Lab 2: Vibrations of a Cantilever Beam

A Lab Report

Submitted By

Neel Nadpara

ME 3264 Applied Measurements Laboratory

Section 004L

Department of Mechanical Engineering

University of Connecticut

Lab Preformed On:

April 13, 2015

1|Page
Summary:
In any given material, molecules are never static, they undergo constant random motion. This
translates to every object, even stationary ones to be vibrating at a certain frequency. This in turn
is deemed to be the objects natural frequency. A given object mat have many different natural
frequencies and there are numerous factors that aid in determining an object’s natural frequency
which include material properties and geometric parameters. Moreover, when an object is
exposed to a driving frequency it starts vibrating more noticeably. When that driving frequency
is equivalent to one of the natural frequencies of the object in question a phenomenon known as
resonance occurs. In resonance, the amplitude of the vibration climbs to a very high value due to
in phase vibration waves. Understanding the location and the magnitude of resonant frequencies
of a system is critical in the design of real world parts and structures such as bridges, automotive
components, and the majority of machines that incorporate periodic or rotational motion. In
structural examples, resonance may even be destructive since it lets oscillations grow out of
control. Thus, resonant frequencies should be avoided when designing parts or structures.
However in certain special cases, resonance may be beneficial such as in automotive intakes and
exhaust systems. In this lab, the harmonic resonant frequencies of a simple cantilever beam were
determined. The experiment consisted of a cantilever beam made of 1060 carbon steel with an
accelerometer affixed to the end, a shaker rig, a controller, and computer with a LabVIEW VI
setup. As the beam is first struck with a rubber mallet, the beam will undergo a forced vibration
and the first harmonic frequency can be observed as an output on LabVIEW. It was seen that as
the beam length was increased the natural frequency decreased and the distance of the node
locations increased.

Experimental Procedure:
To start this lab, the beam was secured to the shaker apparatus. Afterwards, the beam’s
mechanical properties (material type, Young’s Modulus, and density) were noted. Using a digital
caliper, the cross-sectional area of the beam is measured by multiplying transverse length by
vertical height. The span of the beam from the end of the clamping fixture to the beam tip was
measured using a tape measure or ruler. Following this, the LabVIEW VI specific to the lab was
started so that the accelerometer data was live and displayed outputs on the screen. This data was
an output of the FFT of the accelerometer signal. A rubber tipped hammer was utilized to lightly
strike the beam, this in turn would display peak amplitudes on the FFT plot. The first three peaks
on the FFT plot were identified and highlighted using the cursor as this was the approximate
location of the first three natural frequencies. Following this a sinusoidal voltage was applied to
the shaker apparatus using a signal generator by adjusting the frequency. The frequency was
adjusted by fine-tuning the signal generator and amplitude. A low frequency was used initially in
order to prevent damage to the accelerometer at the end of the beam. Also, it was seen that some
time was required for the beam to respond to the excitation particularly when a low frequency
was used. Similarly, some time was necessary for energy to transfer from the shaker apparatus to
the beam. Thus the frequency was gradually increased until the frequency that was marked
earlier during the hammer strike was reached. However, the frequencies acquired from the
hammer strike were not extremely precise and therefore were not the exact natural frequency
found by using the shaker. Therefore, the hammer strike allows one to quickly find an

2|Page
approximate range that the natural frequency would be located in. Then the signal generator and
shaker can be used to precisely locate the natural frequency by incrementally increasing the
frequency until the highest part of the peak is achieved (the highest peak of the FFT plot is the
natural frequency). Subsequently, this process can be repeated to find the second and third
frequencies. Additionally, when the beam’s natural frequencies are found the beam has locations
where the beam is not vibrating. These locations are known as nodes and there is one node for
the 2nd natural frequency while there are two nodes for the 3rd natural frequency. These locations
can be precisely determined by running a pencil tip along the beam until a location is discovered
where the beam is not vibrating. This location is marked and the distance from the start of the
beam (the base of the shaker) is measured to determine the node locations. The full length of the
beam is also measured to preform theoretical calculations and plot a log-log graph of beam
length vs. natural frequency and node location. This entire process may be repeated for two more
beams of different lengths of beams. The length of the beam may be changed via the socket head
cap screws on the cantilevered support mounted atop the shaker rig. Finally, in order to
determine the damping ratio additional data is required. To obtain this data, a free vibration
signal was acquired by striking the beam with a rubber tipped hammer. Data was recorded by
LabVIEW which compiles all the data received from the accelerometer on the beam into an
Excel file. In order to determine the damping ratio, the data that was recorded was graphed and
two different amplitudes separated by any number of complete cycles were found. Then the
logarithmic decrement was utilized to determine the damping ratio of the beam.

Calculations:

Calculating Natural Frequency of a Cantilever Beam:

𝑬𝑰
𝝎𝒏 = 𝜷𝟐𝒏 √ 𝐸𝑞𝑢. 1
𝝆𝑨𝑳𝟐

ωn = Natural Frequency (Hz)

βn = Constant value corresponding to which natural frequency (n) being solved for
β1 L = 1.8751
β2 L = 4.6941
β3 L = 7.8547
E = Modulus of Elasticity (Pa)
I = Area Moment of Inertia (m4 )
kg
ρ = Density ( 3 )
m
A = Area (m2 )
L = Length of Beam (m)

Calculating Area Moment of Inertia:

𝒃𝒉𝟑
𝑰= 𝐸𝑞𝑢. 2
𝟏𝟐

I = Area Moment of Inertia (m4 )

b = Width (m)

3|Page
h = Thickness (m)

Calculating Frequency:
𝝎
𝒇= 𝐸𝑞𝑢. 3
𝟐𝝅
f = Frequency (Hz)

Calculating Logarithmic Decrement:

𝟏 𝑿𝟎
𝜹= 𝐥𝐧 ( ) 𝐸𝑞𝑢. 4
𝒏 𝑿𝒏
δ = Logarithmic Decrement
n = Number of complete cycles
X0 = Amplitude 1
Xn = Amplitude 1

Calculating Damping Ratio:

𝜹
𝜻= 𝐸𝑞𝑢. 5
𝟐𝝅

ζ = Damping Ratio
δ = Logarithmic Decrement

Figure 1: Image of the experimental set-up (please note that the beam material was 1060
steel, not aluminum).

4|Page
Results and Discussion:

Material Properties and Beam Parameters

Parameters Value
Thickness, h (m) 0.00316
Width, b (m) 0.01904
Cross Section, I (m4) 5.00665E-11
Modulus of Elasticity, E (Pa) 2.1E+11
Density, p (kg/m3) 7850
̅
𝐦 0.472
Length 1 (m) 0.59055
Length 2 (m) 0.5095875
Length 3 (m) 0.31115
β1 1.875104
β2 4.694091
β3 7.854757
Table 2: Material Properties and Beam Parameters
.

Natural Frequency Measurements and Calculations

Beam Length Mass Natural Measured Theoretical Percent
of Beam (kg) Frequency Frequency Frequency (Hz) Error
(m) (rad/sec) (Hz)
Beam 1 0.5906 0.2789 47.57 6.6 7.5 12.82
0.5906 0.2789 298.10 46.5 47.4 1.99
0.5906 0.2789 834.69 123.4 132.8 7.11
Beam 2 0.5096 0.2407 63.88 8.7 10.1 14.43
0.5096 0.2407 400.35 56.5 63.7 11.33
0.5096 0.2407 1120.98 161.1 178.4 9.70
Beam 3 0.3112 0.1470 171.35 22.3 27.2 18.23
0.3112 0.1470 1073.83 152.1 170.9 11.00
0.3112 0.1470 3006.75 427.8 478.5 10.60
Table 2: Natural Frequency measurements, theoretical calculations, and percent error
calculations.

5|Page
 Node Measurements and Calculations
Length Natural
Measured Node Theoretical Node Percent
Beam of Beam Mode Freq.
Location (m) Location (m) Error
(m) (Hz)
2 ω1 0.400 0.462 13.5
Beam 1 0.5906 ω1 0.260 0.295 11.8
3
ω2 0.448 0.513 12.7
2 ω1 0.462 0.399 15.9
Beam 2 0.5096 ω1 0.298 0.255 17.1
3
ω2 0.589 0.442 33.2
2 ω1 0.251 0.244 3.0
Beam 3 0.3112 ω1 0.079 0.156 49.0
3
ω2 0.260 0.270 3.6
Table 3: Node measurements, theoretical calculations, and percent error calculations.

Experimental and Theoretical Natural Frequency vs. Beam

Length
500
Log Natural Frequency (Hz)

50

5
0.28 0.33 0.38 0.43 0.48 0.53 0.58
Log Length of Beam (m)

Theor n1 Theor n2 Theor n3

Measured n1 Measured n2 Measured n3

Graph 1: Experimental and Theoretical Natural Frequency vs. Beam Length

6|Page
 Second Mode Experimental and Theoretical Nodes vs. Beam
Length
0.5

0.45
Node Location (m)

0.4

0.35

0.3

0.25

0.2
0.31115 0.5095875 0.59055
Beam Length (m)

Node 1 Node 1 Theoretical

Graph 2: Second Mode Experimental and Theoretical Nodes vs. Beam Length

Third Mode Experimental and Theoretical Nodes vs. Beam

Length
0.7

0.6

0.5
Node Location (m)

0.4

0.3

0.2

0.1

0
0.31115 0.5095875 0.59055
Beam Length (m)

Node 1 Node 1 Theo Node 2 Node 2 Theo

Graph 3: Third Mode Experimental and Theoretical Nodes vs. Beam Length

7|Page
 Vibration of Cantilever Beam 3 (L = 3.112)
0.5

0.4

0.3

0.2
Accelerometer (m)

0.1

-0.1

-0.2

-0.3

-0.4

-0.5
2000 2500 3000 3500 4000 4500
Time (milli-sec)

Graph 4: Vibration of Cantilever Beam 3 (where length is .3111 m)

Dampening Ratio Calculation:

First, Equation 4 was employed to calculate the logarithmic decrement:

1 𝑋0
𝛿= ln ( )
𝑛 𝑋𝑛

𝑋0 = .3686

𝑋𝑛 = .3453

𝑛 = 149

1 0.3686
𝛿= ln ( )
149 0.3453
𝛿 = 4.395 × 10−4

Afterwards, the logarithmic decrement was utilized to calculate the damping ratio using Equation
5:

𝛿 4.395 × 10−4 Because the value of ζ is less than 1, the

𝜁= = = 𝟔. 𝟗𝟗𝟓 × 𝟏𝟎−𝟓
2𝜋 2𝜋 system is underdamped.

8|Page
Discussion:

From the tabulated data, it is clear that the predicted values of the natural frequencies are quite
close to the actual experimental values of the setup. The measured and theoretical natural
frequencies fluctuate from each other by 2 to 18%. Looking at Graph 1, it is evident that the
measured natural frequency is continually less than the theoretical natural frequency for each of
the three trials. This discrepancy may exist for several reasons. The mathematical model
analyzes only a perfectly cantilevered beam, without any measurement equipment or additional
wires like in the experiment. The accelerometer and the wires add additional mass to the beam
𝑘
and also affect factors such as the overall moment of inertia. Therefore, because 𝜔𝑛 = √ and k
𝑚
remains constant (a material property of the beam), the increase in mass will decrease the natural
frequency (easily observable at the higher resonant frequencies in the experimental values).
Another reason behind the differences is the lack of precision in the controller for the shaker.
The increments were only to the nearest whole Hz; had the resolution of the scale been more
precise, such as to the nearest 0.1, the values would be more accurate. Finally, the differences
among the results can be attributed to the measured frequency of the beam being a damped
frequency as opposed to a true natural frequency- that is, 𝜔𝑑 = 𝜔𝑛 √1 − ζ2. It is important to
𝑐
recall that the damping ratio can be determined from the equation ζ = 2√𝑘𝑚 . The fixture,
measurement equipment, and table likely added some damping to the system, as opposed to
letting the beam simply vibrate freely on its own. The table and end play in the shaker fixture
likely contributed coulomb friction, contributing to “c,” and the accelerometer, wires, and
mounting adhesives contributed to additional mass, effectively increasing the damping ratio
overall. Regardless, Graph 1 also showed that as the length of the beam was increased, the
natural frequency decreased. Moreover, Graph 2 and Graph 3 displayed that as the length of the
beam increased the distance of the node locations were also enlarged. Additionally, the locations
of the nodes appear to follow a linear increasing trend as the length of the beam increases.

Data and Error Analysis:

Like any experiment, the data was acquired by taking measurements. However, there is always a
certain uncertainty associated with every measurement taken. This uncertainty translates to error
and the error propagates as equations are used to do various further calculations. For this
experiment measurement error can be attributed to the h (thickness) measurement and L (length)
measurement.

ℎ = ± 2.57 × 10−5 𝑚
𝐿 = ± .0015876 𝑚

𝑩𝟐𝒏 𝒉 𝑬
𝝎𝒏 = √ 𝑬𝑞𝑢. 6
𝟐𝝅𝑳𝟐 𝝆

9|Page
 𝛿ℎ 2𝛿𝐿
𝛿𝜔𝑛 = 𝜔𝑛 − 𝜔𝑛 𝐸𝑞𝑢. 7
ℎ 𝐿

𝜹𝒉 𝟐 𝟐𝜹𝑳 𝟐
√
𝑹𝑴𝑺 = 𝝎𝒏 ( ) − ( ) 𝐸𝑞𝑢. 8
𝒉 𝑳

Error Propagation
Natural
Length of Measured Height Length RMS
Beam Frequency
Beam (m) Frequency (Hz) Partial Partial Error
(rad/s)
47.57 6.6 0.008038 0.005376 0.0394
Beam
0.5906 298.10 46.5 0.008038 0.005376 0.2779
1
834.69 123.4 0.008038 0.005376 0.7374
63.88 8.7 0.008038 0.006231 0.0442
Beam
0.5095875 400.35 56.5 0.008038 0.006231 0.2869
2
1120.98 161.1 0.008038 0.006231 0.8180
171.35 22.3 0.008038 0.010205 0.2897
Beam
0.31115 1073.83 152.1 0.008038 0.010205 1.9758
3
3006.75 206 0.008038 0.010205 2.6760

There are numerous errors that are not due to uncertainty for this experiment. First, there were
several things that could have damped the system that were not accounted for. Examples include:
the tape and mass of the accelerometer on the beam, the table shanking or anyone leaning or
putting weight on the table, and the shaker not being affixed to the table by fasteners. Moreover,
some material property error exists as well as the exact Elastic Modulus is not known and it can
be between 190-250 GPa. Also, vibration over long periods of time can cause strain hardening in
a material which can increase the Elastic modulus. Moreover, it is very possible that the beam
that was used could have been plastically deformed at one point leading to some error. This
experiment could easily be improved by the use of 12 inch calipers instead of a tape measure as
it is capable of providing much more accurate data with much less uncertainty. Overall, the
majority of the percent errors and RMS values were quite small leading one to make the
conclusion that the data was reliable.

10 | P a g e
Conclusion:

Overall the experiment represented forced mechanical vibrations of a cantilevered beam very
well. The experiment demonstrated that the natural frequency of a beam decreases as its length
increases. The experimental natural frequencies of the beam at varying lengths tended to remain
very close to the theoretical values, as seen on the included data tables. The largest reason for the
discrepancies was a result of damping being present in the experimental values- damping came
from the mass of the accelerometer, the friction in the shaker mount, and air friction to name a
few. Another potential reason for any frequency differences is the modulus of the 1060 steel
beam, which may not be at exactly 210 GPa as assumed- it may very likely be higher or lower.
In a real-world design the modulus would either be assumed to be on the high or low side if
certain resonant frequencies were to be avoided. Additionally, it was seen that as the beam
increased in length the node locations also got larger. Overall, the purpose of the experiment was
fulfilled.

11 | P a g e