Q1: d2x dx where x is the displacement of the mass; dx/dt is the velocity; d2x/dt2 is the acceleration.

Free vibration
m c kx 0
2 dt
dt
means no external force drives the system. The vibration is due to the initial conditions. In Mathcad, you need to define the
constants, for example m:= 1(mass) c: =1(damping)k:=100(skiffness). Then you need to define a vector function D which is a
y
1
D( t y ) 1
m c y1 k y0
function of t and y and represents the differential equation. You need to input the acceleration according to
d2x 1 dx
kx cy1 ky o
1
c
dt 2 m dt m In Mathcad, you need to input the initial displacement and velocity. i.e.y 0 : = 0.01, y1 : = 0.
Then, use the built-in function rkfixed to solve the differential equation R rkfixed( y 0 Tlength Npt D) The calculation
of y is from t =0 and t = Tlength = 10. Npt is the number of points calculated within the period t =0 and t = Tlength. D is the

function which represents the differential equation. Damping factor is the parameter to control the
vibration at the natural frequency. In practice, it is not easy to increase the damping factor of a building. It is because common
buildings are huge in size. Maintenance, installation, space for damping treatment would be the practical constraints. If we
increase the mass, what can be observed?(right line : m = 1, left dotted line: m = 2)If the excitation frequency is high, should we
increase the mass to make the amplitude smaller? (in higher frequency, increase mass leads to amplitude smaller, but in low
frequency it is no use)./ If we increase the stiffness, what can be observed ?( right line : k = 100, left dotted line: k = 200). *Only
when excitation frequency close to natural frequency, damping is useful!)
10
10 1
1

1
Q4: where x is the displacement of the mass;
AmpM 1 0.1
dx/dt is the velocity; d2x/dt2 is the acceleration.
AmpM 2
0.1
AmpK100
Free vibration means no external force drives the
AmpK200
0.01
system. The vibration is due to the initial
0.01
conditions
3 3
3.34101 10 3 6.67101 10 3
0 10 20 30 40 0 10 20 30 40
2 40 2 40

Q5: FDM. Firstly, we compare the differences between the solution forms of a numerical method and an analytic method. +
= 0 . Initial conditions : =0; u=1, at t =0 . The analytic solution form is () = . The solution is a function of time. If
we want to calculate u at time t =1, we just simply input t =1 into the function to find it (i.e. u(1) = cos 1). The central difference
method consists of expressing the velocity and acceleration at time tj in terms of displacement at time tj-1, tj and tj+1 using central
finite difference formulae. These are obtained by approximating the response curve shown in the following Figure, by a quadratic
polynomial within the interval (tj-1, tj+1) (three time steps are required). That is(2) = 2 + + . At =
, 0, . ) into the matrix form and solve it for a, b, c

1
()2 1 1 ()2 1 1
[ 0 0
1] () = ( ) > ( ) = [ 0 0 1] ( )
()2 1 +1 ()2 1 +1

Differentiating (2) give the velocity and acceleration at the j th time step

. If uj and uj-1 are known, uj+1 is known. Set j =0.

Eliminating u-1 gives

can be found by (5) and set j =0

Q2: WRM. Good approximation means minimum overall residual. Thus we need to minimize the overall residual by selecting
appropriate a1, a2

x
1 10
( x) x
x 1

( x) sin
x
N1( x) sin x N2( x) sin 2 x
( 10) 10.0909 ( 0) 2 2
10.0909 2 2 3
( x) x 2 N1( x) x ( x 10) N2( x) x ( x 10) N3( x) x ( x 10) ( x a1 a2) a1 N1( x) a2 N2( x) ( x)
10

( x a1 a2 a3) ( x) a1 N1( x) a2 N2( x) a3 N3( x) 2

d
R( x a1 a2) ( x a1 a2) ( x a1 a2)
2
R( x a1 a2 a3) ( x) ( x a1 a2 a3) dx

a1 0 a2 0.1 a3 0
a1 0 a2 0
Given
10

( x) ( x) a1 N1( x) a2 N2( x) a3 N3( x) N1( x) d x 0

0 Given

10

( x) ( x) a1 N1( x) a2 N2( x) a3 N3( x) N2( x) d x 0

1 1

( R( x a1 a2) ) N1( x) d x 0 ( R( x a1 a2) ) N2( x) d x 0
0

0 0
10

( x) ( x) a1 N1( x) a2 N2( x) a3 N3( x) N3( x) d x 0 0.271

find ( a1 a2)
0
0.029
0.15

Find ( a1 a2 a3) 0.012 x 0 .2 10 x 0 0.02 1
2.144 10 3

1
15

10
( x)
( x 0.271 0.029) 0.5

x 0.15 0.012 2.14410
3
5

0 0
0 2 4 6 8 10 0 0.2 0.4 0.6 0.8 1
x x

Q3: Comparison of FEM and WRM. ue,1 and ue,2 are similar to the unknown coefficients a m in the classical approximation. In
FEM ue,1 and ue,2 have more physical meaning, representing the nodal values. In classical approximation, am is just a numerical
coefficient, no physical meaning. What other advantage for FEM? Ans: the shape functions are standardized. WRM

M FEM N u N u N u There are two unknowns ue,1 and ue,2 in we which is

a m N m a1N1 a 2 N 2 .... e e,1 e,1 e, 2 e, 2 e, n e, n
m 1 n 1,2
linear. For higher order approximation, there are more unknowns in the approximation (i.e. we need more information about at
nodes).
w e ( x ) N e,1 ( x )u e,1 N e,2 ( x )u e,2
The subscripts 1 and 2 represent the LOCAL left and right ends of the element or Element nodes; ue,1, and ue,2 are the two
nodal values of the element, xi and xj are the GLOBAL locations of the left and right ends for each element; i = 0,1,2; j = 1,2,3

Original governing differential equation & boundary conditions. Select the approximation for this case Linear

Set up the residual equations and express them into the weak form
xj xj xj
xj
d 2 e dW d e d Rewrite the equations into the matrix format

W a
dx 2
c e q dx 0 a
dx dx
cW e dx Wqdx Wa e
dx
0
xi xi xi xi

k1e,1 k1e,2 u e,1 f e,1 Qe,1

e
k 2,1 k e2,2 u e,2 f e,2 Qe,2