You are on page 1of 8

MSD 2008-2009 solutions

3 EI 3.207 × 109 0.025 ×0.01253


1a) k = 3 = × =20200N/m
L 0.53 12

Where I =4.07× 10−9

Beam mass=7850× 0.025× 0.0125 ×0.5=1.22 kg (Note length of beam is 0.5m not 1m)

1.22
Effective mass=0.5+ =0.805 kg
4

k 20200
Running speed =
√ √
meff
=¿
0.805
=158.4 rad s−1=25.2 Hz=1500rev /min ¿

Likely errors: assumption of rigid clamp and rigid end mass; rigid attachment of end mass; end
mass acting as point mass exactly at end of beam

b) Force=mh ω2=0.805× 0.02× 158.42=404 N

Total Force=404 N ×2=808 N

0.0125
404 ×0.5 ×
c) Max stress = M y max 2
= =311 MPa
I 4.07 × 10−9

d) Stress is over fatigue limit and given major uncertainties in (a), plus likely effect of fretting at
support, very likely to suffer fatigue problems

e) Here, they are talking about the cantilever beam length decreasing from 0.5m to 0.4m.
0.5 3
Stiffness increases by =1.95
0.4

Therefore, to keep same tuning frequency, effective mass increases to 1.95 ×0.805=1.57 kg

0.4
Beam mass returns to 1.22 × =0.476 kg
0.5

0.976
New end mass = 1.57− =1.33 kg
4

f) Force is constant (speculation: effective mass increases, but amplitude of displacement, h,


decreases) but moment reduced by 20% as beam length reduces.

Therefore new max stress 0.8 ×311=244 MPa


Therefore less likely to suffer from fatigue but still allow factor of safety given approximation
made

g) Increases absorber mass will increase separation of resonant frequencies

ω 2n=ω 20 ¿ )

2 0.5
Gap between resonances=ω0 2 μ+
μ
4 (√ )
2a) Model as Cantilever (Clamped-Free)

dv
Clamped end, x=0: v=0 , =0
dx

x=0 v=0 => C1 + C3 = 0 => C3 = -C3

x=0, dv/dx=0 => C2 + C4 = 0 => C2 = -C4

v=C1 cosh λx+C 2 sinh λx−C1 cos λx−C 2 sin λx

d2 v d 3 v
Free end: x=L , = =0
d x2 d x3

d2 v
=0 at L=¿ C1 cosh λL+C 2 sinh λL+C 1 cos λL+C 2 sin λL=0
d x2

C 1 ( cosh λL+ cos λL ) +C2 ( sinh λL+sin λL )=0


C 2=−C 1 ( cosh λL+cos λL
sinh λL+ sin λL )

d3 v
=0 at L=¿ C1 sinh λL+ C2 cosh λL−C1 sin λL+ C2 cos λL=0
d x3

C 1 ( sinh λL−sin λL ) +C 2 ( cosh λL+cos λL )=0

( cosh λL+cos λL )2
C 1 ( sinh λL−sin λL ) −C1 =0
(sinh λL+sin λL)

sinh 2 λL−sin 2 λL−cosh2 λL−cos 2 λL−2 cosh λL cos λL=0

−2−2cosh λL cosλL=0

cosh λLcos λL=−1

2b) Impact and measure at free end.


Point measurement so there is an anti resonance

π 4 4 π
d) I = ( d o −d i ) = ( 0.14 −0.084 ) =2.9 ×10−6 m 4
64 64

π 2 2 π
A= ( d −d )= ( 0.01−0.0064 )=2.83 ×10−3 m2
4 o i 4

3.52 207 ×10 9 × 2.9× 10−6


ω=
2.5 2
√ −3
2.83 ×10 ×7850
=92.6 rad s−1 =14.73 Hz

2e) Likely to not be rigidly clamped at the ground so effective length may be longer than 3m.
Also get shear deformation of cylindrical __on so again overestimate frequency. Frequencies
look fairly consistent though 4 however has higher damping – probably loose in ground.
Columns 1, 2 and 3 are surprisingly good prediction

3a) 6 DoF

I) Vertical z due to piston


II) If cylinder not in centre than all
III) y is due to crank pin effect
IV) Rotation around x axis due to crank pin effect

b) Resonance at 11.2 + 4 = 11.6 Hz = 72.88 rads -1

kN
k actual=m ω2 =200 ω2n=1062
m

1062 kN
k single mount = =265.6
4 m

c) At resonance, read off graph, reacceptance x/F = 9.3 x 10 -6

1 kN
k eff = =107.53
9.6 ×10 −6
m

K actual 1062
Q= =9.87
K eff 107.53

Ratio ζ=0.05=5 %

d) 0−Peak :25 m s−2

3000
ω pow= × 2 π =314.16 rad s−1
60

ẍ=ω 2 x
25
x= =0.253 mm ¿ 0− peak
ω2pow

kx 1
=
F 2

( 1−r ) + ( 2ζr )2
2

1 1
¿ =
√309+ 0.185 17.58
F=17.58 kx=4723 N (max)

mr ω2pow =4723

With r =0.2 m

4723
m= =0.239 kg
0.2 ×314.162

1+ ( 2ζr )2
e) T =
√ 2
( 1−r 2 ) + ( 2 ζr )2
=0.062

Force = 0.062 x 4732 = 294N

Reduce mounts k by about 50% machine stability reduced to rocking

4ai) Visual, ultrasonic, magnetic particle, X-ray, eddy current, dye penetrant

ii) Coil just above component surface: pass ac current through coil; eddy current induction on
surface; flow of current impeded by cracks and so magnetic fields above surface detected.
Detected by change in coil impedance or via field sensor. Better for surface breaking cracks as
skin defects means ac current confined to surface. Yeah I don’t really know what he just said.
Just look up Inductive displacement gauges in the blue book.

4bi) 2048 points at 50 kHz, so T = 0.04096 sec

1/T = 24.41 Hz so you cannot resolve the two resonances


ii) Zero padding increases the record length by eight times. So 1/T is divided by eight to get
3.05 Hz

Effectively interpolates result of part I, but there’s no new information.

iii) Now we have real 3.05 Hz resolution from 50kHz divided by 16384 points, so resolve the
resonance. About eight points between them

5a) Script sets up transfer function and plots step response for PID controller that has Ki, Kd
both set to zero. So P is the only part of the PID controller in effect. Response plotted from t=0
to t=20 seconds.

1
The plant transfer function is 2
J s + cs

KP
θout = (θ¿ −θout )
J s2 +cs

J s 2+ cs+ K p θ out =K P θ ¿

θout Kp
= 2
θ¿ J s +cs+ K P

From canonical form of characteristic transfer function of a second order system (Not in 3M
notes,you need to check it in your 2M notes): ( - Don’t worry! see Databook Pg 27 C.6.2

θout K
=
θ¿ 1 2 2ζ
s + s +1
ω 2n ωn

KP 30
ω n=
√ √
J
=
30
=1rad s−1=0.16 Hz

c 2
ζ= = =0.33
2 √ K P J 2 √ 30 ×30

Frequency of oscillation:
ω D =ω n √ 1−ζ 2=0.151

1
Period: =6.62 seconds
0.151

Note zero slope in the beginning

6a) I θ̈ ¿ +c θ˙¿ + K P θ ¿=K P θout =K P b t

Particular integral: θ¿ = pt+ q

θ̇¿ = p

θ̈¿ =0

Substituting back into the first equation

0+ cp+ K P ( pt+ q ) =K P b t

So p=b

cb
q=
Kp

−cb
S teady state error= is proportional ¿ c , b∧inversely proportional ¿ K P
Kp

b) Integral controller removes steady state error but can increase settling time. Would settle to
zero steady state error eventually. Initial oscillation frequency is similar to part a, so unlikely to
have ____ Kp

c) May be possible, but settling time c_______ further and danger of instability
d) Kp increases. Steady state error substantially reduced. Oscillation frequency increases.
[Some weird hieroglyphics] ~ 3 secs in part c and 6 seconds in part a, so Kp increase by factor
of about 4-5

FROM MIKE RISTIC FOR Q6

a) From the ramp response the system may be assumed to be at least similar to a dc motor
driving a load, requiring a finite positional error to produce an error-proportional voltage
applied to the motor in order to drive it at constant speed, i.e if the positional error was zero
than the motor voltage would be zero and the motor would stop.

b) Integral control is added. Integrator will maintain a constant output once its input
(positional error) has reduced to zero and the graph indicates this. Integral controller clearly
increases settling time as it tends to make the overall system more oscillatory.

c) Increasing K_i will make the integral-of-error portion of the voltage signal driving the motor
larger, making the system approach the steady state faster (when error has reduced to zero).
However, integration also reduces the stability, making the system more oscillatory, so the
oscillations and the settling time may again be increased.

d) Fig 6c shows that the system reaches the steady state in about as many oscillations as in Fig
6b, but those oscillations are faster. Increasing K_p has the effect of reducing any steady state
errors and increasing the natural frequency, so it appears that this is what was done here.

You might also like