Você está na página 1de 20

Engineering Mathematics

DT022/4
DT024/4
DT027/4

12907 MATH 4111


13020 CBEH 4312
13149 CBEH 4001

DUBLIN INSTITUTE OF TECHNOLOGY


BOLTON STREET, DUBLIN 1
-

Bachelor of Engineering (Honours) in


Mechanical / Structural / Civil Engineering

FOURTH YEAR: JANUARY 2010


SEMESTER 1

ENGINEERING MATHEMATICS

Pat Carroll M.Sc

Answer all of the following 4 questions


All questions carry equal marks
Time Allowed : 3 Hours

Formula Sheet and Mathematical Handbook

Formula

Sheet: Fourth Year January 2010

Q1: Central Differences

dy yi+1 -yi-1
=
dx
2h
d2 y 1
= 2 { yi +1 -2yi +yi-1}
2
dx h

FOURIER SERIES :

a sin ax+cos ax
a2

xcosax dx=

x sin ax dx=

-a cos ax+sin ax
a2

Half Range
Half Range cosine series :f(x) defined only from 0 to L

n
1
x
f(x)= a0 + an cos
L
2
1

n
2
2
a0 =
f ( x)dx; an
f ( x) cos
xdx
=

L
L0
L0
Half Range sine series :0 to L
n
n
2
x; b n = f ( x) sin
xdx
f(x)= bn sin
L
L0
L
1

2T 2T
xy is to be solved in the region (see sketch) defined by
+
=
x 2 y 2
the following boundaries:
1. The equation

T=3x
3

2
A

T=9 y2
1

T = y2

x
0

T
=-T
y

the x axis from x=0 to x=3 ; the y axis from y=0 to y=3; the lines y=3 and x=3.
The Boundary Temperatures are as follows:
T=3x along the upper edge y=3; T= 9-y 2 along y axis, T=y 2 along x=3;
T
and
=-T along the x axis.
y
(a) Use the central difference approximations for the second derivatives to formulate
a computation scheme for the temperatures at the internal mesh points.
(5)
(b) Use the central difference approximations for first derivatives and the given
T
condition
=-T to formulate a computation scheme for the temperatures at
y
points on the x axis boundary.
(5)
(c) Set up linear equations for each of these 6 points and express them in the form of
a matrix equation.
(10)
(d) Outline briefly what significant adjustment would required to this procedure if the
plate was a 4 x 2 rectangle
(5)

2. (i)

x2
0 (x 2 + 4)2 dx into a complex integral
clearly specifying the appropriate path.
(5)
(b) Locate and classify the poles and find the residue of each.
(5)
(c) Evaluate the integral;
(2)
(ii)
8z 2 2
(a) Find the poles of the complex function f(z) =
and plot them
z(z+1)(z-1)
on the Argand diagram.
(2)
(b) Identify which of these poles are inside the triangle joining the points
z=2, z=i, z=-1-i. Hence evaluate the integral
f(z) dz over this path. (6)

(a) Transform the real integral

(iii) Find the value of the complex integral

z-i =2

2z+1
dz where c is the path
z2 + z
(5)

3. An insulated copper bar of length 10 cm is heated in an oven till the


temperature distribution along the bar is given by u = 2x+30.The ends of the
o

bar are then placed in ice and maintained at 0 C .


Assume that the flow of heat in the region 0 < x < 10 can be modeled by the

heat diffusion equation

u 2 u
=
t
x 2

(i) Set out the boundary conditions at both ends and the initial condition at
t=0
(5)
(ii) Use the method of separation of variables to solve this problem.
(15)
(iii)What adjustment is required to adapt this solution to a different set of end
o
o
conditions, namely the end A maintained at 0 C and end B at 30 C .
(5)

(i) Expand the function f(x) =100x , 0 x 1 as a Fourier cosine series.


(see attached formula sheet )
(9)
2
2
T T
(ii) A general solution of the Laplace equation
0 for a
+
=
x 2 y 2
rectangular plate 0 x 1; 0 y 2 is:
T= (A1 cos kx + B1 sin kx)(A 2 cosh ky + B2 sinh ky )

You are to adapt this solution to match the following boundary conditions
(a) The edges along x=0, y=0 and x=1 are all insulated
(8)
(b) The temperature along remaining edge y=2 is given by the function
which was expanded in part (i) above, f(x) =100x , 0<x<1 (4)
(c) Explain briefly the principal difference if the 3 edges mentioned in
(4)
part (a) are all at 0o C .

SOLUTIONS

Question 1 Solution
(a)

Computational Molecule for a square:


Increment in x=h
Increment in y is same=h

f(a+h)-f(a)
h
1
f (a)= 2 [ f(a+h)-2f(a)+f(a-h) ]
h

f (a)=

1
2T
=
[ T(x+h,y)-2T(x,y)+T(x-h,y) ]
h2
x 2
1
2T
= 2 [ T(x,y+h)-2T(x,y)+T(x,y-h) ]
2
h
y
In this problem h=1
2T 2T
+
=xy
x 2
y 2
T(x+h,y)-2T(x,y)+T(x-h,y)+T(x,y+h)-2T(x,y)+T(x,y-h)=xy
4T(x,y)=T(x+h,y)+T(x-h,y)+T(x,y+h)+T(x,y-h)-xy
(b) and (c)
6 pts : A B C D E F

For pts E and F along x axis we need to create virtual pts above both : Below E call
this pt P while the point below F is called Q.

Pt A: x=1,y=2 xy=2
4A=5+B+3+C-2
4A-B-C=6

(1)

Pt B: x=2,y=2 xy =
4
4B=A+4+6+D-4
A-4B+D=-6

(2)

Pt C: x=1,y=1 xy =
1
4C =8 + D + A + E 1
A-4C+D+E=-7
(3)
Pt D: x=2,y=1 xy =
2
4D = C + 1 + B + F 2
B+C-4D+F=2

(4)

At E: h=1; x=1, y=0 xy =


0
T
P-C P-C
=T
=
=-E
y
2h
2
P=C-2E
4E = 0 + F + P + C xy = 0 + F + C 2 E + C 0
6 E 2C F =
0
(5)
at F: h=1, x=2,y=0 xy =
0
T
Q-D Q-D
=T
=
==-F
y
2h
2
Q=D-2F
4F=E+0+Q+D-0=E+D-2F+D=E-2F+2D
2D+E-6F=0
(6)

Put the equations in matrix form:


4 1 1 0

1 4 0 1
1 0 4 1

0 1 1 4
0 0 2 0
0 0 0 2

0 0 A 6

0 0 B 6
1 0 C 7
=
0 1 D 2
6 1 E 0

1 6 F 0

Notice this equation set has diagonal dominance so that the Gauss Seidel procedure
can be used to find the solutions.
(d)In a 4 x 2 rectangle the mesh could be based on 4 divisions along x axis and 2
along y axis.

h = 1 along x axis:
1
k= along y axis:
2
This would require an adjustment to the computational scheme:
1
=
h 2 1=
while k 2
4

2T 2T
+
=
xy
x2 y 2
1
1
T ( x + h, y ) 2T ( x, y ) + T ( x h, y ) ] + 2 [T ( x, y + h) T ( x, y ) + T ( x, y h) ] =
xy
2 [
h
k
1
1
=
1=
and 2 4
2
h
k
[T ( x + h, y) 2T ( x, y) + T ( x h, y)] + 4 [T ( x, y + h) T ( x, y) + T ( x, y h)]
T ( x + h, y ) + T ( x h, y + T ( x, y + h) + T ( x, y h) xy
10T ( x, y ) =
The mesh still involves 6 points and each point has the revised computational scheme
applied to it.

Solution to Q2
(i)

x 2 dx
I= 2
( x + 4) 2
0

x 2 dx
1
I=
2 ( x 2 + 4) 2
z 2 dz
Evaluate
C ( z 2 + 4) 2
z2 + 4 =
0 when (z+2i)(z-2i)=0
z=-2i and 2i
Only z=4i in the upper half plane

We have a double pole at z=2i


lim d

z2
2
Residue at z=2i=
(
z
2
i
)

z 2i dz
( z 2i ) 2 ( z + 2i ) 2
lim d z 2
=
z 2i dz ( z + 2i ) 2
lim ( z + 2i ) 2 2 z z 2 2( z + 2i )
{
}
=
z 2i
( z + 2i ) 4
16(4i ) (4)(8i )
=
(16)(16)
i
=
8
1
i
I= 2 i( ) =
2
8
8

(ii)

I =

8z 2 2
dz
z(z+1)(z-1)

c is path of given triangle.


poles at z=0 , z= 1, z=-1
two of these z=0 and z=1 are inside path.
Find residue at z=0 : Simple pole
R(0)=lim z 0
R(1)=lim z 1

8z 2 2
( z )
=2
z(z+i)(z-1)

8z 2 2

z
1
3
)
(
=
z(z+i)(z-1)

I = 2 i [sum of Residues ] = 10 i

(iii)
I=

2z+1
2z+1
dz=
dz
2

z +z
z(z + 1)
C

path is z-i = 2
2 poles: z=0, z=-1
z=0: z-i =
2
0-i =
1: pole is inside c
z=-1: z-i =
2 -1-i =2 : also inside c

2z+1
R(0)=lim z 0 ( z )
=1
z(z + 1)

2z+1
=
R(1)=lim z 1 ( z + 1)
1
z(z + 1)

I = 2 i [sum of Residues ] = 4 i

Solution to Question 3:
The general solution is the sum of 2 components v(x) and w(x,t).
v(x) represents the steady state element that is finally reached as t increases and is
hence independent of time.

The w(x,t) is the transient element which decays over time to zero.
The initial set up at t=0 can be expressed as:
u=2x+30

The amended set up is for both A and B at zero and is the steady state function v(x)
independent of time
This can be expressed as follows
v(x)=0

The general solution is :


u(x,t)= v(x)+w(x,t)
u (x,t)= 0+w(x,t)
at x=0: u=0
0 =0+w(0,t)
w(0,t)=0
Similarly at end point B: x=10
u(x,t)= 0+w(x,t)
x=10: 0=0+w(10,t)
w(10,t)=0
for time t=0: u=2x+30
u(x,0)= 2x+30=0+w(x,0)
w(x,0)=2x+30

The w(x,t) solution can, therefore, be obtained with zero boundary conditions at the
end pts, since the contribution from both these points are covered in the steady state
element v(x).

Transient Solution:

2w w
=
x2 t
subject to boundary conditions:
w=0 at x=0 for t>0
w=0 at t=10 for t>0
w=2x+30 at t=0
w=XT
dT 2 w
d2X
w
T
= X=
;
dt
dx 2
t
x2
d2X
dT
1 d 2 X 1 dT
X
T
=

=
dx 2
dt
X dx 2 T dt
both sides are equal to -k 2

1 d2X
d2X
2
=

+ k2X =
k
0
X dx 2
dx 2

=
X A1 cos kx + B1 sin kx
2
1 dT
A2 e-k t
=
k 2 dt T =
T dt

w=XT=e-k t (Acoskx+Bsinkx)
2

BC no 1: w=0 at x=0: A=0


w=e-k t (Bsinkx)
2

BC no 2: w=0 at x=10:
sin 10k=0 10k=n
n
k=
10

n 2 2
t
100

n
x with n=1,2,3..
10
General Solution: Sum of individual solutions

w=Be

w= e

n 2 2
t
100

sin

(Bn sin

n
x);n=1,2,3...
10
2

n 2 2
t
100

For time t=0: e


w= (B sin n5 x);n=1,2,3...

=1

This is a sine series: and matches the t=0 initial condition


(2x+30) through a Fourier half range sine series expansion of (2x+30)
for x=0 to 10:

Final Solution is:


2

u= Bn e

n 2
t
100

sin

nx
;
10

n=1,2,3..

10

2
n
with Bn = (2x+30) sin
x dx
10 0
10

(iii)
The adjusted set up is for A to be at 0 and B to be at 300 C and is the steady state
function v(x) independent of time
This can be expressed as follows
v(x)=3x

The changes for w(x,t) follow:

u(x,t)= v(x)+w(x,t)
u(x,t)= 3x+w(x,t)
at x=0: u=0 =0+w(0,t)
w(0,t)=0
Similarly at end point B: x=10
u(x,t)= 3x+w(x,t)
x=10: u=30=30+w(10,t)
w(10,t)=0
for time t=0: =4x
u(x,t)= 2x+30 =3x+w(x,0)
w(x,0)=30-x

Solution:
2

n 2
t
100

nx
;n=1,2,3..
10
10
2
nx
with
Bn
(30
x)
sin
dx
=

10 0
10

=3x+ Bn e

sin

Question 4: Solution
(i)

For a half range cosine series:

1
nx
f ( x) = a0 + an cos

2
l
n =1

l
1

2
2
x2
100 xdx =
200 100
=
a0
=
f ( x)dx
=
1 0
2 0
l 0

1
50
a0 =
2

an = 2 100 x cos n xdx


0

n sin n x + cos n x
= 200
n 2 2
0
1

200
(cos n 1)
n 2 2
= 0 if n is even

-400
if n is odd
n 2 2

Hence the half range Cosine series for 100x for 0<x<1 is
100x= 50-

400
1
1

cos x+ cos 3 x+
cos 5 x+...

9
25

(ii)
T= (A1 cos kx + B1 sin kx)(A 2 cosh ky + B2 sinh ky )

Boundary condition 1: Edge along x = 0 is insulated


T
0 along x=0
=
x
T
= (-kA1 sin kx + k B1cos kx)(A 2 cosh ky + B2 sinh ky )
x
Put x=0

This makes B1 =0
T=cos kx(A cosh ky + B sinh ky )

Second BC is insulation along y=0


T=cos kx(A cosh ky + B sinh ky )
T
= 0 along y=0
x
T
= cos kx(A k sinh ky + Bk cosh ky )
x
put y=0
This means B=0
T=A cos kx cosh ky

Condition no 3: Insulation along x= 1


T=A cos kx cosh ky
T
= 0 along x=1
x
T
= Ak sin kx cosh ky
x
Put this =0 for x=1
sin k=0
k=n
T=A cos n x cosh n y

This is valid for all integer values of n:


n=1, 2, 3 etc
The sum of all these individual solutions is also a solution:
T= A n cos n x cosh n y
For 4th edge we have : y=1 where T=100x
T= A cos n x cosh n along y=1

i.e ( A n cosh n ) cos n x = 100x

100x is expanded as a cosine series: Cn cos n x


1

2
where Cn = 100 x cos n x dx
10
Values of this integral is in part (i):
Cn = A n cosh n
Cn
cosh n
Final Soln : T= A n cos n x cosh n y

An =

where A n =
1

Cn
cosh n

and Cn = 2 100 x cos n x dx


0

in this case the expansion of 100 x along the y=2 boundary will be a sine series
rather than a cosine series.
T= (A1 cos kx + B1 sin kx)(A 2 cosh ky + B2 sinh ky )
BC no 1: u=0 along x=0:
A1 =
0
T= sin kx(A cosh ky + B sinh ky )
Similarly: A=0 because of BC no 2 for y=0:
T= Bsin kx sinh ky
Also : as before k=n

T= Bn sin n x sinh n y
1

hence along y=1:

T= Bn sin n x sinh n
1

This means a sine series expansion for the y=2 edge


is appropriate

Você também pode gostar