hD
Nu =
k
Nuk
h=
D
W
100(.026 )
h= mK = 8.67 W
0.3m m2 K

Now we can use Newton’s Law of Cooling to calculate the rate of heat loss.

Q = ha (Tair − Ts )
W
Q = (8.67 )(0.3π )(1.8)m 2 (5 − 30) K
m2K
Q = −367.6W
b.) We need to repeat the process from part a, starting by recalculating the Reynolds number.

20miles 1m 1hr m
v= = 8.94
hr 0.0006214mile 3600 s s
m
(8.94 )(0.3m)
vD s
Re = = = 1.7 ⋅ 10 5
ν m 2
(1.6 ⋅ 10 −5 )
s
So again we should use the values of constants as C=0.26, m=0.61. Prandtl for air is 0.72, so we
can use n=0.37. Prandtl at the surface should be the same value.

Pr 1 / 4
Nu = C (Re) m (Pr) n ( )
Prs
Nu = 0.26(1.7 ⋅ 10 5 ) 0.6 (0.72) 0.37 (1)1 / 4
Nu = 317

And we can use this value of the Nusselt to calculate h.

hD
Nu =
k
Nuk
h=
D
W
317(.026 )
h= mK = 27.4 W
0.3m m2 K

Now we can use Newton’s Law of Cooling to calculate the rate of heat loss.

Q = ha (Tair − Ts )
W
Q = (27.4 2
)(0.3π )(1.8)m 2 (5 − 30) K
m K
Q = −1.16kW

c.)
Neglecting evaporation…
We need to assume a thickness for the water layer. Let this thickness be 1μm.

We can then use the following to determine the overall heat transfer coefficient.
1
U=
1 r ln(rout / rin ) rout
+ out +
hout k wall hin rin

where h out is the heat transfer coefficient between the air and the liquid layer, h in is the heat
transfer coefficient between the liquid layer and the surface of the body, and the water layer has a
thermal conductivity of 0.6 W/mK (Table 5.1).

r in =0.15m, r out =(0.15+1e-6)m

The Nu number will be the same for the outer boundary of the water layer as it was in part b.
Thus, Nu=317.

Calculate h out .

Nuk
hout =
D
W
317(.026 )
mK W
hout = = 27.4 2
0.3 + (2e − 6)m m K

Since we are assuming the temperature does not change within the boundary layer (constant at
30C), h in is negligible.

Therefore,

1
U=
1 0.300001m ln(0.300001m / 0.3m)
2
+
27.4W / m K 0.6W / mK

U=27.4W/m2K

Therefore,

Q = ha (Tair − Ts )
W
Q = (27.4 2
)(0.3π )(1.8)m 2 (5 − 30) K
m K
Q = −1.16kW

As you can see, there is not much difference in the rate of heat loss between a dry boundary and
a wet one when evaporation is not considered and the layer of water is very small.
Accounting for evaporation…
Since we are given the relative humidity of the air at 5C, we can calculate the bulb pressure. At
5oC, P vap =872 Pa
P
%humidity = bulb
Pvap
Pbulb = (%humidity )( Pvap )
Pbulb = 0.75(872 Pa)
Pbulb = 654 Pa
And now we can calculate the wet bulb concentration.

ρ water
C bulb =
MWwater
⎛ kg ⎞⎛ 1m 3 ⎞
⎜1000 3 ⎟⎜⎜ ⎟
⎝ m ⎠⎝ 1000 L ⎟⎠
C bulb =
⎛ g ⎞⎛ 1kg ⎞
⎜18.01 ⎟⎜ ⎟
⎝ mol ⎠⎜⎝ 1000 g ⎟⎠
C bulb = 55.5mol / L

We can calculate the concentration of water in air using the Ideal Gas Law.

Pair
C air =
RTair
872 Pa
C air = 3
m Pa L
(8.314 )(278 K )(1000 3 )
molK m
mol
C air = 3.77 ⋅ 10 − 4
L

Now we can use the Chilton-Colburn analogy to calculate K m .

jH = jD
h K
(Pr) 2 / 3 = m ( Sc) 2 / 3
ρC pV∞ V∞
h Pr 2 / 3
Km = ( )
ρC p Sc

To calculate the Lewis (Sc/Pr) number, we can use:


kg 1000 g
1.7617E-5 ( )
μ m.s 1kg
Scair = = = 0.68
ρD g −5 m
2
1000 L
(1.225 )(2.1 ⋅10 )( )
L s 1m3

where D is the species diffusivity.

And the Prandtl number is a function of the material (0.72 as we determined earlier). Use h from
part b.

W
27.4
m2 K 0.72 2 / 3
Km = ( )
g J 1000 L 0.68
(1.225 )(1 )( )
L gK 1m 3
m
K m = 2.32 ⋅ 10 −2
s

And finally we can calculate the heat loss.

Q = ΔH vap K m a(C w,bulb − C w,air )


kJ m 1000 L mol
Q = (44 )(2.32 ⋅ 10 − 2 )(0.3π )(1.8)m 2 ( 3
)(55.5 − 3.77 ⋅ 10 − 4 )
mol s 1m L
Q = 96,000kW

This value is significantly higher than that obtained without humidity. This indicates that
evaporation is a big contributor to heat loss for this system.

NOTE: In the solutions given above, I found the rate of heat loss from the SURFACE of the
body, which was set at 30oC. Since we know the average temperature of a human body is 37oC,
it is possible to find the overall heat loss from the core of the body.

1
Zhukauskas, A. “Heat Transfer from Tubes in Cross Flow,” in J.P. Hartnett and T.F. Irvine, Jr.,
Eds., Advances in Heat Transfer, Vol. 8, Academic Press, New York, 1972.
Problem 6.11

a.) If we are interested in pressure-driven fluid flow in the z-direction, neglecting gravity, the
Navier-Stokes equation for a 1-D flow profile becomes

∂P ∂ 2v
0=− + μ [ 2z ]
∂z ∂y

But the pressure change will be constant and linear with distance down the tube, so we can make
a further simplification and integrate to obtain:

ΔP ∂ 2vz
= μ[ 2 ]
L ∂y
∂ 2 v z ΔP
=
∂y 2 μL
∂v z ΔP
= y+A
∂y μL
ΔP 2
vz ( y) = y + Ay + B
2 μL

Our boundary conditions are symmetry at the center of the slit and no slip at the walls:

∂v z
= 0@ y = 0
∂y

vz = 0 @ y = H .

Therefore,

∂v z ΔP
= ( 0) + A = 0
∂y μL
A=0
ΔP 2
vz (H ) = H +B=0
2 μL
ΔP 2
B=− H
2 μL

So the profile looks like:

ΔP 2
vz ( y ) = (y − H 2).
2 μL
Now we can look to the energy balance in 2 dimensions to solve for the temperature distribution
in the duct.

ρC p TΔyΔz (1) t + Δt = ρC p TΔyΔz (1) t + ρC p Tv z Δy (1)Δt z − ρC p Tv z Δy (1)Δt z + Δz + q y Δy (1)Δt y − q y Δy (1)Δt y + Δy

Dividing through by delta x, y, z and t and using the definition of the derivative, we get:

∂ ( ρC p T ) ∂ ( ρC p Tv z ) ∂q y
=− −
∂t ∂z ∂y
∂T ∂T ∂q y
ρC p = − ρC p v z ( y ) −
∂t ∂z ∂y

Assuming steady state, we get:

∂T ∂q y
0 = − ρC p v z ( y ) −
∂z ∂y

and applying Fourier’s Law of Heat Conduction for q, we get:


∂T
∂ (−k )
∂T ∂y
0 = − ρC p v z ( y ) −
∂z ∂y
∂T ∂ 2T
ρC p v z ( y ) =k 2
∂z ∂y

Now from the problem statement we can assume the derivative of temperature with respect to z
is constant and replace it with a constant value,

∂ 2T ΔT
= ρC p v z ( y ) .
∂y 2
kL

We also must use the velocity profile we calculated earlier when it’s time to integrate.

∂ 2T ΔP ΔT
= ρC p [ ( y 2 − H 2 )]
∂y 2
2μL kL
∂T ρC p ΔPΔT 2 ρC p ΔPΔT 2
∫ ∂ ∂y = ∫ 2kμL2 y ∂y − ∫ 2kμL2 H ∂y
∂T ρC p ΔPΔT y 3 ρC p ΔPΔT 2
= − H y+ A
∂y 2kμL2 3 2kμL2

Integrating the equation again yields the general solution


ρC p ΔPΔT y 4 ρC p ΔPΔT 2 y 2
T ( y) = − H + Ay + B .
2kμL2 12 2kμL2 2

Now we need to use boundary conditions. Again from symmetry,

∂T
= 0@ y = 0
∂y

And since temperature is changing linearly as we travel down the length of the plates, we know
the wall temperature will equal:

ΔT
T = Tw ( z ) = Tw (0) + z@ y = H
L

Applying B.C.

∂T ρC p ΔPΔT 03 ρC p ΔPΔT 2
= − H (0) + A = 0
∂y 2kμL2 3 2kμL2
A=0

And,
ρC p ΔPΔT H 4 ρC p ΔPΔT 2 H 2 ΔT
T ( y, z ) = − H + B = Tw (0) + z
2kμL 2
12 2kμL 2
2 L
ΔT ρC p ΔPΔT H 4 ρC p ΔPΔT H 4
B = Tw (0) + z− +
L 2kμL2 12 2kμL2 2
ΔT ρC p ΔPΔT 5H 4
B = Tw (0) + z+
L 2kμL2 12

So the complete profile becomes:

ρC p ΔPΔT y 4 ρC p ΔPΔT 2 y 2 ΔT ρC p ΔPΔT 5 H 4


T ( y, z ) = − H + T ( 0 ) + z +
2kμL2 12 2kμL2 2kμL2
w
2 L 12
ρC p ΔPΔT y 4 2 y
2
5H 4 ΔT
T ( y, z ) = [ − H + ] + Tw (0) + z
2kμL 2
12 2 12 L
ΔPΔT y 4 2 y
2
5H 4 ΔT
T ( y, z ) = [ − H + ] + Tw (0) + z
2αμL 12
2
2 12 L

Finally, we can calculate the average temperature using the mean value theorem:
H
− 1
H ∫0
T ( z) = T ( y, z )dy

ΔPΔT y 4 2
5H 4 ΔT
H
− 1 2 y
T ( z) = ∫ {
H 0 2αμL 12
2
[ − H
2
+
12
] + Tw (0) +
L
z}dy

− 1 ΔPΔT y 5 2 y
3
5H 4 ΔT
T ( z) = { [ − H + y ] + Tw (0) y + zy} |0H
H 2αμL 60
2
6 12 L
− 1 ΔPΔT H 5 2 H
3
5H 4 ΔT
T ( z) = { [ −H + H ] + Tw (0) H + zH }
H 2αμL 60
2
6 12 L
− ΔPΔT H 4 H 4 5H 4 ΔT
T ( z) = [ − + ] + Tw (0) + z
2αμL 60
2
6 12 L
− 2 ΔPΔT 4 ΔT
T ( z) = H + Tw (0) + z
15 αμL 2
L

b) The equation for Nusselt number is:

hH
Nu =
k

and we can calculate h from Newton’s law of cooling:


q = h(Tw − T )
q
h= −
(Tw − T )

We can use Fourier’s Law to calculate q, evaluated at the surface of the wall from our
temperature profile from the previous part:

∂T
q = −k y=H
∂y
ρC p ΔPΔT H 3 ρC p ΔPΔT 2
q = −k ( − H H)
2kμL2 3 2kμL2
ρC p ΔPΔT 2 H 3 ρC p ΔPΔT H 3
q= =
2μL2 3 μL2 3
Therefore,
ρC p ΔPΔT H 3
μL2 3
h= −
(Tw − T )

Here we can plug in the equations for Tw and T and use this to simplify. We know Tw is simply
the temperature profile evaluated at y=H.

ΔPΔT H 4 H 4 5 H 4 ΔT
T (H , z) = [ − + ] + Tw (0) + z
2αμL 122
2 12 L
ΔT
Tw = T ( H , z ) = Tw (0) + z
L

And we know:

− 2 ΔPΔT 4 ΔT
T ( z) = H + Tw (0) + z
15 αμL2
L

So the difference becomes:

− ΔT 2 ΔPΔT 4 ΔT
(Tw − T ) = [Tw (0) + z] − [ H + Tw (0) + z]
L 15 αμL2
L
− 2 ΔPΔT 4
(Tw − T ) = − H
15 αμL2

Plugging into our equation for h, we get:

ρC p ΔPΔT H 3
μL2 3
h=
2 ΔPΔT 4
− H
15 αμL2
5k
h=
2H

So therefore the Nusselt number equals:

hH 5k H
Nu = =
k 2H k

5
Nu =
2
Problem 6.12

a) We can use the same velocity profile as we obtained in Problem 6.11 on Homework 9.

ΔP 2
v z ( y) = (y − H 2)
2 μL

But this time we need to do a mass balance rather than energy balance.

C A ΔyΔz =C A ΔyΔz t + C A v z ΔyΔt z − C A v z ΔyΔt + J y ΔyΔt − J y ΔyΔt


I I I I
t + Δt z + Δz y y + Δy

Dividing through by delta y, z and t and using the definition of the derivative, we get:

∂ (C A ) ∂ (C A v z ) ∂J y
I I

=− −
∂t ∂z ∂y

Assuming steady state, we get:

∂ (C A ) ∂J y
I

0 = −v z ( y ) −
∂z ∂y

And applying Fick’s Law for J, we get:

∂C A
I

∂ (− D A )
∂C A ∂y
I

0 = −v z ( y ) −
∂z ∂y
∂C A ∂ 2C A
I I

v z ( y) = DA
∂z ∂y 2

Again we can assume the derivative of concentration with respect to z is constant and replace it.

ΔC A ∂ 2C A
I I

v z ( y) = DA
L ∂y 2
∂ 2C A ΔC A
I I

= v z ( y)
∂y 2
DA L

We also must use the velocity profile we calculated earlier when it’s time to integrate.
∂ 2C A ΔP 2 ΔC A
I I
=[ ( y − H 2 )]
∂y 2
2 μL DA L
∂C A ΔPΔC A 2 ΔPΔC A 2
I I I

∫∂ ∂y
=∫
2 D A μL2
y ∂y − ∫
2 D A μL2
H ∂y

∂C A ΔPΔC A y 3 ΔPΔC A
I I I
= − H2y + A
∂y 2 D A μL 3 2 D A μL
2 2

ΔPΔC A y 4 ΔPΔC A 2
I I
2 y
C A ( y) = − + Ay + B
I
H
2 D A μL2 12 2 D A μL2 2

Now we need to use boundary conditions. Again from symmetry,

∂C A
I

= 0@ y = 0
∂y

And since temperature is changing linearly as we travel down the length of the plates, we know
the wall temperature will equal:

ΔC A
I

C A = C A w ( z ) = C A w (0) + z@ y = H
I I I

Applying B.C.

∂C A ΔPΔC A 0 3 ΔPΔC A
I I I

= − H 2 (0) + A = 0
∂y 2 D A μL2 3 2 D A μL2
A=0

And,
ΔPΔC A H 4 ΔPΔC A 2
ΔC A
I I I
2 H
C A ( y, z ) = − + = w (0) +
I I
H B C z
2 D A μL2 12 2 A μL2
A
2 L
ΔC A ΔPΔC A H 4 ΔPΔC A H 4
I I I

B = C A w ( 0) + z− +
I

L 2 D A μL2 12 2 D A μL2 2
ΔC A ΔPΔC A 5 H 4
I I

B = C A w ( 0) + z+
I

L 2 D A μL2 12

So the complete profile becomes:

ΔPΔC A y 4 ΔPΔC A 2
ΔC A ΔPΔC A 5H 4
I I I I
2 y
C A ( y) = − + + +
I I
H C w ( 0) z
2 D A μL2 12 2 D A μL2 2 D A μL2 12
A
2 L
ΔPΔC A y 4 2
5H 4 ΔC A
I I
2 y
C A ( y, z ) = − + ] + C A w (0) +
I I
[ H z
2 D A μL 12
2
2 12 L

Finally, we can calculate the average concentration using the average property equation.
− H
1
C A ( z ) = ∫ C A ( y, z )dy
I I

H 0

1 ΔPΔC A y 4 2
5H 4 ΔC A
H I I
2 y
C A ( z) = ∫ { − + ] + C A w (0) +
I I
[ H z}dy
H 0 2 D A μL 12
2
2 12 L

1 ΔPΔC A y 5 3
5H 4 ΔC A
I I
2 y
C A ( z) = { [ −H + y ] + C A w (0) y +
I I
zy} | 0H
H 2 D A μL 60
2
6 12 L

1 ΔPΔC A H 5 3
5H 4 ΔC A
I I
2 H
C A ( z) = { − + H ] + C A w (0) H +
I I
[ H zH }
H 2 D A μL 60
2
6 12 L

ΔPΔC A H 4 H 4 5 H 4 ΔC A
I I

C A ( z) = − + ] + C A w (0) +
I I
[ z
2 D A μL 60
2
6 12 L

2 ΔPΔC A ΔC A
I I

C A ( z) = H 4 + C A w (0) +
I I
z
15 D A μL2
L

b) The equation for Sherwood number is:

KmH
Sh =
DA

And we can calculate K m from the rate law for mass transfer:


J = K m (C A − CA )
I I
w

J
Km = −
w − CA )
I I
(C A

And we can use Fick’s Law to calculate J, evaluated at the surface of the wall.

∂C A
I
J = − DA y=H
∂y
ΔPΔC A H 3 ΔPΔC A
I I
J = − DA ( − H 2H )
2 DA μL2 3 2 DA μL2
ΔPΔC A 2 H 3 ΔPΔC A H 3
I I
J= =
2 μL2 3 μL2 3
Therefore,
ΔPΔC A H 3
I

μL2 3
Km = −
− CA )
I I
(C A w

I I I
Here we can plug in the equations for C A w and C A and use this to simplify. We know C A w is
simply the concentration profile evaluated at y=H.

ΔPΔC A H 4 H 4 5 H 4 ΔC A
I I

C A ( H , z) = − + ] + C A w (0) +
I I
[ z
2 D A μL 12
2
2 12 L
ΔC A
I

= C A ( H , z ) = C A w (0) +
I I I
CA w z
L

And we know:


2 ΔPΔC A ΔC A
I I

C A ( z) = H 4 + C A w (0) +
I I
z
15 D A μL2
L

So the difference becomes:


ΔC A 2 ΔPΔC A ΔC A
I I I

w − C A ) = [C A w (0) + z] − [ H 4 + C A w ( 0) +
I I I I
(C A z]
L 15 D A μL2
L

2 ΔPΔC A
I
(C A w − C A ) = − H4
I I

15 D A μL2

Plugging into our equation for K m , we get:

ΔPΔC A H 3
I

μL2 3
Km =
2 ΔPΔC A
I

− H4
15 D A μL2
5D A
Km =
2H

So therefore the Sherwood number equals:

K m H 5 DA H 5
Sh = = ⇒ Sh =
DA 2 H DA 2
Problem 6.13

The first step is to determine the mass transfer coefficient from the salt/water experimental data
we are given. From that, we can use the Chilton-Colburn analogy to calculate the heat transfer
coefficient for our rock system. We can start by performing the familiar component mass
balances for the two components. Since we know phase 2 (salt tablets) are solid, we can use the
density rather than the concentration.

I
dV I C A
= K m a[C A − Mρ II ]
I

dt
dV II ρ II
= − K m a[C A − Mρ II ]
I

dt

As we did in chapter 4, we can assume that Mρ II = C A, sat which is much greater than C A for all
I

times. We can assume that V I and ρ II are constant, and that the area (for spherical tablets) can
be represented using the shape factor of 4.836. Therefore,

a = αN 1 / 3 [V II ] 2 / 3 = 4.836(30)1 / 3 [V II ] 2 / 3
a = 15.036[V II ] 2 / 3

Therefore, we can simplify the model equations to:

I
dC A
V I
= (15.036[V II ] 2 / 3 ) K m [ Mρ II ]
dt
dV II
ρ II = − K m (15.036[V II ] 2 / 3 )[ Mρ II ]
dt

We can integrate this second equation to obtain an equation for V II .

dV II
= − K m (15.036[V II ] 2 / 3 )[ M ]
dt
V II t

∫ [V ] dV = − ∫ 15.036MK m dt
II − 2 / 3 II

VO II 0

3(V II )1 / 3 − 3(VO )1 / 3 = −15.036 MK m t


II

V II (t ) = [(VO )1 / 3 − 5.012 MK m t ]3
II

I
We can plug this equation into our other model equation to obtain an expression for C A .
15.036 K m [ Mρ II ]
I
dC A
= [(VO )1 / 3 − 5.012 MK m t ] 2
II
I
dt V
CAI
15.036 K m [ Mρ II ] t
∫ dC A = ∫ [(VO )1 / 3 − 5.012 MK m t ] 2 dt
I II
I
C A,O I
V 0

We can use u substitution to evaluate the right side of the integral.


Let u = (VO )1 / 3 − 5.012 MK m t
II

du = −5.012 MK m dt
−1
du = dt
5.012MK m

So integrating:

CAI (VO II )1 / 3 −5.012 MK m t


3 ρ II
∫ dC =− I ∫u
I 2
A du
C A,O I
V (VO II )1 / 3

ρ II
C A − C A, 0 = − {[(VO )1 / 3 − 5.012 MK m t ]3 − (VO )}
I I II II
I
V
ρ II
C A = C A, 0 − {[(VO )1 / 3 − 5.012MK m t ]3 − (VO )}
I I II II
I
V

But we want this in a form that can be plotted to find the (linear) slope and therefore the mass
transfer coefficient. Since C A, 0 = 0
I

ρ II
CA = − {[(VO )1 / 3 − 5.012MK m t ]3 − (VO )}
I II II
I
V
I I
CA V
= VO − [(VO )1 / 3 − 5.012MK m t ]3
II II

ρ II

I
CA V I
VO − = [(VO )1 / 3 − 5.012MK m t ]3
II II

ρ II

I
CA V I
{VO − }1 / 3 = (VO )1 / 3 − 5.012MK m t
II II

ρ II

I
CA V I 5.012MK m t
(VO )1 / 3 {1 − }1 / 3 = (VO )1 / 3 [1 −
II II
]
VO ρ (VO )1 / 3
II II II

I
CA V I 5.012MK m t
{1 − }1 / 3 = [1 − ]
VO ρ (VO )1 / 3
II II II
I
CA V I ) 5.012MK m
After all this algebra, we can now call = c and = ω . The equation becomes:
VO ρ (VO )1 / 3
II II II

)
{1 − c}1 / 3 = 1 − ωt

)
We can plot {1 − c}1/ 3 versus t and the slope will give us -ω, which we can use to calculate K m ,
the mass transfer coefficient.

Experimental Data for Solid-Liquid Mass Transfer Experiment

1.2

0.8

y = -0.0014x + 0.9932
(1-c_hat)^1/3

R2 = 0.9937

0.6

0.4

0.2

0
0 50 100 150 200 250 300
time (s)

We find that ω is equal to 0.0014 s-1. Therefore, the mass transfer coefficient can be calculated to
be:
I
5.012 MK m 5.012C A, sat K m
= = ω = 0.0014
(VO )1 / 3 (VO )1 / 3 ρ II
II II

g
5.012(0.360
3
)K m
0.0014 = cm
g
(8.889cm 3 )1 / 3 (2.16 3 )
cm
cm m
K m = 3.47 ⋅ 10 −3 = 3.47 ⋅ 10 −5
s s

Now we can use the Colburn analogy to calculate the heat transfer coefficient.
jH = jD
h K
(Pr) 2 / 3 = m ( Sc) 2 / 3
ρC pV∞ V∞
Sc 2 / 3
h = K m ρC p ( )
Pr
And we know:

Prwater = 6.5
g
0.0065
μ cm.s
Scwater = = = 298
ρD g −5 cm 2
1000 L
(990 )(2.2 ⋅ 10 )( )
L s 1003 cm 3
g
ρ water = 990
L
J
C p water = 4.18
gK

So we can estimate h as:

m g J 298 2 / 3 1000 L
h = (3.47 ⋅ 10 −5 )(990 )(4.18 )( ) ( )
s L gK 6.5 1m 3
W
h = 1839
m2 K

The final step is to determine heating time. We can perform an energy balance around the water
and integrate.

ρ water C p waterVwater Twater t + Δt = ρ water C p waterVwater Twater t + Δt +q


dTwater
ρ water C p waterVwater =q
dt
dTwater
ρ water C p waterVwater = ha(Tstone − Twater )
dt
Twater ( boil ) t
dTwater ha

Twater ( 0 )
=∫
(Tstone − Twater ) 0 ρ water C p waterVwater
dt

(Tstone − Twater (0)) ha


ln[ ]= t
(Tstone − Twater (boil )) ρ water C p waterVwater
ρ water C p waterVwater (Tstone − Twater (0))
t= ln[ ]
ha (Tstone − Twater (boil ))
Plugging in,
g J
(990 )(4.18 )(6 L)
L gK (450 − 25)
t= ln[ ]
W 1m 2
(450 − 100)
(1839 2 )(100πcm )(2
)
m K 10000cm 2
t = 83s = 1.4 min
Problem 6.14

(a) We are given that

l
τ=
u

1
⎛ν 3 ⎞ 4
u = (ν L ε ) 4 .
1
where l = ⎜⎜ L ⎟⎟ and
⎝ε ⎠

(ν L is the kinematic viscosity)

Therefore, τ =
νL
ε .

Since the liquid-side mass transfer coefficient can be estimated with

k L = DL
τ

we get an equation for the mass transfer coefficient as a function of the kinematic viscosity and
the energy dissipation.

k L = DL ⎛⎜ ε ⎞⎟
4

⎝ νL ⎠

We can then use the relationship for the average liquid velocity, u (given above), to find the
mass transfer coefficient as a function of u and ε.

DL ε
kL =
u

(b) We are given that

2 f (u )
3
0.0791
ε= and f = 1
for turbulent flow through a pipe.
d pipe Re 4

Plugging these values into the mass transfer coefficient derived above gives
0.3977 DL u
kL =
d pipe (Re )
1
8

The diameter of the pipe in the above equations is assumed to be the hydraulic diameter for the
stream.

We want to find a relationship for the Sherwood number using this equation, where

k L d pipe 0.3977 d pipe u


Sh = =
DL (Re )
1/ 8
DL

d pipe u νL
Since Re = and Sc = ,
νL DL

0.3977(Re ) (Sc ) 12
1
2
Sh =
(Re ) 18

Sh = 0.3977 Re 0.375 Sc 0.5


Problem 6.15

The physical situation is indicated in Figure 6.14. First we will define a residence time
distribution function E(t), with an average residence time τ. The probability that a certain surface
element is at the interface for a time t is then given by E(t)dt. Thus, the probability for a certain
surface element to have been at the surface for a time t, E(t)dt, has to be equal to the probability
to find one that has been around for time t-dt, minus surface elements that leave the interface
during time dt. The probability that an element leaves the surface is given by dt/τ times the
probability of the element being on the surface (E(t)dt. This word statement is expressed in the
equation below:
( E (t )dt )t +Δt = ( E (t )dt )t − k1 ( E (t )dt )t Δt
lim
Δt →0

dE (t )
= −k1 E (t )
dt

When integrated, this yields: E (t ) = Ae − k1t . The constant of integration can be determined from
the requirement that the sum of the probabilities has to be 1 (because the interface needs to be
full of surface elements t any given time):

∫ E (t )dt = 1
0

Thus, the final expression for the residence time distribution at the surface E(t):

E (t ) = k1 exp ( − k1t )

The average residence time can be found by taking the first moment of the residence time
distribution.

∞ ∞ − exp(−∞)(∞ + 1) exp(0)(0 + 1) 1
τ = ∫ t ⋅ E (t )dt = ∫ tk1 exp(−k1t )dt = + =
0 0 k1 k1 k1

Therefore, our final residence time distribution equation is

E (t ) = (1 τ ) exp(− t τ ) .

The flux from phase I into each surface element is the product of the flux and the probability that
that surface element will reside at the surface for a time span t. The average flux is then
expressed as:
∞ ∞
( )
⎛ exp −t
τ k ρ C p vmax
(

qz = ∫ E (t ) qz z =0 dt = ∫ ⎜ b ) dt
I ⎟
T II
− T
⎜ τ πx ⎟
0 0
⎝ ⎠

J A, z = ∫ E (t ) J A, z z =0 dt = ∫
⎛ exp −t


τ ( )
DAvmax
(

MC A − C A,b ⎟ dt
II I
)
0 0⎜
⎜ τ π x ⎟⎟
⎝ ⎠
where q z and J A, z are the expressions found from penetration theory. The fraction
z =0 z =0
x/v max is an expression of how long the surface element has been exposed at the interface. It is
therefore replaced by the time t that the surface element has spent at the interfacial area. These
expressions can then be integrated readily:

qz = k ρ C p ( T − T

)∫
( τ )dt
exp −t
J A, z = DA ( MC − C

( τ )dt
exp −t
I
b
II

τ πt
I
A
II
A ,b )∫ τ πt
0 0

k ρCp
=
τ
(T I
− TbII ) =
DA
τ
( MC I
A − C AII,b )

Comparison with the constitutive equations for heat and mass transfer, yield expressions for the
heat transfer coefficient h and mass transfer coefficient K m :

qz = h (T I − TbII ) J A, z = K m ( MC AII − C AI ,b )
k ρC p DA
∴ h= ∴ Km =
τ τ
The same square root dependence of h and K m on the diffusivities is found as for the penetration
theory. The difficulty however arises with the average residence time τ. This residence time is
unknown and has to be either found from detailed, and complex, turbulent fluid mechanics
calculations, or it has to be experimentally fitted by comparison with experimental data.
Problem 6.16

This problem is solved in P.V. Dankwerts, “Absorption by Simultaneous Diffusion and


Chemical Reaction,” Trans. Faraday Soc. 46 300-304, 1950.

Using the word statement, applied to the physical picture on page 275, we find
C AI ΔxΔyΔz |t +Δt = C AI ΔxΔyΔz |t +Vmax C AI ΔyΔzΔt |x −Vmax C AI ΔyΔz Δt |x +Δx
+ N A, z ΔxΔyΔt |z − N A, z ΔxΔyΔt |z +Δz −k1C AI ΔxΔyΔz Δt
Dividing by ΔxΔyΔzΔt and taking the differential limit yields:
∂C AI ∂C AI ∂N A, z
+ Vmax + + k1C AI = 0
∂t ∂x ∂z
dC AI
using N A, z = J A. z = − DA
dz
∂C A
I
∂ CA
2 I
∴Vmax − DA + k1C AI = 0
∂x ∂z 2

Where we are considering a steady state process. The boundary conditions are:
x = 0 C AI = 0
z = 0 C AI = C AI ,i = MC AII
z = ∞ C AI = 0

This equation can be integrated (see for example, Carslaw and Jaeger (1947, p. 111). The
solution (which can be checked by back substitution) is:
⎛ ⎞
⎜ ⎟
CAI
1 ⎛ k1 ⎞ ⎜ z xk1 ⎟
= exp ⎜⎜ − z ⎟ *erfc ⎜ −
MC AII 2 ⎝ DA ⎟⎠ xDA Vmax ⎟
⎜⎜ 2 ⎟⎟
⎝ Vmax ⎠
Note, erfc(x)=1-erf(x). The rate of adsorption can thus be found by differentiating and
calculating the flux at z=0, yielding:
⎛ ⎛ xk1 ⎞ ⎞
⎜ exp −
⎜ V ⎟⎟ ⎟

dC AI ⎜ ⎛ xk ⎞ ⎝ max ⎠ ⎟
N A , z | z = 0 = − DA |z =0 = MC AII DA k1 ⎜ erf ⎜ 1
⎟⎟ + ⎟
dz ⎜ V
⎜ ⎝ max ⎠ π 1
xk ⎟
⎜ Vmax ⎟
⎝ ⎠
Note that, in the absence of reaction, we found (6.4.11)
DAVmax
N A, z | z = 0 = MC AII
πx
Thus, the flux across the boundary is enhanced by the ratio:
⎛ ⎛ xk1 ⎞⎞
⎜ exp ⎜⎜ − ⎟⎟ ⎟
⎜ ⎛ xk1 ⎞ ⎝ Vmax ⎠⎟
DA k1 ⎜ erf ⎜⎜ ⎟⎟ + ⎟
V ⎛ ⎛ xk1 ⎞ ⎞
⎜ ⎝ max ⎠
xk
π 1 ⎟ ⎜ exp ⎜⎜ − ⎟⎟
⎜ ⎟ Vmax ⎟⎠ ⎟
N A, z | z = 0 ⎝
Vmax ⎜ ⎛ xk1 ⎞
⎠ = Vmax xk1 ⎝
no rxn
= ⎜ erf ⎜⎜ V ⎟⎟ + ⎟
N A,z |z =0 Vmax
⎜ ⎝ max ⎠
DAVmax xk
π 1 ⎟
πx ⎜ Vmax ⎟
⎝ ⎠
This ratio is denoted the “Hatta number”, a dimensionless parameter that gauges the importance
of reaction on the rate of mass transfer.

Note that, if the reaction rate is small, one can show:


DAVmax ⎛ V ⎞
N A, z | z = 0 ≈ MC AII ⎜1 + k1 max ⎟
πx ⎝ x ⎠
Thus, we see reaction increases the rate of mass transfer. Finally, we can write the effective
mass transfer coefficient in analogy to (6.4.13) as
DAVmax ⎛ V ⎞
km = 2 ⎜ 1 + k1 max ⎟
πL ⎝ 2L ⎠

⎛ V ⎞
The dimensionless Hatta number would simplify to ⎜1 + k1 max ⎟ for this simplification.
⎝ x ⎠
Problem 6.17

This problem is different than most heat transfer problems encountered so far in this course in
that we must account for the variation in thermal properties of the various materials involved.
For example, we see that, depending on which fluid we are considering at a certain temperature,
the viscosity varies by as much as an order of magnitude! Since it is much too mathematically
cumbersome to explicitly account for the variation of temperature (and thus material properties)
with both radial and axial position in the exchanger, we will make continued use of average
temperatures at which to evaluate thermal properties.

For evaluation of bulk fluid properties, we will refer to the average temperatures in the process stream
(phase 1) and the cooling water stream (phase 2), which are most accurately computed as the log mean
temperature in each fluid

T1 f − T1 T2 f − T2
T1 = and T2 = .
⎛T ⎞ ⎛ T2 f ⎞
ln ⎜ 1 f ⎟ ln ⎜ ⎟
⎝ T1 ⎠ ⎝ T2 ⎠

We will also refer to the log mean temperature difference, which we already know for a counter-current
exchanger is given by

(T2 f − T1 ) − (T2 − T1 f )
ΔTlm = .
⎛ T −T ⎞
ln ⎜ 2 f ⎟⎟
⎜ T2 − T1 f
⎝ ⎠

Finally, we will assume that the average inside and outside wall temperatures of the contact area,
respectively, can be estimated by

Tw,in = T1 − ΔTlm and Tw,out = T2 + ΔTlm .

In order to use these average temperatures to compute thermal properties of the fluids and the wall, the
data given in the tables were interpolated to come up with an analytical expression for the various thermal
properties of the materials. Plots of the data with best fit expressions are summarized below (where all
temperatures are in degrees celsius).

Water:
ρ ⎡⎣ kg/m3 ⎤⎦ = -0.0037 ⋅ T 2 - 0.0472 ⋅ T + 999.81
Cˆ p [ J/kg K ] = 3 ×10−6 ⋅ T 4 - 0.0008 ⋅ T 3 + 0.0826 ⋅ T 2 - 3.5516 ⋅ T + 4225.9
k [ W/m K ] = -1×10-5 ⋅ T 2 + 0.0023 ⋅ T + 0.5572
μ [ cP ] = 4 ×10−8 ⋅ T 4 - 1×10-5 ⋅ T 3 + 0.0011 ⋅ T 2 - 0.0576 ⋅ T + 1.7933
Stainless steel:
k [ W/m K ] = 0.0163 ⋅ T + 15.675
THERMAL PROPERTIES OF WATER

1010 4230

Heat capacity (J/kg K)


1000 4220 y = 3E-06x4 - 0.0008x3 + 0.0826x2 -
Density (kg/m^3)

3.5516x + 4225.9
990 4210 R2 = 0.9995
980 4200
970 4190
y = -0.0037x 2 - 0.0472x + 999.81 4180
960
R2 = 0.9988
950 4170
0 20 40 60 80 100 0 20 40 60 80 100
T (C) T (C)
Thermal conductivity (W/m

0.8 2
0.7 y = 4E-08x4 - 1E-05x3 + 0.0011x2 - 0.0576x
Viscosity (cP)
0.6 1.5 + 1.7933
0.5 R2 = 0.9999
1
K)

0.4
y = -1E-05x 2 + 0.0023x + 0.5572
0.3
R 2 = 0.9995
0.2 0.5
0.1
0 0
0 20 40 60 80 100 0 20 40 60 80 100
T (C) T (C)

a.) In order to choose an appropriate heat transfer correlation to use for the design of the heat
exchanger, we must first determine the operating flow regime of the exchanger. The log mean
temperature of the process stream is

373 K − 353 K
T1 = = 89.9o C .
⎛ 373 K ⎞
ln ⎜ ⎟
⎝ 353 K ⎠

At this temperature, we find the following thermal properties,

ρ = 965.66 ⎡⎣ kg/m3 ⎤⎦
Cˆ p = 4201[ J/kg K ]
k = 0.6832 [ W/m K ]
μ = 0.2425 [ cP ]

and for a 3” inner and 5” outer schedule-40S stainless steel pipe, we have from Table E3.6,
Di ,in = 0.0779 [ m ] , Do ,in = 0.0889 [ m ]
.
Di ,out = 0.1281[ m ] , Do ,out = 0.1410 [ m ]
For these values and the given process flow rate of 100 L/min, we obtain for the process stream:

ρVDin 4ρ q
Re = =
μ πμ Din
4 ( 965.66 [kg/m3 ])( 0.1 [m3 /min]/60 [s/min])
Re =
π ( 0.2425 ×10−3 [Pa ⋅ s]) ( 0.0779 [m])
Re = 1.085 ×105

This Reynolds number is well in the turbulent regime, so we can be fairly confident in any heat
transfer coefficient calculated using a correlation developed for turbulent flow. Thus, we can
use either the Ditus-Boelter or the Colburn correlations:
(NOTE: These are the correlations given on p. 368. Others from Table 6.5 or another source
may also be used as long as your data satisfy the constrictions of the given correlation.)

Ditus-Boelter: Nu = 0.023 Re 0.8 Pr 0.4

Colburn: Nu = 0.023 Re 0.8 Pr1/ 3

b.) The minimum flow rate required to carry out the operation is achieved when the cooling
water is exits the heat exchanger at the maximum allowable temperature of 30°C (NOTE: as this
temperature is lower than the equilibrium temperature, it will require a finite area).

First, we must calculate the thermal properties of cooling water at the average temperature

30o C − 5o C
T2 = = 17.3o C
⎛ 303 K ⎞
ln ⎜ ⎟
⎝ 278 K ⎠
ρ = 997.9 ⎡⎣ kg/m3 ⎤⎦
Cˆ p = 4186 [ J/kg K ]

k = 0.5940 [ W/m K ]
μ = 1.0686 [ cP ]

Thus, the minimum flow rate of cooling water can be computed using a level I balance around
the counter-current exchanger:
ρ1Cˆ p1 ⎛ T1 f − T1 ⎞
q2,min = q1 ⎜ ⎟
ρ 2Cˆ p 2 ⎜⎝ T2,max − T2 f ⎟⎠
(965.7 kg/m3 )(4201 J/kg ⋅ K) ⎛ 100o C − 80o C ⎞
q2,min = (100 L/min) ⎜ ⎟
(997.9 kg/m3 )(4186 J/kg ⋅ K) ⎝ 30o C − 5o C ⎠
q2,min = 77.7 L/min

Using this flow rate, we can calculate the Reynolds number for the cooling water in the outer
pipe,

ρVout Di ,out 4 ρ q2 Di ,out


Re 2 = =
μ (
πμ D 2 − D 2 o , in i , out
)
4 ( 997.9 [kg/m3 ])( 0.0776 [m3 /min]/60[s/min])
Re 2 =
(
π (1.0686 × 10−3 [Pa ⋅ s]) ( 0.1281 [m]) − ( 0.0889 [m])
2 2
)
Re 2 = 1.67 × 104

where we have taken the relevant length scale to be the outer diameter of the inner pipe, as this is
the length scale over which the majority of the heat transfer is occurring between the cooling
water and the pipe. The Prandtl number is (REMINDER: the properties are evaluated for the
fluid)

Pr2 =
μ Cˆ p
=
(1.0681×10 −3
[Pa ⋅ s]) ( 4186 [J/kg K])
k ( 0.5940 [W/m K]) .
Pr2 = 7.53

Using the Dittus-Boelter and Colburn correlations, we obtain

hout , DB Di ,out
Nu2, DB = = 0.023Re0.8 Pr 0.4
k
Nu2, DB = 0.023(1.67 ×104 )0.8 (7.53)0.4
Nu2, DB = 123.2
123.2 ( 0.5576 [W/m K])
⇒ hout , DB =
( 0.0901 [m])
⇒ hout , DB = 990.4 [W/m 2 K]

and
hout ,CB Di ,out
Nu2,CB = = 0.023Re0.8 Pr1/ 3
k
Nu2,CB = 0.023(1.67 ×104 )0.8 (13)1/ 3
Nu2,CB = 107.7
107.7 ( 0.5576 [W/m K])
⇒ hout ,CB =
( 0.0901 [m])
⇒ hout , DB = 865.7 [W/m 2 K]
Similarly, we can compute the heat transfer coefficient for the process stream in the inner pipe if
we calculate the Prandtl number for the inner pipe:

Pr1 =
μ Cˆ p
=
( 0.2425 ×10 −3
[Pa ⋅ s]) ( 4201 [J/kg K])
k ( 0.4787 [W/m K])
Pr1 = 2.13

Using this value of the Prandtl number, we obtain

hin , DB Di ,in
Nu1, DB = = 0.023Re0.8 Pr 0.4
k
Nu1, DB = 0.023(1.085 ×105 )0.8 (2.13)0.4
Nu1, DB = 332.2
332.2 ( 0.4787 [W/m K])
⇒ hin , DB =
( 0.0779 [m])
⇒ hin , DB = 2041 [W/m 2 K]

and

hin ,CB Di ,in


Nu1,CB = = 0.023Re0.8 Pr1/ 3
k
Nu1,CB = 0.023(1.085 ×105 )0.8 (2.13)1/ 3
Nu1,CB = 315.9
315.9 ( 0.4787 [W/m K])
⇒ hin ,CB =
( 0.0779 [m])
⇒ hin ,CB = 1941 [W/m 2 K]

We can also evaluate the thermal conductivity of the inner pipe wall at the appropriate average
temperature, which in is given by
T1 + T2 89.9o C+14.0o C2
Twall = Tw,in − Tw,out = =
2 2
Twall = 52.0o C
⇒ k wall = 16.52 [W/m ⋅ K]

We now have all we need to calculate the overall heat transfer coefficient, U o , for use in the model
equation for a counter-current heat exchanger,

−1
⎡ 1 aout ln ( rout / rin ) aout ⎤
UO = ⎢ + + ⎥ .
⎣ hout 2π Lk wall hin ain ⎦

Using ain = π Di ,in L, aout = π Di ,out L for the areas of heat exchange on each side of the inner pipe,
the above expression reduces to

Di ,out ln ( ri ,out / ri ,in )


−1
⎡ 1 Di ,out ⎤
Uo = ⎢ + + ⎥
⎣⎢ hout 2k wall hout Di ,in ⎦⎥
−1
⎡ ⎛ 0.0889 [m] ⎞ ⎤
⎢ ( 0.0889 [m]) ln ⎜ ⎟ ⎥
⎢ 1 ⎝ 0.0779 [m] ⎠ ( 0.0889 [m]) ⎥
⇒ U o , DB = + +
⎢ ( 990.4 [W/m 2 K]) 2 (16.52 [W/m ⋅ K]) ( 2041 [W/m K]) ( 0.0779 [m]) ⎥⎥
2


⎣ ⎦
⇒ U o , DB = 561.2 W/m 2 K

−1
⎡ ⎛ 0.0889 [m] ⎞ ⎤
⎢ ( 0.0889 [m]) ln ⎜ ⎟ ⎥
1 ⎝ 0.0779 [m] ⎠ + ( 0.0889 [m])
⇒ U o ,CB =⎢ + ⎥
⎢ ( 865.7 [W/m 2 K]) 2 (16.52 [W/m ⋅ K]) (1941 [W/m 2
K] ) ( 0.0779 [m] ) ⎥
⎢ ⎥
⎣ ⎦
⇒ U o ,CB = 516.1 W/m K 2

We can now use the value of U o to complete the design of the heat exchanger. We begin by
evaluating the heat load required to cool the process stream:

Q& load = ρ1q1Cˆ p1 (T1 − T1 f )


where we evaluate the thermal properties at the average process stream temperature of 68.3°C,
Q& load = ( 0.1 [m3 /min] / 60 [s/min])( 965.66 [kg/m3 ]) ( 4201 [J/kg K]) ( 20o C )
.
Q& load = 1.352 ×105 W

The appropriate model equation to determine the heat exchanger length is the level IV balance
given by equation 3.5.2

Q& load = U o ain ΔTlm = 2π Di ,in LU o ΔTlm


Q& load .
⇒L=
π Di ,inU o ΔTlm

The log mean driving force for the counter-current exchanger is

(100o C − 30o C) − (80o C − 5o C)


ΔTlm = = 72.47o C
⎛ 100 C − 30 C ⎞
o o
ln ⎜ ⎟
⎝ 80 C − 5 C ⎠
o o

Thus, the length of exchanger required is

LDB =
(1.352 ×10 [W]) 5

π ( 0.0779 [m]) ( 561.2 [W/m K]) ( 72.47 [ K ])


2

LDB = 13.58 m

and

LCB =
(1.352 ×10 [W]) 5

π ( 0.0779 [m]) ( 516.1 [W/m K]) ( 47.45 [ K ])


2

LCB = 14.77 m

c.) In order to calculate the length required for any cooling water exit temperature, one can use
the algorithm as it is done in the previous part, as follows:

Step 1: Pick a cooling water outlet temperature.

Step 2: Compute the log mean cooling water and wall temperatures using

T2 f − T2 T +T
T2 = , Twall = 1 2 .
⎛T ⎞ 2
ln ⎜ 2 f ⎟
⎝ T2 ⎠

Step 3: Compute the properties of the cooling water at T2 , and of stainless steel at Twall .
Step 4: Compute the flow rate of cooling water using a Level I balance

ρ1Cˆ p1 ⎛ T1 f − T1 ⎞
q2 = q1 ⎜ ⎟ .
ρ 2Cˆ p 2 ⎜⎝ T2,max − T2 f ⎟⎠

Step 5: Evaluate Re and Pr for the cooling water at T2 .

Step 6: Calculate h out using one of the two heat transfer correlations

Dittus-Boelter: Nu = 0.023 Re 0.8 Pr 0.4


Colburn: Nu = 0.023 Re 0.8 Pr1/ 3 .

Step 7: Calculate U o for the new values of h out and k wall

Di ,out ln ( ri ,out / ri ,in )


−1
⎡ 1 D ⎤
Uo = ⎢ + + i ,out ⎥ .
⎢⎣ hout 2kwall hout Di ,in ⎥⎦

Step 8: Calculate the log mean temperature difference

(T2 f − T1 ) − (T2 − T1 f )
ΔTlm = .
⎛ T −T ⎞
ln ⎜ 2 f ⎟⎟
⎜ T −T
⎝ 2 1f ⎠

Step 9: Calculate the new length of heat exchanger required

Q& load
L= .
π Di ,inU o ΔTlm

Using the above algorithm, the length of heat exchanger and the cooling water flow rate required were
calculated using both the Ditus-Boelter (solid line) and Colburn (dashed line) correlations as a function of
the outlet cooling water temperature.
Design summary for counter-current heat exchanger

d.) Looking at the above plot, we immediately see the tradeoff that is presented by the design. If
we wish to use a low flow rate of cooling water, we are forced to use a longer heat exchanger,
and vice versa. A good range to use for the final design would be on the flatter part of the
blue curve at higher cooling water outlet temperatures, where changes in the length of the
exchanger will not significantly affect the cooling water flow rate required.

The final design choice will ultimately come down to cost. It is clear that there is a utility cost
associated with the flow rate of cooling water, such that we wish to minimize the utility flow
rate. However, there will also be equipment costs for the heat exchanger and, more importantly,
utility costs to operate the pump that flows liquid through the exchanger, which will increase as
the pipe length increases. Thus, a careful design will take these economic points into
consideration and present a reasonable compromise between them.
7.1) qTCE = 100 L/min

qH2O = 100 L/min → contains 5 wt % acetone

dB = 2 mm, 5 mm, 1 cm

Compare the energy requirements (kW)

a) Tank-type mass contactor

From eq. (4) from the provided reference by Wichterle [Chem. Eng. Sci., 1995],
we find that

ε ∝ N3 ⇒ N ∝ ε1/ 3 .

Thus, for a given tank geometry with fixed impeller diameter (Lp), the impeller
Weber number (eq. 7.2.3) is given by

WeI ∝ N 2 ∝ ε 2 / 3 .

Using the correlation developed by Paul et al. (eq. 7.2.2) for the Sauter mean drop
diameter for fixed Lp, we see that

d 32 ∝ We−I 3/ 5 ∝ ε −2 / 5 .

Note that this is equivalent to the result by Hinze (eq. 7.1.4).

∴ε ∝ d −5 / 2
ε 2mm ∝ ( 2 mm )
−5 / 2
= 0.177
ε5mm ∝ ( 5 mm )
−5 / 2
= 0.0178
ε10mm ∝ (10 mm )
−5 / 2
= 0.00316
∴ ε 2 mm = 10 × ε5mm = 56 × ε10mm

b) Co-current horizontal tubular contactor

Here, the maximum drop diameter will be given by the correlation in eq (7.3.3).
In this correlation, all three dimensionless groups (Weber, Reynolds, and
Capillary numbers) will depend on the fluid velocity, and thus on the energy input
per unit mass. Because Wecrit depends on the fluid stress, we have (eq. 7.1.3)

WeI ∝ τ ∝ ( v I ) ∝ ε 2 / 3 .
2
We also know that the Capillary number is proportional to the fluid velocity, so
similarly we have

Ca ∝ v I ∝ ε1/ 3 .

Assuming that there is turbulent flow in the tubular exchanger (typically a


prerequisite for generating droplets), we can use eqs. (7.3.1,2) to express the
Reynolds number in terms of the energy input pet unit mass. Given eq. (7.3.2),
we have

f ∝ Re−1/ 4 .

Substituting this result into eq. (7.3.1), we obtain

( )
3
∆P ∝ f v I ∝ Re −1/ 4 ε

Finally, substituting this result into the expression for power per unit mass in
Section 7.3.1, we have

ε ∝ ∆Pq I ∝ ∆Pv I ∝ ( Re−1/ 4 ε ) ε1/ 3


⇒ Re ∝ ε 4 / 3

Substituting the proportionalities for Wecrit, Re, and Ca obtained above, we obtain

d max ∝ We3/crit5 Re−1/ 2 Ca −3/ 5 ∝ ( ε 2 / 5 )( ε −2 / 3 )( ε −1/ 5 )


⇒ d max ∝ ε −7 /15

Thus, for a contactor with fixed pipe dimensions, we have

∴ε ∝ d −15/ 7
ε 2mm ∝ ( 2 mm )
−15 / 7
= 0.226
ε5mm ∝ ( 5 mm )
−15 / 7
= 0.0317
ε10mm ∝ (10 mm )
−15/ 7
= 0.00720
∴ ε 2mm = 7.1× ε5mm = 31.5 × ε10mm
Problem 7.2

An approximate value of Km for transport of oxygen between the water and nitrogen phases can be
calculated by approximating the flow of the water stream over the 25mm Raschig rings as a film of water
traveling with bulk velocity vl in the positive-z direction over the surface of the ring in contact with a film
of nitrogen traveling with bulk velocity vg in the negative-z direction. Here, it is assumed that the
thickness of the water film is small with respect to the diameter of the ring, such that the physical
situation can be approximated by flow over a flat plate (shown below).

Water Nitrogen
Air y
z
Water
vg

Raschig
Ring
vl

For this composite film system, the overall mass transfer coefficient is given by

1
Km = .
1 M
+
kl k g

We are told to assume that the resistance to mass transfer is primarily in the liquid phase (as the
diffusivity of oxygen in gaseous nitrogen is several orders of magnitude greater than that in
water). Furthermore, we are told that penetration theory is accurate for estimating kl, such that

DOl 2 vmax
K m ≈ kl ≈ 2
πL

where L = 25 mm is the length of a single Raschig ring. The maximum velocity here is given by
the velocity of the nitrogen stream relative to the water stream. Thus,

vmax = vl + vg

DOl 2 ( vl + vg ) .
⇒ K m = kl = 2
πL

Below is a table of the material properties and flow rates of the nitrogen and water streams,
respectively (we have assumed that the properties of the nitrogen stream are similar to those for
air at 25°C and 1 atm).
Property Value
Nitrogen (g) Water (l)
Density, ρ 1.261 kg/m3 999.3 kg/m3
Oxygen diffusivity, DO2 1.76× 10 m /s 1.8× 10 -9 m2/s
-5 2

Superficial velocity, v ??? 0.25 m/s

We are given vl = 0.25 m/s. Thus, all that remains in order to calculate Km is superficial gas
velocity, vg. For this, we will use the correlation for interfacial area in a packed counter-current
tower, eq. (7.3.9):

av
= 0.54G ′0.31 L′0.07 .
aw

Here, we are told that the packed area of the exchanger is aw = 185 m2/m3. From Chemical
Engineers’ Handbook, the specific surface area of Raschig rings is av = 190 m2/m3. The
superficial liquid flow rate, L’, has units of [kg/m2s], and for the conditions given here is equal to

L′ = ρl vl = ( 999.3 kg/m3 ) ( 0.25 m/s ) = 249.8 kg/m 2s .

Rearranging eq. (7.3.9), we find that the corresponding superficial gas flow rate is

1
 ( 190 m 2 /m3 )  0.31
1
 aw  0.31
  = 2.287 kg/m 2s .
G ′ = ρ g vg =  0.07 
=
 0.54av L′   0.54 (185 m 2 /m3 )( 249.8 kg/m 2s ) 
0.07

 

Thus, the superficial gas velocity is

G′ ( 2.287 kg/m s )
2

vg = = = 1.81 m/s .
ρg (1.261 kg/m3 )
Therefore, using penetration theory, the overall mass transfer coefficient for transport of oxygen
from the water to the nitrogen stream is

Km = 2
DOl 2 ( vl + vg )
=2
(1.8 ×10 −9
m 2 /s ) ( 0.25 m/s + 1.81 m/s )
πL π ( 0.025 m ) .
K m = 4.35 ×10−4 m/s
7.3)

This problem illustrates the nature of log-normal size distributions and the difference between a
mean size and the sauter mean size, which is what is needed to calculate the area average size
per volume of the gas phase.

Two mixers produce mean bubble sizes (d50) of 5mm, but one produces a narrow distribution
(10%) and the other a broad distribution (50%) . The area a = 6 *Vd is given by the sauter
d 32
d3
mean size (equations on page 302) d 32 = . Using the relationship given on page 302,
d2
d 32 = d50 exp 5 2 ln (σ LN )  , where σ LN is the standard deviation of the log-normal distribution,
2

 
(which means, ln (σ LN ) = 0.1, 0.5 ), we find that the sauter mean diameter is nearly twice for
the 50% (broad) distribution as for the narrow distribution

For the same volume of the dispersed phase, this leads to a much smaller area (see the
spreadsheet calculation), again, by nearly a factor of two. The calculation is done for 100 liters
of dispersed phase. We can understand this result as there will be a large number of relatively
large liquid drops, with commensurately less area per volume. The distributions are plotted
(equation 7.0.1) in the spreadsheet below, where although the most probably drops are actually
smaller for the broadly dispersed system, there is less area as most of the volume is contained in
the larger drops.
Finally, note that the broadly dispersed distribution also contains smaller bubbles, which
may be entrained and will be harder to coalesce and remove in the separator.
d_50 (mm) 5 area (m^2) x f(x)dx(10%) f(x)dx(50%)
sigma_LN d_32 (mm) (based on V=0.1m^3) 0.1 0 4.06556E-13
0.10 5.13 117.04 0.3 1.7606E-171 3.54564E-07
0.50 9.34 64.23 0.5 5.925E-115 3.96095E-05
0.7 6.53702E-84 0.000500293
0.9 6.22E-64 0.002475041
Log Normal Distributions 1.1 5.98018E-50 0.00740006
1.3 1.21134E-39 0.016287287
1 1.5 8.87625E-32 0.029294442
probability

1.7 1.25374E-25 0.045772117


0.8 f(x)dx(10%) 1.9 9.82703E-21 0.064564992
f(x)dx(50%) 2.1 8.65203E-17 0.084344712
0.6 2.3 1.39724E-13 0.10386177
2.5 5.88932E-11 0.122091061
0.4 2.7 8.40976E-09 0.138288411
2.9 4.95624E-07 0.151987427
0.2 3.1 1.40391E-05 0.162963355
3.3 0.000215418 0.171183365
0 3.5
3.7
0.001969517
0.011586669
0.176755412
0.179882265
3.9 0.046701343 0.180823475
0 2 4 6 8 10 4.1 0.135812616 0.179865869
4.3 0.297498628 0.17730188
diameter (mm) 4.5 0.508915774 0.17341453
4.7 0.700932121 0.168467742
4.9 0.797721201 0.162700717
5.1 0.767051617 0.156325301
5.3 0.635194808 0.149525444
5.5 0.460564185 0.142458079
5.7 0.296638844 0.135254872
5.9 0.171863064 0.128024483
6.1 0.090561562 0.120855042
6.3 0.043825719 0.113816675
6.5 0.01964552 0.106963936
6.7 0.00821958 0.100338091
6.9 0.003231645 0.093969189
7.1 0.001201169 0.087877927
7.3 0.000424357 0.082077285
7.5 0.000143188 0.076573954
7.7 4.6346E-05 0.071369563
7.9 1.4446E-05 0.066461723
8.1 4.3516E-06 0.061844915
8.3 1.27086E-06 0.057511231
8.5 3.60874E-07 0.053450986
8.7 9.98984E-08 0.049653231
7.4) dB = 0.0087m

using Rushton impeller

motor → 50 rpm & eq 7.2.2 applies

impeller diameter?

would you purchase it?

what are inherent difficulties?

O2 transfer to H2O T = 20o C


well-mixed
Km a
= 0.0007
H2O → Co2 = 1.7 × 10-3 kg/m3 VI
6.5m
V I = 13.6m3
6.0m T = 11.9°C K m a = 0.0095
Cs I = 5.6 ×10 −3 kg / m3
q air = 3.9 × 10 -3 m 3/s

1.7m

d 32
= 0.053(We I ) −3/ 5
Lp
ρN 2 L3p
WeI = N → impeller speed (s -1 )
σ
rev  1min  rev
N = 50   = 0.833 Lp → impeller diameter (m)
min  60sec  s
N
d 32 = 0.0087m σ → surface tension  
m
For air-water:
ρ = 1.2 kg m3
 1×10 − s N   100cm 
σ = 71.97 dyn cm    = 0.07197 N m
 1 dyne   1m 
−3/ 5
d 32  ρN 2 L p 3 
= 0.053 
Lp  σ 
 
−3/ 5
d 32  ρN 2 
= 0.053   Lp −9 / 5
Lp  σ 
−3/ 5
 ρN 2 
d 32 = 0.053   L p −4 / 5
 σ 
−5 / 4
 
 
 
= ( Lp −4 / 5 )
d 32 −5 / 4

 2 −3/ 5 
 0.053  ρN  
   
  σ  
−5 / 4
 
 
 d 32 
Lp = 
2 −3 / 5 
 0.053  ρN  
   
  σ  
−5 / 4
 
 
 
 0.0087m 
Lp =  −3/ 5 
  1.2kg / m3 ( 0.833s −1 )  
2

 0.053   
  0.07197N / m  
   
L p = 1.5m

You might also like