ASEN5022 HW04 2004 Sol
ASEN5022 HW04 2004 Sol
ASEN5022 HW04 2004 Sol
From Lecture Notes 13, we find (There were two terms missing in equation (11) of that lecture
notes, viz., kw1 and kw2 terms ):
t2
{[E I w(L , t)x x ]x + kw2 w(L , t) Q 2 } w(L , t)dt
t1
t2
+ {[E I w(0, t)x x ]x kw1 w(0, t) + Q 1 } w(0, t)dt
t1
t2
{[E I w(L , t)x x ] + k 2 w(L , t)x M2 } w(L , t)x dt
t1 (1)
t2
+ {[E I w(0, t)x x ] k 1 w(0, t)x + M1 } w(0, t)x dt
t1
t2 L
{ m(x)w(x, t)tt + [E I w(x, t)x x ]x x
t1 0
f (x, t) } w(x, t) d xdt = 0
The above Hamiltons principle yields the governing equation of motion of the form
Thus, the governing differential equation and the four natural boundary conditions are as follow.
1
m(x) w(x, t)tt + [E I w(x, t)x x ]x x = 0
{[E I w(L , t)x x ]x + kw2 w(L , t) } =0
{[E I w(0, t)x x ]x kw1 w(0, t) } =0 (4)
{[E I w(L , t)x x ] + k 2 w(L , t)x } =0
{[E I w(0, t)x x ] k 1 w(0, t)x } =0
In order to formulate the beam vibration problem, first, we assume w(x, t) in the form
2 m
W (x) + W (x)x x x x = 0, 0x L
EI
k
W (0)x x x w1 W (0) =0
EI
k 1
w(0)x x W (0)x =0 (6)
EI
k
W (L)x x x + w2 W (L) =0
EI
k 2
w(L)x x + W (L)x =0
EI
= L , ki = ki /(E I /L)
2
After computing the first mode of the 5 ideal boundary conditions using BeamModeShapeFinder.m,
the mode and mode shape of the present problem are found to fall between the fixed-fixed and
the simply supported-simply supported beams as shown in Figure 1 below. Notice the frequency
parameter L of the problem is L = 4.25 whereas the fixed-simple support and the fixed-fixed
beams are L = 3.9266 and L = 4.73, respectively. In addition, the maximum mode shape
amplitude of the problem occurs at x/L = 0.55 which is between that of the fixed-simply supported
beam(x/L = 0.60) and of the fixed-fixed beam (x/L = 0.5).
0.9
0.8
0.7
0.6
Mode shape
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Beam span
Fig. 1 Mode and mode shapes of fixed-fixed and fixed-simple support beams
Hence, the stage is set for finding the rotational spring parameters k 1 and k 2 . A possible beam
model is shwon in Figure 2 below.
z
w
3
4.2 Model this beam in terms of discrete springs and masses employing the discrete modeling
approaches discussed in the class. Can you have a rough estimate of the boundary springs from
your discrete model? Show your rationale as to how well your proposed discrete model can guide
you to a reasonable set of model parameters.
mx + kx = 0, x = [1 , 2 ]T
m 3 4 3
m= , m = A
420 3 4 (8)
EI 4 + 1 2 k
k= , =
2 4 + 2 EI
z
w
One FEM beam elements with rotational boundary springs
1 2
k1 k 2
x
EI, m(x)
[k 2 m]x = 0
(9)
4 + 1 2 ( L)4 4 3 2 m
det| | = 0, 4 =
2 4 + 2 420 3 4 EI
4
1 Search for rotational springs with 2-dof model
10
Two end rotations of a fem element and two rotational support springs
Determinant Magnitude
0
10
-1
10
-2
10
3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4
k theta 1 = k theta 2
In order to get an idea of an order of magnitude estimate, we set = 1 = theta2 and searched
for its value with the desired L = 4.25. The result is shown in Figure 3, which shows that for the
case of 1 = theta2 it hovers around
. Note, however, that this equal rotational springs will result in a symmetric mode shape, meaning
that the maximum mode shape amplitude will occur at x/L = 0.5.
z
w
Two FEM beam elements with rotational boundary springs
k1 0.55 L 0.45 L k 2
x
EI, m(x)
5
mx + kx = 0, x = [1 , wm , m , 2 ]T
m 1 2 m 1 21
13m 1 1
1
0
105 420 140
13m 1 1 13(m 1 +m 2 ) 11(m 2 2 m 1 1 )
13m 2 2
420
m=
35 210 420
m 2
1 1 11(m 2 2 m 1 1 ) (m 2 22 +m 1 21 )
m 2 22
140 210 105 140
2 2
m 2 22 m 2 22
0 13m
420
140 105
(11)
4k1 21 + k 1 6k1 1 2k1 21 0
6k1 1 12(k1 + k2 ) 6(k1 1 k2 2 ) 6k2 2
k=
2k1 21 6(k1 1 k2 2 ) 4(k1 21 + k2 22 ) 2k2 22
0 6k2 2 2k2 22 4k2 22 + k 2
EI EI
m 1 = A1 = m1 , m 2 = A2 = m2 , k1 = , k2 =
31 32
4.4
4.3
4.2
beta*L magnitude
4.1
beta*L value without rotational spring = 3.1488
4
3.9
3.8
3.7
3.6
3.5
2 4 6 8 10 12 14 16 18 20
Rotational spring k1 ( k2 = (0.45/0.55) *k1 ) value
6
4.2.3 Classical approach with two end rotational degrees of freedom: A classical approach
(i.e., assumed mode approximation) of the following form may be utilized for the present purposes:
and setting (x, t) = w(x, t)x , [c1 (t), c2 (t)] can be expressed by the two end rotations, [ 1 (t) =
(0, t), 2 (t) = (L , t) ]. Therefore, the total kinetic and potential energy can be obtained by
L
T = 1
2
m[w(x,
t)]2 d x, m = A
0
L
(14)
V = 1
2
E I [w(x, t)x x ] d x + 2 1
k 2
2 1 1
+ 1
k 2
2 2 2
0
which is a function of [1 , 2 , 1 , 2 ].
Carrying out the necessary variational process, one finds the following homogeneous equation:
mx + kx = 0, x = [1 , 2 ]T
5/32 3/32
m=m
3/32 5/32
(15)
5k/8 + k 1 3k/8
k=
3k/8 5k/8 + k 1
m = m L 3 / 2 , k = 2 E I /L
4.25
beta*L magnitude
4.2
4.15
4.1
4.05
5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7
Rotational k1 [k2 = (0.45/0.55)*k1] value
Fig. 5 Classical assumed mode model with two rotational support springs
7
(1 )(2do f classical) 6.4 (16)
4.3 Utilizing the technical insight you gained from the above two tasks, develop a strategy of how
you can employ the continuum model derived in problem (4.1) to arrive at the uncertain boundary
condition parameters.
An analysis for the estimation of the support rotational springs from Problem 4.2 has provided the
following:
One-element with two end rotational DOFs: 1 3.45
Two-element with one interior translation and two end rotational DOFs: 1 12.0 (17)
Classical approach with two end rotational DOFs: 1 6.4
It should be noted that the above estimates are based on the assumption that the peak amplitude of
the mode shape occurs at the beam mid-span.
Armed with these estimates, an iterative search for 1 and 2 was launched by utilizing the theoretical
(4x4)-characteristic matrix routine, CmatrixBeamGeneral.m, while varying 1 in the outer loop
and 2 in the inner loop. A limiting solution has been found if 1 = k 1 L/E I is chosen to be
excessively large, e.g.,
1 = 106
k 2 L (18)
2 = 4
EI
0.9
0.6
Mode shape
0.5 L = 4.25
0.4 k1/(E I /L) = 192
0.3 k2/(E I /L) =4.47
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
Beam span
Fig. 6 Classical assumed mode model with two rotational support springs
A further iterative search discovered that a solution (not unique!) has been discovered as
k 1 L
1 = 192
EI (19)
k 2 L
2 = 4.35
EI
8
The mode shape of this choice is plotted in Figure 6.
k1 L
26.5
EI (20)
k2 L
1.95
EI
which shows that, for a drop of frequency 11 percent (note that the frequency is proportional to ( L)2 ), an appreciable
reduction of the support springs will result. In practice, the frequency drop is an indication of how much the bridge has
deteriorated, a damage indicator. It is for this reason the support spring models play a crucial role in assessing the health of
the bridge via dynamic testing. }
4.3 Discuss complementary features of both the rigorous continuum modeling approach and ruthless discrete modeling
approaches, if any. What have you learned?
The estimate of 2 = k2 L
E I by the two to four-DOF models has been proved to be fairly useful. Judging from that fact that
the peak amplitude of the mode shape occurs at x?L = 0.55, we see that the left-end rotational support spring (1 ) must
be stiffer than the right-end support spring (2 ). In using the exact continuum formula (7), we have found that in fact one
could treat the left-hand boundary condition is very close to an ideal clamped or fixed support. Therefore, if one is charged
to inspect or repair the bridge, one should carefully look into any loosened connections on the right-side of the bridge and
the right-end end condition.
Finally, with the preliminary estimate, a refined finite element model can be constructed, which can be used to study not
only the end conditions but also the many connectors in the bridge structures. For this to be meaningful, there should be
more measured data along the beam span. This subject is called structural health monitoring which has been a growing
activity around the world.