Pset 2 solutions

3.044 2013
Problem 1
Recall our old friend from pset 1, the solution to the heating/cooling of a 1D
sheet with fixed temperature boundary conditions:
theta1D@x_, t_, nmax_D :=
SumA4 ê HPi nL Sin@n Pi xD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E

To produce the solution for a infinite square prism, we can use superposition,
and Q(x, y, t) = Q(x, t) Q(y, t). To get Q(y, t), we can simply swap out y for x
because the boundary conditions are the same in x and y. So, Q(x, y, t) is:
theta2D@x_, y_, t_, nmax_D :=
SumA4 ê HPi nL Sin@n Pi xD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E
SumA4 ê HPi nL Sin@n Pi yD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E

To produce the solution for a cube, we can use superposition again: Q(x, y, z, t)
= Q(x, t) Q(y, t) Q(z, t).
theta3D@x_, y_, z_, t_, nmax_D :=
SumA4 ê HPi nL Sin@n Pi xD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E
SumA4 ê HPi nL Sin@n Pi yD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E
SumA4 ê HPi nL Sin@n Pi zD ExpA- n2 Pi2 tE, 8n, 1, 2 nmax - 1, 2<E

Now to plot it, using contours, at various times. Assuming that we are cooling an
initally hot body, hot colors = hot contours (pink is hottest) and cold colors = cold
contours (purple is coldest) (flip the color scheme if we are heating an initally
cold body):
First, define the desired contours, contour colors, dimensionless times, and how many terms to use in
the series at each time:
contours = 80.01, 0.1, 0.25, 0.5, 0.75, 0.9, 0.99<;
colors = 8Purple, Blue, Green, Yellow, Orange, Red, Pink<;
times = 80.001, 0.005, 0.01, 0.1, 0.2<;
terms = 830, 20, 10, 2, 2<;
Then make a list of equations at each time:

Printed by Wolfram Mathematica Student Edition

1
2 pset2solns.nb

plot1d = Table@ Htheta1D@x, t, terms@@jDDD ã contours@@iDDL ê. t Ø times@@jDD,

8j, Length@timesD<, 8i, Length@contoursD<D;

plot2d = Table@ Htheta2D@x, y, t, terms@@jDDD ã contours@@iDDL ê. t Ø times@@jDD,

8j, Length@timesD<, 8i, Length@contoursD<D;

plot3da =
Table@ Htheta3D@x, y, 0.25, t, terms@@jDDD ã contours@@iDDL ê. t Ø times@@jDD,
8j, Length@timesD<, 8i, Length@contoursD<D;

plot3db =
Table@ Htheta3D@x, y, 0.5, t, terms@@jDDD ã contours@@iDDL ê. t Ø times@@jDD,
8j, Length@timesD<, 8i, Length@contoursD<D;

plot3d = Table@ Htheta3D@x, y, z, t, terms@@jDDD ã contours@@iDDL ê. t Ø times@@jDD,

8j, Length@timesD<, 8i, Length@contoursD<D;
Create some labels:
fontsize = 28;
labelRow = 8Text@Style@"Time", fontsizeDD,
Text@Style@"1D", fontsizeDD, Text@Style@"2D", fontsizeDD,
Text@Style@"3D, z = 0.25", fontsizeDD, Text@Style@"3D, z = 0.5", fontsizeDD<;
labelColumn = 8Text@Style@"t = 0.001", fontsizeDD,
Text@Style@"t = 0.005", fontsizeDD, Text@Style@"t = 0.01", fontsizeDD,
Text@Style@"t = 0.1", fontsizeDD, Text@Style@"t = 0.2", fontsizeDD<;
And put it all together in a graphics grid:
GraphicsGrid@Join@8labelRow<, Transpose@
8labelColumn, Table@ContourPlot@Evaluate@plot1d@@jDDD, 8x, 0, 1<, 8y, 0, 1<,
ContourStyle Ø colors, ContourLabels Ø TrueD, 8j, Length@timesD<D,
Table@ContourPlot@Evaluate@plot2d@@jDDD, 8x, 0, 1<, 8y, 0, 1<,
ContourStyle Ø colors, ContourLabels Ø TrueD, 8j, Length@timesD<D,
Table@ContourPlot@Evaluate@plot3da@@jDDD, 8x, 0, 1<, 8y, 0, 1<,
ContourStyle Ø colors, ContourLabels Ø TrueD, 8j, Length@timesD<D,
Table@ContourPlot@Evaluate@plot3db@@jDDD, 8x, 0, 1<, 8y, 0, 1<,
ContourStyle Ø colors, ContourLabels Ø TrueD,
8j, Length@timesD<D <DD, ImageSize Ø FullD

Printed by Wolfram Mathematica Student Edition

2
 pset2solns.nb 3

3D, z 3D,
Time 1D 2D
= 0.25 z = 0.5
1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

t = 0.001 0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

t = 0.005 0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

t = 0.01 0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

t = 0.1 0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

t = 0.2 0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Notice that as you add dimensions, the heat transfer goes faster. This makes
sense: adding ways for heat to get in/out should accelerate the process. In 3D,
that the center is a little behind the z = 0.25 because it takes longer for heat to
penetrate all the way to the center.

Problem 2 Printed by Wolfram Mathematica Student Edition

3
4 pset2solns.nb

Problem 2
Problem: Find the time required to heat a 25cm by 1m (long in the third dimen-
sion) steel billet to 900C at its center in a 1000C furnace.

Geometry: the center of the slab is at (0, 0). Lx = 0.125m, Ly = 0.5m.

BC’s: There are a total of 4 boundary conditions. Convection at x = Lx and y =

Ly , symmetry J ¶∂T
¶∂x
= 0 and ¶∂T
¶∂y
= 0N at x = 0 and y = 0.

IC: at t = 0, T ~ 0C inside the billet, the furnace is at 1000C (just like we

assumed in class)

Govening equation:
h Lx I100 W ëm2 K M H0.125 mL
Bix = k
= H35 W êmKL
= 0.36
h Ly I100 W ëm2 K M H0.5 mL
Biy = k
= H35 W êmKL
= 1.4
Therefore, we must use transient solutions from the charts for both dimensions,
plus superposition.
Q(x, y, t) = Q(x, t) Q(y, t)

Use the Poirier & Geiger chart for a 1D plate, x/L = 0, for both dimensions.

Solve:
The goal is to find when Q(x, y, t) = 0.1. Take a guess that it will be less time
than it was for the 1D case we analyzed in class, which was 22300 sec, so try
15000 sec:
at I5.6 x 10-6 m2 ësM H15 000 sL
Fox = = = 5.4
Lx 2 H0.125 mL2
at I5.6 x 10-6 m2 ësM H15 000 sL
Foy = = = 0.34
Ly 2 H0.5 mL2

read Q’s off the chart for the respective Biot numbers:
Qx º 0.18
Qy º 0.82
Qcenter = Qx Qy = 0.15

This is still a little too cold, so try a later time of 18000 sec:
at I5.6 x 10-6 m2 ësM H18 000
PrintedsL
by Wolfram Mathematica Student Edition
Fox = = = 6.5 4
Lx 2 H0.125 mL2
at I5.6 x 10-6 m2 ësM H18 000 sL
Foy = 2 = 2 = 0.40
 Qx º 0.18
Qy º 0.82
pset2solns.nb 5
Qcenter = Qx Qy = 0.15

This is still a little too cold, so try a later time of 18000 sec:
at I5.6 x 10-6 m2 ësM H18 000 sL
Fox = = = 6.5
Lx 2 H0.125 mL2
at I5.6 x 10-6 m2 ësM H18 000 sL
Foy = = = 0.40
Ly 2 H0.5 mL2

read Q’s off the chart for the respective Biot numbers:
Qx º 0.14
Qy º 0.77
Qcenter = Qx Qy = 0.11

This is still just a little too cold, so try a later time of 19000 sec:
at I5.6 x 10-6 m2 ësM H19 000 sL
Fox = = = 6.8
Lx 2 H0.125 mL2
at I5.6 x 10-6 m2 ësM H19 000 sL
Foy = = = 0.43
Ly 2 H0.5 mL2

read Q’s off the chart for the respective Biot numbers:
Qx º 0.13
Qy º 0.75
Qcenter = Qx Qy = 0.10

This time is equivalent to 5.3 hours.

In class, we learned that the time to heat a 25cm square billet (long in the third
dimension) is about 3.1 hours, and the time to heat a 25cm thick billet (long in
the other two dimensions) is about 6.2 hours. Our analysis showed that the time
to heat a 25cm x 100cm billet (long in the third dimension) is in between these
values, at 5.3 hours.
Time to reach 900ÎC:

3.1hrs 5.3 hrs 6. hrs

These times make sense because the contribution in heat from the sides
decreases as as the width of the billet increases.

Problem 3

Printed by Wolfram Mathematica Student Edition

5
6 pset2solns.nb

Problem 3
We want a solution for a 1D slab, with uniform heat generation within, and con-
vection on the surfaces.

Carslaw & Jaeger, Chapter on linear heat flow in a solid bounded by two planes,
section 3.14: the slab with heat produced within it, equation 12, p. 132:
1 L h x h
T= 2k L
:Hx + LL Ÿx „ h Ÿ-L AHxL „ x + Hx - LL Ÿ-L„ h Ÿ-L AHx L „ x >

where T = temperature, k = thermal conductivity, A(x) is the rate of heat produc-

tion (which we will take to be a constant in our case that only depends on the
current), and L = sheet thickness. h and x are “dummy” variables for integration,
so they effectively go away and don’t mean anything. You are not required to do
the next bit, however, it is cool:

In our case, A(x) = A = constant rate of heat production, so

x h x 2 h=x
Ÿ-L„ h Ÿ-L AHxL „ x = A Ÿ-LHh + LL „ h = A Ih ë 2 + L hM h=-L =
A Ix 2 ë 2 + L x + L2 ë 2M
and
L h L
Ÿx „ h Ÿ-L AHxL „ x = A Ÿx Hh + LL „ h =
h=L
A Ih2 ë 2 + L hM h=x = A I-x 2 ë 2 - L x + 3 L2 ë 2M.

Therefore,
T = 2 k1 L 9Hx + LL A I-x 2 ë 2 - L x + 3 L2 ë 2M + Hx - LL A Ix 2 ë 2 + L x + L2 ë 2M=
A
= 2k L
9L3 - L x 2=
A
= 2k
9L2 - x 2=
where A is the rate of heat generation, so it is just a parabola! This is a solution
that you can do by hand, it turns out.

Problem 4
We want a solution for a single wall, so that is a 1D slab, with a sinusoidally
varying temperature at one surface.

C & J, Chapter on linear heat flow in a solid bounded by two parallel planes,
Section 3.6: the slab with periodic surface temperature, equation 5, page 105:

Printed by Wolfram Mathematica Student Edition

6
 We want a solution for a single wall, so that is a 1D slab, with a sinusoidally
varying temperature at one surface.
pset2solns.nb 7

C & J, Chapter on linear heat flow in a solid bounded by two parallel planes,

n H-1Ln Ikn2 p2 sin e - wL2 cos eM 2

p 2 t ë L2
T = A sin(wt + e + f) + 2pk ڦ
n=1 2 4 4 2 2 sin npx
e-kn + Tinside
k n p +w L L

Where T = temperature, L = wall thickness, the temperature is kept at Tinside at

2 kx + cos 2 kx 1ê2 k x H1+i L 2
x=0, A = 8 cosh
cosh 2 kl + cos 2 kl
= , f = arg : cosh
cosh kl H1+i L
>, k = I 2wk M , e = phase shift
(which we may take to be zero), w = frequency of temperature variation (which
we take to be 1 day), and k = thermal diffusivity. Note that this is essentially what
you’d expect: if the surface temperature is sinusoidal, the solution is sinusoidal
in space, but exponentially damped in time, i.e., the effects of the outside temper-
ature are exponentially damped as you move into the thickness of the wall.

Problem 5
We want a solution for a solid sphere, with uniform heat generation, and a radia-
tive boundary condition at the surface, because the antenna can cool by radiat-
ing into the vacuum of space, which is at ~0 K. There is no convection in space,
so our answer better not have a convective heat transfer coefficient in it.

C & J, chapter on the flow of heat in a sphere and cone, section 9.8: the sphere
0 § r < a with heat generation, equation 19, page 245:

2
A0 2 h a2 A0 sin r an e-k an t
T= 6hk
2 2
9hIa - r M + 2 a= - rk
ڦ
n=1 a 2 Aa2 a 2 + a h Ha h-1LE sin a a
n n n

where T = temperature, A0= constant heat generation, h = M number divided by

the characteristic length so that it has units of 1/m (note: this is really hard to
find! It’s kind of explained on page 19, but it takes some digging and thinking), k
= themal conductivity, k = thermal diffusivity, a = radius of the antenna, r = radial
coordinate, and an are the roots of
a a cot a a = 1 - a h
for n = 1, 2, ....

Problem 6
i. Geometry & BCs
-Center of laser spot is origin
-laser spot has radius a
-symmetry at r = 0: ¶∂T/¶∂r = 0.
Printed by Wolfram Mathematica Student Edition

7
8 pset2solns.nb

-laser spot has radius a

-symmetry at r = 0: ¶∂T/¶∂r = 0.

ii. Quantitative simplification

Calculation the Biot number for the z direction:

Bi = h L / 2 k = (10 W ë m2K) (0.05m) / (35 W/mK) = 0.014

The Biot number is small -> we can neglect heat transfer via convection, all heat
transfer will occur in the interior of the steel slab.

Therefore, look for a solution T(r, z). Change the difficult convective boundary
condition to zero flux at the z=0, r>a surface.

Note: some folks might be confused why our solution will still depend on z, even
though the Biot number is small in z. This is because the Biot number compares
conduction in z with convection at the surface. Convection is so slow that it won’t
contribute at all to the transfer of heat. However, conduction in z occurs at the
exact same speed as conduction in r because the slab is isotropic. So, if our
solution depends on r, it better depend on z too, because conduction is equally
fast in r and z. All that we can conclude from a small Biot number here is that we
can neglect heat loss to the air, all heat will be dissipated internally.

Printed by Wolfram Mathematica Student Edition

8
 conduction in z with convection at the surface. Convection is so slow that it won’t
contribute at all to the transfer of heat. However, conduction in z occurs at the
exact same speed as conduction in r because the slab is isotropic. So, if our
pset2solns.nb 9

solution depends on r, it better depend on z too, because conduction is equally

can neglect heat loss to the air, all heat will be dissipated internally.

iii. Find the steady-state solution, plot it, and identify melt (if any)
What is the solution going to look like? Many people expect that you should treat
the laser as a point source and treat the body as finite in z. But, the thing that
sets the relevant length scale in any heat transfer problem is not the dimensions
of the objects involved, but the dimensions of the heat transfer process. The
heat is coming in over a little disk of radius 0.1mm and spreading radially out-
ward in r and z. It doesn’t care about the surface at z = 10cm because the length
scale of the actual heat transfer is orders of magnitude smaller. As far as the
heat is concerned, that surface is too far away to matter, so you can treat it a
semi-infinite in z.

This relevant solution can be found in Carslaw & Jaeger, the chapter on the flow
of heat in regions bounded by surfaces of the cylindrical coordinate system,
section 8.2: the steady temperature in an infinite or semi-infinite medium due to
heat supply over a circular area, page 216, equation 14, “the region z > 0 with
constant flux Q over the circular area r < a and zero flux over r > a”:

Q a ¶ -l z
T= k Ÿ0
e J0Hl r L J1Hl aL ê l „ l

where T = temperature, Q = heat flux over the circle 0 < r < a, k = thermal conduc-
tivity, l = dummy variable, z = coordinate into the slab, r = radial coordinate, J0,
J1= zeroth and first kind Bessel J functions. Now, plot this result:
The heat flux is equal to the wattage of the laser divided by the laser spot area
to yield W ë m2 :
Q = 1000 ë I0.00012 PiM

3.1831 µ 1010

Create a function T(r, z) using our solution:

temperature@r_, z_D := Q 0.0001 ê H2 µ 37L
NIntegrate@Exp@- l zD BesselJ@0, l rD BesselJ@1, l 0.0001D 1 ê l, 8l, 0, 1000<D

Plot T(r, z):

trz = Plot@Table@temperature@r, zD, 8z, 80, 0.0005, 0.001, 0.0025, 0.005<<D,
8r, 0, 0.005<, AxesLabel Ø 8Style@"r HmL", 20D, Style@"T H°CL", 20D<D;

Printed by Wolfram Mathematica Student Edition

9
10 pset2solns.nb

Show@trz, Plot@1500, 8r, 0, 0.005<, PlotStyle Ø RedD,

Graphics@8Black, Text@Style@"z=0", 14D, 80.001, 2100<D,
Text@Style@"z=0.5mm", 14D, 80.001, 1700<D,
Text@Style@"z=1mm", 14D, 80.001, 1150<D, Text@Style@"z=2.5mm", 14D,
80.001, 850<D, Text@Style@"z=5mm", 14D, 80.001, 500<D<DD

T H°CL
z=0
2000

z=0.5mm

1500

z=1mm
1000

z=2.5mm

500 z=5mm

0.001 0.002 0.003 0.004 0.005

r HmL
In the above plot, we see that the temperature exceeds the melting point,
1500C, for about 0.75 mm into the slab in z and for a radius out to about 2.2 mm
at the surface (ignoring the latent heat, which would make the melt puddle even
smaller). There is no way this laser could cut this slab of steel.

Printed by Wolfram Mathematica Student Edition

10
MIT OpenCourseWare
http://ocw.mit.edu

3.044 Materials Processing

Spring 2013

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.