Dynamic Analysis of a Cantilever Beam with an Offset Mass by Yurui Zhan Department of Mechanical Engineering and Materials Science Duke University Date: Approved: Brian P. Mann, Supervisor Samuel C. Stanton Earl Dowell Thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Mechanical Engineering and Materials Science in the Graduate School of Duke University 2019
49
Embed
Dynamic Analysis of a Cantilever Beam with an Offset Mass
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Dynamic Analysis of a Cantilever Beam with anOffset Mass
by
Yurui Zhan
Department of Mechanical Engineering and Materials ScienceDuke University
Date:Approved:
Brian P. Mann, Supervisor
Samuel C. Stanton
Earl Dowell
Thesis submitted in partial fulfillment of the requirements for the degree ofMaster of Science in the Department of Mechanical Engineering and Materials
Sciencein the Graduate School of Duke University
2019
Abstract
Dynamic Analysis of a Cantilever Beam with an Offset Mass
by
Yurui Zhan
Department of Mechanical Engineering and Materials ScienceDuke University
Date:Approved:
Brian P. Mann, Supervisor
Samuel C. Stanton
Earl Dowell
An abstract of a thesis submitted in partial fulfillment of the requirements forthe degree of Master of Science in the Department of Mechanical Engineering and
Materials Sciencein the Graduate School of Duke University
The aluminum base consists of a fixed base and a moving part connected by
screws. The fixed base is mounted on the table using screws so that it cannot have
any movement. After plugging the beam to the predetermined location, the moving
19
Figure 3.1: Experimental setup of a cantilever beam with an offset mass
plate is used to clamp the beam to make sure the actual boundary conditions as
close as possible to theoretical assumptions.
The system is made of ABSplus-P430 by 3D printing, which is a production-
grade thermoplastic that is durable enough to perform virtually the same as produc-
tion parts (Stratasys (2018)). When combined with 3D Printers, ABSplus is ideal
for building 3D models and prototypes. Properties of this light-weight and highly
machinable material can be obtained from data sheet.
The offset mass is realized by attaching different bars to the end of main beam
as shown in Figure 3.2. Each bar has a rib to add rigidity and make its own natural
frequency extremely high compared with the cantilever beam, and a slot is used for
attaching to the beam on the other side. To reduce stress concentration and add
flexibility when plugging the beam, fillets are added to the slot as shown in Figure
3.3.
There are two sets of attachment for this experimental setup. The first set in-
cludes five bars with same length but different widths. These five widths given in
terms of beam widths are respectively 0.25b, 0.5b, 0.75b, b and 1.25b, which pro-
20
Figure 3.2: Attachments to the main cantilever beam
Figure 3.3: Fillets and rib on attachments
vides a set of attachments with different mass but same offset. To illustrate these
attachments more clearly, they are named after No.1 - No.5 attachment with the
increase of width. Parameters for the attachment are shown in Table 3.2. The other
set includes one beam with five slots evenly distributed on it. The gap between slots
is 34mm, and mass for this attachment is 5.5g. When the cantilever beam is plugged
into different slots, the attachments are considered as same mass but different offsets.
To illustrate the experiments more clearly, these five slots are named in Figure 3.4.
21
Table 3.2: Attachment parameters(Unit of mass: gram/ Unit of length: mm)
No. Mass Base length Base Width Base Thickness Rib Length Rib Width
1 0.9 50.09 4.99 2.12 2 2.6
2 1.7 50.02 9.94 2.05 2.92 3.67
3 3.1 49.74 15.25 2 5.01 5.51
4 4.4 49.94 19.94 2.13 5.92 6.72
5 5.6 49.96 24.96 2.09 6.96 7.8
Figure 3.4: Schematic of slots for different offset ratio
The natural frequency was determined by the unforced free oscillations of the
cantilever beam by giving an initial displacement to the system. As a result, an
accelerator is mounted to the vibrating beam to measure the position. It is then
connected to a NI data acquisition (DAQ) device, and the position-time plot will
be shown on a pre-designed MATLAB interface. The data saved can be used in
analyzing other system characteristics. Mass of the accelerator used is 0.7g, which
is 7.1% of the beam mass. It is then considered as a distributed mass to make the
theoretical model closer to actual situation.
3.2 Experimental Data and Analysis
The data obtained from experiments was processed in MATLAB to calculate natural
frequency and damping ratio. Fast Fourier Transform(FFT) rapidly computes the
22
transformation for a signal from time domain to frequency domain. Code using FFT
is generated to analyze the experimental data, and one of the results for the system
is shown in Figure 3.5.
Figure 3.5: Time series and FFT results
The decrease in amplitude from one cycle to the next depends on the extent
of damping in the system. Because the successive peak amplitudes bear a certain
specific relation ship involving the damping of the system, the method of logarithmic
decrement is formed to evaluate the damping ratio of a underdamped system:
δ “ lnx1x2“
2πξa
1´ ξ2, (3.1)
where x1, x2 is the first two displacement. So the damping ratio is
ξ “δ
a
p2πq2 ` δ2. (3.2)
Time response for each experiment gives two displacement as shown in Figure
3.5. Damping ratio can be derived from the method of logarithmic decrement.
3.3 Uncertainty Analysis
The measurements of the variables have uncertainties associated with them, and the
values of the material properties obtained from reference resources also have uncer-
23
tainties (Coleman and Steele (1999)). General uncertainty analysis is an approach to
consider only the uncertainties in each variables, neglecting random errors. If result
r is a function of J variables Xi:
r “ rpX1, X2, ..., XJq,
the general uncertainty is defined as
U2r “ p
Br
BX1
q2U2
X1` p
Br
BX2
q2U2
X2` ...` p
Br
BXJ
q2U2
XJ,
where the UXiare the absolute uncertainties in the variables Xi. So the definition of
relative uncertainties is
ˆ
Urr
˙2
“
ˆ
X1
r
Br
BX1
˙2ˆUX1
X1
˙2
`
ˆ
X2
r
Br
BX2
˙2ˆUX2
X2
˙2
`...`
ˆ
XJ
r
Br
BXJ
˙2ˆUXJ
XJ
˙2
,
(3.3)
where the uncertainty magnification factors (UMFs) are defined as
UMFi “Xi
r
Br
BXi
. (3.4)
In the cantilever beam system, the derivation equation of natural frequency is
ωi “ pβiLq2
c
EI
mL4
“ pβiLq2
d
E 112ct3
ρctL4
“ 0.2887pβiL2qt
L2
d
E
ρ,
(3.5)
where c, t are width and thickness of the beam. The general uncertainty expression
24
becomes
ˆ
Uωω
˙2
“
ˆ
t
ω
Bω
Bt
˙2ˆUtt
˙2
`
ˆ
L
ω
Bω
BL
˙2ˆULL
˙2
`
ˆ
E
ω
Bω
BE
˙2ˆUEE
˙2
`
ˆ
ρ
ω
Bω
Bρ
˙2ˆUρρ
˙2
.
(3.6)
The UMFs are
UMFt “t
ω
Bω
Bt“
t
ω0.2887pβiLq
2 1
L2
d
E
ρ(3.7a)
UMFL “L
ω
Bω
BL“L
ω0.2887pβiLq
2t
d
E
ρp´2q
1
L3(3.7b)
UMFE “E
ω
Bω
BE“E
ω0.2887pβiLq
2 t
L2
c
1
ρ
1
2
1?E
(3.7c)
UMFρ “ρ
ω
Bω
Bρ“ρ
ω0.2887pβiLq
2 t
L2
?E
ˆ
´1
2
˙
ρ´32 . (3.7d)
Substituting Equation 3.7 into Equation 3.6,
pUωωq2“
¨
˝
t
0.2887pβiL2q tL2
b
Eρ
0.2887pβiLq2 1
L2
d
E
ρ
˛
‚
2
pUttq2
` 4
¨
˝
L
0.2887pβiL2q tL2
b
Eρ
0.2887pβiLq2t
d
E
ρ
1
L3
˛
‚
2
pULLq2
`1
4
¨
˝
E
0.2887pβiL2q tL2
b
Eρ
0.2887pβiLq2 t
L2
c
1
ρ
1?E
˛
‚
2
pUEEq2
`1
4
¨
˝
ρ
0.2887pβiL2q tL2
b
Eρ
0.2887pβiLq2 t
L2
?E
ˆ
´1
2
˙
ρ´32
˛
‚
2
pUρρq2
“pUttq2` 4p
ULLq2`
1
4pUEEq2`
1
4pUρρq2.
(3.8)
Relative uncertainties for natural frequency can be obtained from Equation 3.8 using
the relative uncertainties of thickness, length, Young’s modulus and density. From
25
Equation 3.8, the system is most sensitive to beam’s length, while Young’s modulus
and density have same and smallest effect on the results.
26
4
Analysis and Discussion
4.1 Experimental Results
Experiments were carried out using the above-mentioned setup. Experiments of
the beam with each attachments were carried out five times repeatedly to get rid
of random error. When conducting the FFT, the sampling frequency is chosen to
be 1000Hz, and as shown in results, all the FFT results for the one experimental
setup are consistent. After the calculation, as shown in Figure 4.1, the results of
each experimental natural frequency are read directly from the plot and recorded.
Calculating the average of all the results leads to the final experimental natural
frequencies. For beam with same offset but different mass, the offset ratio is
ζ “23
196.92“ 0.117,
and result of FFT for the third trial with attachment 1 is shown in Table 4.1.
For beam with different offset but same mass, results are shown in Table 4.2 after
applying the same approach to experimental data. Mass ratio here is
µ “5.5
9.9“ 0.556.
27
Figure 4.1: FFT result of the beam with No.1 attachment
Table 4.1: Experimental Results of natural frequencies with same offset ratio
Set Mass Ratio µ Frequency(Hz)Beam 0 12.2Beam with No.1 Attachment 0.091 10Beam with No.2 Attachment 0.172 8.714Beam with No.3 Attachment 0.313 7.286Beam with No.4 Attachment 0.444 6.5Beam with No.5 Attachment 0.566 5.9
Table 4.2: Experimental Results of natural frequencies with same mass ratio
Set Offset Ratio ζ Frequency(Hz)Beam with No.1 Slot -0.361 5.5Beam with No.2 Slot -0.254 5.7Beam with No.3 Slot 0 6.1Beam with No.4 Slot 0.254 5.8Beam with No.5 Slot 0.361 5.4
28
4.2 Theoretical Result
Following the steps discussed in Chapter 2.2 , MATLAB code is generated to derive
the natural frequency theoretically and the results are shown in Table 4.3 and Table
4.4. Because all these theoretical results come from dimensionless equations, they
don’t have any unit here.
Table 4.3: Theoretical Results of natural frequencies with same offset ratio
Set Natural FrequencyBeam 3.516Beam with No.1 Attachment 3.0007Beam with No.2 Attachment 2.69Beam with No.3 Attachment 2.3129Beam with No.4 Attachment 2.07Beam with No.5 Attachment 1.91
Table 4.4: Theoretical Results of natural frequencies with same mass ratio
Set Natural FrequencyBeam with No.1 Slot 1.76Beam with No.2 Slot 1.8994Beam with No.3 Slot 1.9446Beam with No.4 Slot 1.8994Beam with No.5 Slot 1.76
From Table 4.3, natural frequency decreases when offset is constant and mass
increases. Similarly, it also shows a downtrend when offset increases. It is reasonable
from a physical point of view because the beam will obviously have lower natural
frequency when the mass becomes larger.
4.3 Damping Ratio
Damping ratio is considered as an influencing factor for the system. Damping ratio
of each system is calculated using the method of logarithmic decrement as discussed
29
in Chapter 3.2 for each time response, and then taking the average of results for
five trials leads to the final damping ratios as shown in Table 4.5 and Table 4.6.
Table 4.5: Damping ratio of natural frequencies with same offset ratio
Set x1 x2 lnpx1{x2q Damping ratio ξBeam 0.0727 0.0646 0.1158 0.0184Beam with No.1 Attachment 0.0748 0.0674 0.1046 0.0166Beam with No.2 Attachment 0.0709 0.0669 0.0581 0.0093Beam with No.3 Attachment 0.0729 0.0672 0.0809 0.0129Beam with No.4 Attachment 0.0745 0.0676 0.0965 0.0154Beam with No.5 Attachment 0.0758 0.0685 0.0995 0.0158
Table 4.6: Damping ratio of natural frequencies with same mass ratio
Set x1 x2 lnpx1{x2q Damping ratio ξBeam 0.0727 0.0646 0.1158 0.0184Beam with No.1 Slot 0.0658 0.0600 0.0913 0.0145Beam with No.2 Slot 0.0608 0.0561 0.0822 0.0131Beam with No.3 Slot 0.0595 0.0548 0.0824 0.0131Beam with No.4 Slot 0.0501 0.0463 0.0787 0.0125Beam with No.5 Slot 0.0488 0.0435 0.1143 0.0182
Undamped natural frequency, which is more similar to the condition of theoretical
assumption, can be obtained using
ωn “ωd
a
1´ ξ2, (4.1)
where ωn is undamped natural frequency, and ωd is the damped natural frequency
obtained from the experiments.
Taking damping ratio into consideration gives to the undamped natural frequency
as shown in Table 4.7 and 4.8. From the tables, the undamped frequencies have very
little difference with damped frequencies, so we can conclude that damping ratio in
this system is small in this system and has little effect on the system.
30
Table 4.7: Comparison of natural frequencies with same offset ratio including damp-ing ratio
µ ωed(Experimental) ωed ωenrad/s
Beam 0 76.655 3.555 3.556
Beam with No.1 Attachment 0.091 62.832 2.913 2.914
Beam with No.2 Attachment 0.172 54.752 2.539 2.539
Beam with No.3 Attachment 0.313 45.779 2.123 2.123
Beam with No.4 Attachment 0.444 40.841 1.894 1.894
Beam with No.5 Attachment 0.566 37.071 1.719 1.719
Table 4.8: Comparison of natural frequencies with same mass ratio including damp-ing ratio
ξ ωed(Experimental) ωed ωenrad/s
Beam with No.1 Slot -0.361 35.673 1.608 1.608
Beam with No.2 Slot -0.277 36.186 1.678 1.678
Beam with No.3 Slot 0 37.442 1.736 1.737
Beam with No.4 Slot 0.277 36.814 1.707 1.707
Beam with No.5 Slot 0.361 34.301 1.590 1.591
4.4 Moment of Inertia
When calculating the moment of inertia for the attachments, the slot will be ne-
glected, so that they are considered as standard plates with ribs. Non-dimensionalizing
the moment of inertia gives to Equation 4.2.
I “IMML2
“
112mp4h2 ` w2q ` 1
12m2p4h
22 ` w
22q `m2d
2
ML2,
(4.2)
where m, h and w are the mass, length and width of the plate, m2, h2 and w2 are
the parameters of the rib, and d shows the offset between the plate and rib.
Including moment of inertia of offset bar gives to the results as shown in Table
4.9 and Table 4.10. The tables indicate that moment of inertia, similar to damping
31
ratio, doesn’t have significant influence on results.
Table 4.9: Natural Frequency for same offset ratio with I
Set µ ωtn ωtn with inertia
Beam 0 3.516 3.516
Beam with No.1 Attachment 0.091 3.001 2.979
Beam with No.2 Attachment 0.172 2.69 2.661
Beam with No.3 Attachment 0.313 2.313 2.283
Beam with No.4 Attachment 0.444 2.07 2.045
Beam with No.5 Attachment 0.566 1.91 1.882
Table 4.10: Natural Frequency for same mass ratio with I
Set ξ ωtn ωtn with inertia
Beam with No.1 Slot -0.361 1.760 1.743
Beam with No.2 Slot -0.277 1.890 1.810
Beam with No.3 Slot 0 1.945 1.914
Beam with No.4 Slot 0.277 1.890 1.810
Beam with No.5 Slot 0.361 1.760 1.743
4.5 Poisson’s Ratio
Poisson’s ratio is the ratio of transverse contraction strain to longitudinal extension
strain in the direction of stretching force. It has an effect on the Young’s modulus
according to the analysis of Arafat (Arafat (1999)). As shown in his dissertation,
Lame constants are introduced as
µ “E
2p1` νq
λ “Eν
1´ ν ´ 2ν2,
where ν is Poisson’s ratio for material. For a simple beam,
Q1111 “ λ` 2µ
“ Epν
1´ ν ´ 2ν2`
1
p1` νqq.
32
Poisson’s ratio for ABS plastic is about 0.35, which makes Young’s modulus E bigger.
Poisson’s ratio will lead to an increase in the Young’s modulus, which is not
consistent with the trend of experimental results. Therefore, it will not be taken into
consideration as an influencing factor for the deviation.
4.6 Result Discussion
The mass of beam becomes 9.9` 0.7 “ 10.6g after including the mass of accelerator.
As a result, the density has a 7% increase to 1.25 g/cm3. Based on other parameters
in the experiment, the scaling constants Tc is
Tc “
c
ρAL4
EI
“
c
1.25ˆ 103 ˆ 0.000043164ˆ 0.196924
2.2ˆ 109 ˆ 1.70944ˆ 10´11
“ 0.0463 1{s
Applying the scaling constants for time and damping ratio calculated to experimental
results leads to the final comparison of experimental and theoretical results. For
experiments of beam with different masses where ζ “ 0.117, results are shown in
Table 4.11 and Table 4.12.
ωed(Experimental), ωed, ωen and ωtn represents damped natural frequencies from
natural frequencies and theoretical natural frequencies.
Results for each natural frequency are shown in Figure 4.2 and Figure 4.3.
4.7 Uncertainty
Uncertainty can be derived from Equation 3.8 as discussed in Chapter 3.3. Taking
results from same offset ratio as an example, the uncertainty comes from many
33
Table 4.11: Comparison of natural frequencies with same offset ratio
µ ωen ωtnBeam 0 3.556 3.516
Beam with No.1 Attachment 0.091 2.914 2.979
Beam with No.2 Attachment 0.172 2.539 2.661
Beam with No.3 Attachment 0.313 2.123 2.283
Beam with No.4 Attachment 0.444 1.894 2.045
Beam with No.5 Attachment 0.566 1.719 1.882
Table 4.12: Comparison of natural frequencies with same mass ratio
ξ ωen ωtnBeam with No.1 Slot -0.361 1.608 1.743
Beam with No.2 Slot -0.277 1.678 1.810
Beam with No.3 Slot 0 1.736 1.914
Beam with No.4 Slot 0.277 1.707 1.810
Beam with No.5 Slot 0.361 1.591 1.743
Figure 4.2: Plot of natural frequency with same offset ratio
34
Figure 4.3: Plot of natural frequency with same mass ratio
aspects, including the material properties and geometric features of the beam. From
the equation of scaling time constants, uncertainties of other parameters can be
minimized using methods like repeated measurement except Young’s modulus. Thus
all weight of uncertainty is put on Young’s modulus. The theoretical property is
calculated by tuning the Young’s modulus to match the last experimental result
with the theoretical one, which has the largest deviation of 9.47%. Matching the last
point in the experiment to the theoretical data
ωe5 “ ωt5 “ 1.8823, (4.3)
so the nondimensionalized damped natural frequency is
ωe5 “ 1.8823ˆa
1´ ζ2
“ 1.8823ˆ?
1´ 0.01582
“ 1.8821,
(4.4)
35
then the corresponding time scaling constant is
Tc “ωepnondimensionalq
ωe
“1.8821
37.071
“ 0.05076,
(4.5)
so the tuned Young’s modulus is
E “ρAL4
T 2c I
“1.25ˆ 103 ˆ 4.3ˆ 10´5ˆ 0.19692
0.050762 ˆ 1.709ˆ 10´11
“ 1.84ˆ 109Pa.
(4.6)
From Equation 3.8, the uncertainty for the Young’s modulus is
UE “ Een ´ Etn
“ 2.2ˆ 109´ 1.84ˆ 109
“ 3.64ˆ 108
UEE“
3.64ˆ 108
2.2ˆ 109
“ 16.5%,
(4.7)
so the uncertainty for natural frequency is
Uω1
ω1
“
c
1
4pUEEq2
“
c
1
4ˆ 0.1652
“ 8.25%.
(4.8)
The plot of natural frequency with tuned Young’s modulus is shown in Figure. 4.4.
36
Figure 4.4: Plot of natural frequency with same offset ratio after tuning Young’smodulus
37
5
Conclusion
This thesis investigates the natural frequency of an asymmetric system consisting of
a cantilever beam and different offset masses.
The beam theory of a cantilever beam is first generated using Hamilton’s princi-
ple. After deriving the kinetic and potential energy, equation of motion is written for
the system. Nondimensionalization is applied to simplify the analysis. Essential and
natural boundary conditions are defined to obtain the harmonic motion expression.
The system including the cantilever beam with different offset masses is then ana-
lyzed using the same method. Analysis including moment of inertia is also performed
to get closer to the experimental environment. An experiment is then designed to
verify the theoretical calculation. After repeating with different offset masses, a set
of data is obtained and used to derive other parameters in the system.
The theoretical analysis shows the decreasing trend of first-mode natural fre-
quency when the attachment mass or the offset of attachment increases. From the
plots, the experimental results have a reasonable trend showing the decrease but have
big deviation with the theoretical ones. We take several influencing factors into con-
sideration: damping ratio in the system, moment of inertia of attachments, Poisson’s
38
ratio of the material and beam properties. The method of logarithmic decrement is
applied to derive damping ratio, and it is proved to have little effect on the result.
Moment of inertia for the attachments is also calculated but doesn’t have significant
difference in this system. Poisson’s ratio will make Young’s modulus higher, which
results in opposite tendency of the natural frequency. We conclude that the biggest
influencing factor lays on the beam properties.
ABS plastic is the material chosen in this experiment instead of commonly used
metal setup. All the beam and attachments are printed out by 3D printers, which
increases unknown uncertainty to the process. The material properties of ABS plastic
comes from the data sheet provided by the manufacturer.The beam and attachments
are printed out using a 3D printer whose smallest resolution is 0.5mm. Printing
orientation is calculated by the corresponding software for the printer and defined
by the chosen position on the printing platform. Filling level is also a factor for
the printed beam. 3D printing provides a new way to build models for structural
analysis. It has much flexibility and can be used to build complex system, but it also
has many limitations learned from the analysis. The precision of printer determines
the quality of the printed model, which is important for a system to be consistent
in property for all parts. Making sure that the printed structure is isotropic is also
essential to perform all beam theory analysis, but 3D printing increases uncertainty
from this aspect. As a result, using 3D printing to complete dynamic analysis is
convenient but needs further consideration.
39
Bibliography
Abramovich, H. and Hamburger, O. (1991), “Vibration of a Cantilever TimoshenkoBeam with a Tip Mass,” Journal of Sound and Vibration, 148, 162–170.
Arafat, H. N. (1999), “Nonlinear Response of Cantilever Beams,” Ph.D. thesis, Vir-ginia Polytechnic Institute and State University.
Coleman, H. W. and Steele, W. G. (1999), Experimentation and Uncertainty Analysisfor Engineers, Wiley.
D., Z. L. and H., N. A. (1989), “The Non-linear Response of a Slender Beam carryinga Lumped Mass to a Principal Parametric Excitation: Theory and Experiment,”Int. J. Non-Linear Mech, 24, 105–125.
Moeenfard, H. and Awtar, S. (2014), “Modeling Geometric Nonlinearities in the FreeVibration of a Planar Beam Flexure With a Tip Mass,” Journal of MechanicalDesign, 136.
Pilkee Kim, S. B. and Seok, J. (2012), “Resonant Behaviors of a Nonlinear CantileverBeam with Tip Mass subject to an Axial Force and Electrostatic Excitation,”International Journal of Mechanical Sciences, 64, 232–257.
Shahram Shahlaei-Far, A. N. and Balthazar, J. M. (2016), “Nonlinear Vibrations ofCantilever Timoshenko Beams: A Homotopy Analysis,” Latin American Journalof Solids and Structures, 13, 1866–1877.
Stratasys (2018), “ABSplus-P430 Material Parameters,” .
TO, C. W. S. (1982), “Vibration of a cantilever beam with a base excitation and tipmass,” Journal of Sound and Vibration, 83, 445–460.
Xiao Shifu, D. q. and Bin, C. (2002), “Modal Test and Analysis of Cantilever Beamwith Tip Mass,” ACTA Mechanica Sinica (English Series), 18, 407–413.