£453,440 $9534.. p. m.
.- ... 4.. 4. ...... :I..-..: o 0‘ As.” U’anvac .-

.. . I , ,_ 4 . 4 _ . .. _ .4... _... ...44.4.44H..;...__.444.A.. ....».......:.3./.44.Jo .33»... bursa.”

 
 

            

                           

 

. _ . . . . ......4....... 4
. . 4 .44 . . 4. 4. . 4. .. .... 3“.“ . ‘4. .. t. I to ”1.0.7...
_ 4 .c . .4 .4 ...¢ . .... .. . 4v..P-..~w.~vw...o’
. .. . .4 ... 4 ... .. 4.... ‘ ......v 04. .4.&
. . r 4 .... . v ; a . a . o. . a... . I... ¢ 0 o o. 4:01:
r. 4 4 p. . . . O... .. O44 . .I
. 4 . .. .44 .4 4 ...... .4. ....¢.. . ... ... .4. .f....; ...;:.....v&
4 . . . .. . 4 . 4 . . ..o. ....
. 4 4. . .4. . .4 4 ’. . c o.».’.. .4. 3 4.... .4. .p 4.
. 4 .1 . ... ...l. ...a._ . .4134-
. o. 4 a . 1. 4......._.....4 4.....c4....a..n... 17.39.:-
. 4.. 4 4.. . . . _ . . . .. ..r..
v 4.4 . . co. . q a. 4.1.... o:. . .ovur: .oa.‘
. 4 a 4 . 4 :23: .....v?. 4
4 4 . 4a.4t.
.
4 4...
.
1, . _ . . . ’....o..
a . ..
w —
.
fi 0 . . O;
. .
w . 4 r.
. . '»
4 . . . .
. .. v. .
. 4 .4 . . _ . . 4? ... ......44f
o . 4. .ll

 

           

 

 

 

 

. 4 _ .. . M4444“.64.5.43“453.444.... .»
u . 4 . .31... ... z. 3.. haiku”... 4 . .c
u. m .4 . 4.14.4.4“ .
4. E , .
. M . _
_ 4m. m .
4 m m .
H M m 4 M mu ..H 4 ... .4 4. .. .4. 444...... . m
. 0 E 0 LI mu... . .. a 4.4.474. .M
I e N 4 -. . .. 4.... ..4
M N e nu Mun 4 . 4 .549...._.._4.4.4_..,4.. .W
.. R .nlU Wu 5 . . H . ,4 H
. B T e E E. 7 . . ... . . 4 .. 4 HR...” . 4
. F. R. In S E _. .. 4.” hum. “
o m .1 N c . .. .
. w... M ...». m .... _ ....
. m... 0 . . ._ w .. ..M .....m. .. _

v . .’ o . .. . 4 .. o... . «...-Q. s...,.. 4 4.... 4... . ...P. . ..L.»...$.o.€.o .4......._.‘o . J...
. o . . . . o . .
o .

 

...”...‘....4.4......J._“ ..4w.‘5..3.104;. ...4 m4.4rt4;\03...4.»4¢.2 4.3.34 if... 141%. (.15...

o .4 . o ... o 4.. b.f...o.. .4 .- 0:. {13AM gfg .qSMfianfi. 4

4. 4 . . . _.. . . . . s .u . .. .v . . 4 . . .... ... . ... u .. .4. . ...... 4 ... x ......44 .4 4. cu 44 . r A .. . . 0.- 4 . . . . .c.. 04.14 . .. . O).fu...bn4¢...l \o..l.8.._4 fi...o‘.oLR.“NCo§.h1..wa.flfi.‘%lbfihnhanu~h ... ...v

 

    
   
 

.4v .
l

mM/IWIY/flflffli/i/f/I]Willis/WWW”WWI

3 1293 10537 9949

 

 

 

3,1? ABSTRACT

A CRITICAL ANALYSIS OF VIBRATION
MEASUREMENTS OF THE
TRANSPORTATION ENVIRONMENT

BY

Janece Rae Hausch

A review of vibration and its occurrence in the
transportation environment is presented. Details of the
analysis and presentation methods used are examined. Examples
of real damage situations are given. Implications of vibra-
tion's interaction with the package system and the need for
exact information on input levels and frequencies is stressed.
A comprehensive comparison of three recently published re—
ports measuring the environment was made. The two studies
conducted by the School of Packaging using commercial vehicles
under actual operating conditions were found to have accelera-
tion levels which fall within figures previously reported.
This finding suggests that present interpretation of the
actual environment is overly severe. The predominant fre-
quencies experienced were higher than expected. Areas re-
quiring further investigation for a better understanding in
package design and current controversies concerning vibration

inputs are defined.

A CRITICAL ANALYSIS OF VIBRATION
MEASUREMENTS OF THE

TRANSPORTATION ENVIRONMENT

BY

Janece Rae Hausch

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

School of Packaging

1975

TABLE OF CONTENTS

Page

LIST OF TABLES . . . . . . . . . . . . . . . . . . . iii
LIST OF FIGURES. . . . . .V . . . . . . . . . . . . . iv
I. INTRODUCTION. . . . . . . . . . . . . . . . . 1
II. VIBRATION . . . . . . . . . . . . . . . . . . 4
III. VIBRATION MEASUREMENT . . . . . . . . . . . . 11
IV. METHODS OF ANALYSIS . . . . . . . . . . . . . 16
V. DATA PRESENTATION . . . . . . . . . . . . . . 27
VI. DISCUSSION. . . . . . . . . . . . . . . . . . 35
VII. REPORT COMPARISON . . . . . . . . . . . . . . 42
VIII. CONCLUSIONS . . . . . . . . . . . . .'. . . . 61

REFERENCES . . . . . . . . . . . . . . . . . . . . . 68

ii

L IST OF TABLES

Table Page
1. MSU--Reported Values . . . . . . . . . . . . . 51
2. List of Variables and Instrumentation . . . . 52
3. Foley Reported Values. . . . . . . . . . . . . 55
4. Miscellaneous Reports. . . . . . . . . . . . . 57

5. Report Comparisons . . . . . . . . . . . . . . 59

iii

LIST OF FIGURES

Figure

l. Sine wave . . . . . . . . .
2. Combined waveform . . . . .
3. Complex waveform . . . . .
4. Transmissibility curves . .
5. Types of filters. . . . . .
6. Filter transmission characteristics
7. Low-pass filter cutoff frequency.
8. Sinusoid spectra. . . . . .
9. Complex waveforms spectra .

10. Random wave spectrum. . . .

11. Two filter transmissions .

12. Two filter resolutions. . .

l3. PSD analysis . . . . . .

14. Strip chart presentation. .

15. Envelop presentation. . . .

16. Bar graph presentation . .

l7. Discrete figures presentation

18. VIBRAN presentation . . . .

19. PSD presentation. . . . . .

iv

Page

18
18
19
21
21
22
23
24
26
29
3O
31
32
33

34

I. INTRODUCTION

Every product is subjected to shock and vibration
while proceeding through the distribution system. In gen-
eral shock is usually a result of drops incurred during
handling operations. Vibration is the result of input to
the package through material handling equipment and trans-
portation vehicles. The former is a controllable factor
which can be measured for the individual manufacturing
and warehouse situations. Transportation related vibration
is of a more varied and uncontrollable nature relating to
specific vehicles and Operating conditions. Vibration
occurs as the result of the interaction of the vehicle
structure and the road surface.

I The area of shock input has been extensively re-
searched in order to relate product damage to material
handling factors. These studies include the number of
expected drOps, their height, the weight of the package,
and the relation of the cushioning material to the actual
input to the product. SoPhisticated engineers have utilized
this information in the design of packages for protection.
Unfortunately not all product damage is related to drOp
shocks but has been found to relate to the more complicated
input of vibration to the product and its package.

1

It is important that the engineer understand the
implication of the vibration response of the packaged pro-
duct to the transportation vibration. This less intensive
input damages the product through force factors due to
amplification and fatigue. Abrasion can result in container
loss due to an illegible address. An unsightly package
loses customer appeal. Vibration may cause a breakdown of
the package components so that it is unable to perform its
protective function. This failure may have an effect on the
external surroundings with possible damage due to leakage,
loose product impacts, or broken container hazards.

A considerable amount of information has been pub-
lished describing the vibration environment. Various studies
were conducted to determine the extent of vibration in the
field. These experiments represent a considerable under-
taking from the standpoint of time and money. The modes of
travel covered were truck, rail, air and shipping._ This in-
formation is necessary for a scientific approach to the de-
sign of packages which would be able to withstand the dis-
tribution environment. With the proper data the design of
a package can be evaluated to provide additional protection
or eliminate costly over-packaging. An economic trade-off
may be made between the package cost and cost of losses due
to damage. I

The previous work done in the transportation vibra—

tion measurement field has been full of inconsistencies and

is inconclusive regarding the package environment. Many

of the projects were of a limited nature such as the anal-
ysis of a new type of aircraft or suspension system. Al-
though used as distribution data most measurements were not
meant to examine the effect on cargo but on the structural
integrity of the vehicle.

Government agencies have developed the most infor-
mation on vibration but they have been mainly concerned
with military and space vehicles. Some studies concerned
directly with packaged product distribution have been con-
ducted by the related transportation industries and pack-
aging groups. Composite studies which combine the results
of different vehicles and procedural methods compound prob-
lems and present extreme values. These studies have been
sporadically conducted since the 1950's and many have been
limited in sc0pe according to the researcher's objectives
and costs. Published literature reveals a wide range of
values for expected frequencies and expected acceleration
levels.

The vibration data presented is not static and must
not be construed to be all inclusive of the transportation
environment. There has been a constant change in vehicles
and methods of Operation. Many new modes of travel have
opened up including cargo jets, high speed trains, piggy
back, and container ships. These all present new possibilities
of input to be measured and may be found to be considerably

different than "normal" vibration inputs.

Beside the particular vehicles selected the methods
used in the studies should be questioned. New measurement
instrumentation and new faster methods of analysis have been
develOped, refined and utilized. Examination of the alter-
natives used and data collected will allow future researchers
greater refinement since they will know what to expect and
can anticipate problems.

To be useful to the engineer some sense should be
made of the large amount of information available. Vibra-
tion information gathered should be of a general nature and
presented in a practical form for use in engineering calcula-
tions. Conditions encountered should be representative of
common transport environment applicable to packaging. Any
composite data form should be compiled selectively so as to
avoid overstatement and misinterpretation of the facts.

It is the intent of this paper to examine present
published data representing values established for commercial
truck and rail transportation. To begin with a general re-
view of vibration will be given along with how this phenomena
is measured. Following will be a review of the ways that the
recorded information may be analyzed. Then the various for-
mats for presenting the information and inherent problems
incurred will be examined. Finally a detailed comparison
is made of three of the most recent reports with appropriate

conclusions and further considerations.

II . VIBRATION

Vibration is defined as an oscillating motion. The
movement varies independently over time and magnitude. This
motion is usually represented as a wave on a graph. The
magnitude factor is designated the wave amplitude and is
the maximum vertical distance through which the wave varies
from the origin. The period of the wave is the amount of
time required for the motion to begin repeating itself.

The frequency of the wave is the number of wave cycles
which occur in a second and is the reciprocal of the period.

A pure sine wave has both a constant amplitued and constant

frequency.

+A-

 

 

FIGURE 1.—-Sine wave.

5

Complex waves can be formed by the combining of
individual sine waves. The various input segments are
added to determine the maximum amplitude and the frequency

of occurrence. An example is shown in Figure 2.

 

FIGURE 2.-—Combined wave form.

Other common wave forms are the square wave and sawtooth wave.
These occur as a result of the discrete prOperties of an ob-
ject and its driving mechanism. In practice these may be

simulated by a combination of sinusoids.

 

 

 

 

 

 

 

 

 

 

 

 

FIGURE 3.--Complex waveform.

Wave motion can be described as being stationary or
nonstationary. The latter movements are called transients
which are sudden changes in amplitude from the normal state
such as occurs in a shock input. Since these occur only
occasionally the important aspect is the time element or
duration. Common causes of transient motion in the trans-
portation environment are rail switching, rough roadbeds,
and hard accelerations. The important factor is whether

the package has time to react to the extreme input.

Stationary motion is described as being either
periodic or random. A periodic motion repeats itself in a
continuous pattern. In this case the important factors are
the dominant frequency or frequencies of the input and their
magnitude. Random motion is one which has no set pattern of
either magnitude or frequencies. Since the motion is not
exactly definable, the variable factors must be stated in
terms of averages or maximums. It has been found that the
transportation environment is basically a random situation
with discrete inputs from the drive mechanisms and railbed
plus variable inputs from interactions of the entire system.

The importance of wave motion is the manner in which
a mass reacts to the input energy. Each mass has a natural

frequency at which it reacts.

fn=-2lfl- @735

The variable K is the spring constant or stiffness factor
of the object or package. Many products have various com-
ponents each of which has its own natural frequency due to
its construction or mounting characteristics.

Energy input to an object from a wave of the same
frequency as the object's natural frequency results in ex-
treme amplification of the input force. This condition is
called resonance. Of course an object does not respond at
the theoretical limit at resonance due to the factor of
damping. Damping is the dissipation of amplitude through

the transfer of mechanical energy at different rates or

into thermal energy. The amount of energy transferred to
an object is seen in transmissibility curves for various

fractions of critical damping.

 

 

 

3? I
‘L = 0 c = fraction of
critical
; = 2 damping
2D
Tr
c=5
jfl‘llm.~§
1 ...... -511---
I .
I
I
I
I
0 J L $ A
1. /2 2 3
f/fn

FIGURE 4.--Transmissibility curves.

As can be seen the force experienced by the object
can be many times the input force. A ratio of forcing fre-
quency to the natural frequency in the range of .5 to 1.5
causes serious magnification levels. Other factors affect-
ing the response level at the resonance point are humidity
and temperature. In these cases the characteristics of the
packaging materials may be significantly altered. Corrugated
absorbs moisture thereby losing strength, while cushioning
material changes stiffness at different temperatures.

All parts of a product are affected by a dynamic

input to the system. A transient input results in a

10

vibration which occurs at the natural frequency. However,
it quickly dies away due to internal damping and lack of
continuous input. A periodic input results in the product
vibrating at the particular discrete frequency. A random
excitation results in random response dependent on the natural
frequency of the product and the frequency range of the in-
put vibration. Should the natural frequency be within the
range and the duration of the input long enough, resonance
will develop. Damage occurs as a direct result of vibration
when a resonance condition concentrates high forces on
fragile parts of the product. This force can be beyond the
stress limit of the material and result in breakage. Also
fatigue may occur which reduces the life of the element.
Package resonance can quickly obscure labels and graphic
design through abrasive contact within the cargo hold.

In addition large compressive forces may be developed on
package surfaces due to movement and the loading of product
resulting in crushing of the package. Lateral force can be
especially critical if the product is designed to only with-

stand vertical inputs.

III. VIBRATION MEASUREMENT

All engineering studies should be conducted in accord-
ance with accepted experimental standards. Exact documenta-
tion of the project should be recorded to ensure the valid-
ity of the subsequent data. The investigation should be
planned to proceed in a methodical manner. The information
should be gathered carefully so that the data can be repro-
duced in future studies or refinements of a similar nature.

The aim in this case is to measure a vibration
phenomena within stated objectives and parameters of the
transportation environment. The number of modes available
allow for many variations which should not be compounded to
give a large overstatement of the subject. In a study the
variables of the environment which are to be investigated
should be clearly stated. These include type of vehicle
and its structural characteristics, loading weight of
vehicle, speed, and the run characteristics including sur-
face type and Operations. It is important to know all these
parameters along with the basic methodology of the vibration
measuring and recording system.

Since the two variables of vibration are magnitude

and time, a device or devices will be needed to correctly

11

12

measure and record them and their interaction. Earlier
studies measured magnitude with a simple strain gage and
plotted on a strip chart which moved at a specified rate.
This combination, though not very responsive, gave an indi-
cation of the maximum amplitude and the duration of time.

Modern technology has greatly improved the accuracy
of these measurements. Recent studies have been conducted
using accelerometers which have high natural frequencies
and are capable of very sensitive frequency measurements
across the transportation range. These accelerometers are
either piezoelectric or piezoresistive. Their output is
electrical in direct prOportion to the amount of internal
distortion caused by the force input from the vibration.

The recording of the transducer's output has pro-
gressed both in quantity and quality with the use of magnetic
tape. Tape recording provides an exact record of the vibra-
tion input response over long periods of time and allows
this record to be thoroughly analyzed by different techniques
at a later date.

In any engineering study the objective of the in-
vestigation and cost considerations affect the parameters,
instrumentation, and depth of analysis. Based on the amount
and quality of data desired a decision must be made on the
basic components of the instrumentation system. The engineer
must be aware of the possible problems and limitations in-

volved and insure proper use and experimental design.

13

Accelerometers' frequencies vary from a very few
cps to as high as 30 KHz. A high natural frequency will
allow for a linear response over the anticipated vibration
environment. A more critical point is the sensitivity
rating of the accelerometer. DC response is necessary to
be able to measure and distinguish very low levels of vibra-
tion. Care must be taken not to damage the accelerometer
by an overload from a high input road shock or during in-
stallation. Secure installation is important because the
mounting determines the response characteristics. Since
vibration occurs in three planes, the accelerometers should
be oriented for directional input normal to each plane.
They should be able to withstand significant accelerations
transverse to the sensitive axis, thereby giving a correct
reading without any dissipation or additions.

There are many types of tape recorders available.
The main features to consider are the number of channels
available, sensitivity, range, and ease of use. -The recorder
must have DC response and a signla to noise ratio high enough
to allow accurate resolution of the transmitted signals. This
means correct coupling and cable type so as not to distort or
lose data content. An accurate reading for analysis may be
enhanced by use of low-pass filter within the input line to
eliminate possible high frequency noise.

An additional piece of equipment, the amplifier,

may be necessary to raise the recorded signal to a suitable

14

level for analysis. It is important that this procedure
does not distort the signal characteristics or introduce
new problems such as noise. Again filters may be used to
suppress unwanted signals which may interfer with the
readings.

The correct output of a system requires that all
the components be compatible so that the signal processed
is not altered in any manner. It also depends on the proper
isolation of the recording system, prOper connections, and
correct placement of the measuring devices so that data is
not limited or eliminated. Usually this means having the
recorder and required power sources in a separate compart-
ment, either in the cab or an adjacent car. Placement of
the individual accelerometers should be in the area most
likely to receive rigorous input. Previous studies have
shown that in some frequency ranges lateral forces are
actually larger than the vertical forces. Accelerations in
the front of a trailer may differ significantly from those
recorded in the back. The question of the most severe
location still remains Open.

Another aspect of measuring the transportation en-
vironment is the actual moment of when to do so. There are
many variable inputs which occur normally in transportation
and many considered of a more uncommon nature. Selection
of the relevant occurrences for study must be made. Since

transportation vibration input is random there must be

15

enough data to be able to be analyzed statistically. There
are certain limitations as to the amount which can be
continuously recorded. Objective judgement of the data
must be made to determine the typical examples. Above all
there must be consistency in selections for comparison. Be-
fore taping the requirements of the analyzing machines must
be considered. Some instruments require long data segments
but only use a small percentage of the actual data due to
sampling techniques. Others can give almost instantaneous

results utilizing all elements of a very short data interval.

IV. METHODS OF ANALYSIS

Transformation of the recorded data into hard copy
requires instrumentation to read the signal, separate fre-
quencies, and reduce the amount of the information to a
useable size. The broader the scope of the study the more
difficult it is to analyze because of the total amount of
data available and the number of variables included. This
will yield a broad range of measurements both continuous
and transient. In contrast less conclusive and more exact
measurements will be obtained from a project of limited
scope. The choice of analyzing equipment is dependent on
the Objective of the study and must be considered during
the initial phases of the project planning.

The two independent variables of time and amplitude
must be given a measured value. A normal sine wave exhibits
a variation in amplitude of an equal amount in both the
positive and negative directions. This results in an
average amplitude of zero. The most severe acceleration is
designated the peak amplitude. This is the maximum height
of the sine wave and occurs instantaneously twice per cycle.
The average value of the sine wave is the average of a half-

cycle or .636 of the peak. A more realistic value and the

16

17

standard measure for electronic signals is the square root
of the mean squared or RMS. The mean square value of a wave
is the average of the sum of the squared individual magni-
tude differences from the mean value. In the case of a

sine wave the RMS value is .707 of the peak value. Other
wave shapes have different ratios of the RMS value to the
peak value. If the data measured is Gaussian the RMS value
and the standard deviation, 0, are identical. In general,
the RMS gives the most statistically accurate account of

the wave phenomena.

The time factor is easily converted into frequency
but the problem is accurate presentation. The analyzing
apparatus selected to examine the recorded data must be
sensitive enough to pick up all frequencies and determine
the differences in their magnitude. Filters play a critical
part in this analysis process. A filter is an electronic
instrument or network which allows certain frequencies to
pass through to the next stage of analysis while blocking
out all other unwanted frequencies. A filter may be high-
pass or low-pass which allows frequencies over or under a
specified point to be read or it may be a bandpass which
will only pass those freuqencies within the bandwidth of
the interval. The first two types are usually used in
eliminating unwanted frequences for a clearer signal before
analysis. The bandpass is used extensively in the actual

wave analyzing process.

18

 

 

 

 

 

 

 

 

 

Low—pass . High-pass ' Bandpass
K K h- K p IF‘
Sto Pass Sto
Pass StOp Stop Pass p p
fc fc fcl fc2

FIGURE 5.--Types of filters.

An ideal filter is one which would allow an exact
number of cps to pass through at their full strength. The
transmissiOn characteristic would be unity. _However, a real
filter only passes a portion of the frequencies. The cut-
off frequency of the filter is that frequency where the

voltage gain drOps to .707 of the passband value or -3db.

   
  
  

[M1512
1 ..
Ideal filter
e—vReal filter
.5_, *Half-power
bandwith

 

  

 

 

 

f1 f3 fc f4 f2

FIGURE 6.--Filter transmission characteristics.

l9

 

.707 K _

 

fc

FIGURE 7.--Low-pass filter cutoff frequency.

The transmission characteristic is critical for small
bandwidth filters. The value of a le increment is estimated
by averaging the reading over the actual bandwidth of the
filter used. This method will compound errors in reading
very low frequency ranges. The size of a filter affects
the time needed for the correct reading to be attained.

This settling rate is another characteristic which has a
direct effect on analysis time. The time required is usually
equal to the reciprocal of the filter bandwidth.

Analysis may be performed using a number of machines
which differ in accuracy, time requirements and type of
value measured. The most elementary is the strip chart which
translates the recorded or received signal into a graph of

amplitude over time. The detected signal is proportionally

20

displaced from the center line according to the magnitude
of acceleration. The resolution of the time element is
controlled by the speed of the chart. No frequency is
indicated. 1

This visual representation is rather superficial
but it gives an indication of relative values. It may help
pinpoint areas of interest to be examined in greater detail.
The limitation of the machine is the inherent slow mechanical
response. This leads to individual amplitudes being super-
imposed resulting in misstated acceleration levels. Also
this visual data record results in subjective judgements of
what is a typical occurrence.

A more accurate representation of a wave is given
by spectrum analysis. This is a determination of the power
of a wave as a function of its component parts. A filter is
used to scan the frequency range and the specific frequencies
present are located. The wave is presented on an oscillosc0pe
as the maximum amplitude input for each frequency. For
instance a sine wave is presented as a single energy level
at a single frequency. A composite wave Of two different
sine waves would be presented by two different lines. More
complicated wave forms such as a square or sawtooth are
presented as harmonics which are multiples of the principle

or fundamental frequency.

21

/\/\.2

f3

 

 

 

FIGURE 8.--Sinusoid spectra.

 

 

 

 

 

 

 

A
Square wave
.. I I L
f 3f 5f 7f 9f

A
Composite wave

 

 

 

 

f1 f2

FIGURE 9.--Complex waveforms spectra.

22

For a random wave, which by definition is composed
of various frequencies of varying amplitudes all changing
over time, a wave analyzer is required. This instrument
yields a visual output on an oscilloScope of the accelera-
tion levels over all possible input frequencies. The wave
analyzer determines the wave content by either sending the
signal through a bank of filters of increasing bandwidth or
by sweeping a single filter across the entire range. Records
need to be quite long due to the settling time of the fil-
ters and the time required for sweeping. Although long
data segments are necessary only 1% of the data might be
used due to the sampling rate. An alternative method which
may increase the chances of signal error is the use of a
data lOOp. This would allow the same selected representative
segment to be fully analyzed by repetition. ,A random wave

would contain some measurement of all frequencies.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FIGURE lO.--Random wave spectrum.

Filters of course play a critical part in this in-

strumentation. Further information is required for com-

plete understanding of their implications. The size or band-

width of a filter determines the resolution of adjacent

frequencies in the wave. For instance, two separate fre-

quencies input to an analyzer may yield two entirely dif-

ferent wave shapes and amplitudes. The smaller the filter's

bandwidth the more accuracy obtained. The superimposed

waves will give an overstatement of the magnitude of the

vibration condition. As seen in Figure 11, the amplitude

measurement (ll-d) is dependent on the filter's bandwidth

(lla and llb) and the spacing of the inputs (ll-c).

Filter B1

1’ II

Frequency

(a)

Transmission
Characteristic

 

Sinusoidal
Input

1'1 II

Frequency

(C)

 

 

 

 

 

 

Filter 32
. Transmission
l ' -Characteristic
Frequency —I
(b)
I Measured
.Amplitude []
.‘ Bl
./ .2
Frequency
(d)

FIGURE ll.--Two filter transmissions.

Another aspect of wave resolution is the time re-

quirement of the filter. A small single bandwidth filter

used over the entire range will give fine resolution but re-

quires an extremely long analysis time. Alternately the

24

filters used may be of increasing bandwidth size which will
smooth the measured input at the higher frequency intervals.
Larger filters allow things to proceed at a faster rate.

In addition large frequency ranges are generally presented
on a logarithmic scale and fine resolution in the upper

range is usually lost.

Constant bandwidth filter Filter proportional to
interval analyzed

 

Acceleration
Acceleration

 

 

 

 

Frequency Frequency

FIGURE 12.--Two filter resolutions.

The most sophisticated instrument available today

for analyzing random vibration is the hybrid digital-analog
realtime wave analyzer. This machine Operates on the same
principle as the wave analyzer. However, due to the computer
memOry it can read, average and present the information using
100% of the recorded data in a fraction of the previously re-
quired time. This is a result of a time compression memory
which allows successive data samples to be immediately and

continuously processed within the microseconds of the

25

memory's operational time. The impact of using a 1500

word memory with a circulating period of .0001 sec. at a
sampling rate of 30 Hz is that 50 s of data are in 100 us of
storage or a compression ratio of 500,000 to l. A larger
filter say Of 10 KHz only requires 50 ms to sweep the entire
spectrum and analyze it properly.

Since the vibration environment is a random phenomena
it should be treated statistically. This requires stating
values as a probability of occurrence. The realtime wave
analyzer allows the retention of data so that the amplitude
frequency occurrences over time or density can be computed.
The value is most commonly known under the generic electrical
phrase of power spectral density or PSD. Actually it repre-
sents the mean square acceleratiOn density and the units are
G2/Hz. This measurement gives an indication of the extent
of the dominant frequencies in a random input.

The output on the oscilloscope will be of a similar
shape as that from the wave analyzer except for the units.

A true random signal as exemplified by white noise measure-
ments will have the same density across the entire range.
Discrete input superimposed on a random signal or a narrow
band random vibration input will result in peak occurrences.
The character of the vibration signal is important to know

for design purposes.

26

 

 

 

 

 

 

N
N m
m \\
O?‘ Nb
(.9 >a
a u
u -a
a True Random Wave 2
c m
(D 'O
U c
a I "\ 010
(D O H w-I
H-H M4J
0-0 5:6
5‘0 0*“
0" H U) a)
m m IIA
I-4 c m
c m m U
8 8 £8
2 6
Frequency Frequency

FIGURE l3.--PSD analysis.

Measurements taken from these analyzing machines
can be utilized and presented in many ways. These will be

examined in the next section.

V . DATA PRE SENTAT ION

The presentation of the vibration measurements is
determined by the inherent characteristics of vibration,
the content determined by the method of data collection, and
the intended use of the data. These three areas are inter-
related and should be considered in the planning stages.
The test design for the vibration study should be carefully
planned so as not to include too much or exclude what is
necessary. This will allow a quick and effective analysis.
The important point of presenting any data is that it be
correct and useful.

Since vibration is recorded as a measured displace-
ment level over time, these two elements and their inter-
relationship are of primary interest. The frequency content
of the vibration is important due to the natural frequency
of any Object. The acceleration levels are important in
relation to the fragility level of the packaged product.
The time history of the input to a vehicle is important for
endurance stress considerations. The relative weight of
each occurrence should be considered for its importance as
an input to a specific system.

The objective of the study will set the parameters
which will be included. These may include a study of the

27

28

type of vehicle, effect of loading, speed, and road type.
Initial objectives also set the data requirements. The
test design will determine the level and extent of the
measurements to be taken. The measuring method may limit
the data recorded to peak amplitude or to preselected am-
plitude ranges. Analysis may be done for separate acceler-
ation increments or averages. Intervals analyzed may be
of varied length. This has a bearing on the accuracy,
resolution, range, and completeness of the individual values
presented.

The method of data presentation is dependent on
the information which is required either for engineering
calculations or test purposes. Definitive values for a
specific variable studied may be presented with comparisons
or a more general composite may be made. Engineering use
may require more than one graph or table to apply to in-
dividual circumstances. Specific areas of interest include
the maximum or average amplitude, specific frequencies,
effect of test variables, and duration of input. Aside from
the actual numbers or values of the elements their inter-
action and occurrence should be seen. This will result in
some form of probability or statistical treatment. The
common methods of presenting vibration data are discussed.

The most elementary presentation is the strip chart
recording. It depicts the magnitude versus time history of

the recorded vibration. No frequency can be read. This

29

method allows the variation of possible levels to be seen.

With this readout a typical segment may be isolated for more
detailed study. The accuracy of the measurements is limited
by the mechanical response of the moving needle and the chart

speed.

 

 

 

 

 

 

 

 

 

_LIJILILLILAAIAAILJALII

 

FIGURE l4.--Strip chart presentation.

One of the most common methods to show frequency is
the envelOp graph of acceleration levels versus frequency.
This graph maps the maximum acceleration in G's that is
attained within a measured frequency band against the center
frequency of the band. The levels may be either peak or
RMS. The individual points are then connected to envelop
all frequencies across the range indicating their expected

levels for that test situation.

30

Acceleration

 

 

Frequency

FIGURE 15.—-Envelop presentation.

Individual graphs of different parameters such as
measurement directions or speed may be superimposed to com-
pare magnitudes and dominating frequencies._ A possible
problem is that an overstatement of levels may be made due
to the size of the filter analysis or the inclusion of an
occasional transient. Connecting the fC of large analysis
intervals may obscure dips in the range. Also indiscriminate
combination of values from different conducted tests may
result in errors due to differences in instrumentation,
vehicle, road, and technique. The use of the envelope
usually gives a very conservative picture of the environment.

A similar type of presentation is given by the bar
graph. Here the maximum value attained in the measured fre-

quency interval is presented for the entire interval.

 

31

Different measurement directions are often presented by
different shading techniques. Again the size of the in-
tervals are determined by the instrumentation and the

amplitude resolution determined by accelerometer sensitivity

and filtering.

WI

 

 

Acceleration

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Frequency Intervals

FIGURE l6.--Bar graph presentation.

A more exacting presentation is a numerical table
which breaks out the percent of samples registered for each
discrete amplitude level within each band interval. This
is accomplished with the use of a computer which reads and
sorts the signal. The visual table shows the relative
occurrences in each acceleration level. This is more precise
presentation but it does not allow for quick comparison of
similar levels of different parameters or of similar percen-
tages across the range. Transient occurrences are seen as

a discrete jump a step above the previous level.

32

Distribution in Percent

h
h.
but!" us“.
hhmwhhk
”lb-WhhlbWw-N-M“

II

u:.u. hnl~r'h~.hv'hb'UI-hh'h~
an.k-IhvlwrnunvNu.n~'n~suus..
Mwhh’mbh’huu
up Au-tu.~u-~u~ uhlhvlhubyu..~
hhhhha-W-uph-Mn—

Acceleration Increments
HHHH’
HIIH‘I

 

 

RMS Frequency Intervals
L8V€l~m~wmhmwflmww

FIGURE l7.--Discrete figures presentation.

The VIBRAN diagram developed by Sandia eliminates
some of the above problems. Probability levels are used
in a plot of the expected magnitude against the center fre-
quency of the measurement intervals. This graphic presenta-
tion of relative confidence levels allows quick analysis
of the expected amplitudes. This method shows the large
difference between the peak level and the levels registered
for the major portion of the data. Comparison of different
parameters may be made by overlaying their comparable prob-

ability levels.

33

v, \I \ Peak

99.5%
99.0%
98.0%
95.0%
90.0%

Acceleration
(ggggéé

less
less
less
less
less

than
than
than
than
than

 

 

Frequency

FIGURE 18.--VIBRAN presentation.

A plot of the PSD curves for a random wave input

is obtained directly from the oscilloscope of a realtime
wave analyzer. This graph will show the relative intensity

of dominant frequencies across the analyzing range. The
values may be read as either peak or average to be expected.
The reading of mean square acceleration density for each

frequency or Gz/Hz makes it difficult to convert to accel-

eration levels for comparison with other methods. An ex-

pected RMS level across the entire range may be computed

from the area under the curve. The PSD curves may also be

superimposed for comparison of parameters. This measurement

is unsuitable for rapidly changing wave values. This type
of measurements has the advantage of being extremely quick

and not requiring a computer.

34

 

Gz/Hz

 

 

Frequency

FIGURE l9.--PSD presentation.

All the above methods have been used to publish
data on the vibration environment. However many problems
and misunderstandings arise from a casual interpretation of
the numbers. All studies and resulting values should be
examined critically for the correct statement of range,
sensitivity, technique of measurement, instrumentation
employed, vehicles and other parameters. Variation in any

one could significantly alter or obscure compared data.

VI . DISCUSSION

Although common vibration levels are known to be
less than shock input levels, many problems still arise
from the phenomena. These are due to interaction of the
input frequencies, natural frequency of the product, and
spring constant of the packaging material. A slight change
in any part of the package or products distribution system
may result in a damage situation eventually traced to vi-
bration. For this reason adequate description of the vibra-
tion field is necessary.

A recent example of such a damage situation was high-
lighted in a trade magazine. A switch to a newer freight
car caused fruit shipments to bruise. Closer examination
revealed that the problem only occurred in the top containers.
The cause of damage was attributed to the stiffer suspension
which interacted with the corrugated containers' transmissi-
bility factor. The input experienced at the tOp of the
stacked containers was five times the original input. This
problem was solved by use of a compressible foam inner pad
which prevented attainment of the resonant condition through
damping.

Another aspect of vibration damage was experienced

in rail shipments of a photographic emulsion. Two liquids

35

36

were packed in a plastic cartridge separated by a punctur-
able film divider. Vertical vibration testing according

to accepted standards did not reveal any problems. However,
initial shipments were ruined by premature mixing of fluids
due to a breakdown of the divider's seal. Further investi-
gation determined that the container was very susceptible
to lateral vibrations which caused the fluid to impact
against the plastic divider.

A third example of complications arising from vi-
bration input occurred after the switch to palletized ship-
ments. In this case product which had been block stored
three pallets high with no crushing. When shipped by truck
stacked two high the lower cartons were failing. The
factors attributed were the dynamic load imposed by the
upper pallet weight and the humidity weakening the
corrugated.

These examples reveal the complexity and prevalency
of the vibration problem. Every change in a system should
be thoroughly examined for damage implications. New modes
of travel may change input patterns. New transport designs
are constantly being introduced for both specialized and
general applications. Recent levels expected for rail have
been close to one half the levels quoted twenty years ago.
A transport design considered good for one packaged product

may adversely affect another.

37

It is possible that the costly damage problems
cited could have been completely avoided if complete atten-
tion had been given to the vibration implications. How-
ever, this area is often overlooked both from ignorance and
the lack of decent information. All presentations purport-
ing to define the vibration environment should be examined
critically. Often the data is categorically stated as the
level expected for all possible vibratiOns. However, one
must consider carefully the test method and objectives. As
stated previously an overstatement of the levels can be as
costly as an understatement from the standpoint of package
cost.

Unfortunately it is not always possible to estab-
lish the various components of a study. The chart which is
presented often lacks an accompanying statement as to the
instrumentation or methodology. A statement of the instru-
ments measurement range is needed to explain the upper and
lower cutoffs. The filter bandwidth used in analysis would
indicate resolution and the amount of smoothing encountered
by areas of rapidly changing acceleration. The type of,
measurement used for the acceleration levels should be stated.
The difference between an average and the RMS signal may
not be significant but for testing purposes or calcula-
tions it could be critical. The amount of noise in the

system and the signal to noise ratio should also be stated.

38

A statement of the composition of the data should
be made. As seen in the previous section transients may
be present as indicated by the ommission of sampling
points in an acceleration step of a frequency interval.

This would increase unnecessarily the expected levels.

The length of the recorded data may be important. Conditions
of operation have a significant effect. Reports which are

a composite of previous studies may be a composite of errors.
Different equipment and methods used in each case may shift
the graph of dominant frequencies or highest amplitudes.
Composite reports usually combine diverse values since most
smaller studies deal with noncommercial subjects and special-
ized vehicles. This is particularly true for air and ship
studies which are concerned with structural dynamics instead
of cargo.

Some graphic presentations give the impression that
vibration increases indefinitely at the extreme ends of the
ranges measured. While many studies concentrate on fre-
quencies below 500 cps some have shown an increase of accel-
eration above those levels. All curves showing an increase
in magnitude should be carried out until a leveling or down
turn occurs due to damping of the system.

Special areas of analytical interest in vibration
studies include transients, low frequencies, and the varia-
tion due to rotational inputs. In the analysis of rotating

equipment for bearing failure spectrums are develOped to

39

measure abnormal frequency input and magnitude. To elim-
inate the difference in input due to the free coast down
speed the successive spectrum readings are normalized.
This means the frequencies are shifted in an amount deter-
mined by the ratio to the first order or base speed. It
is possible that a similar shift may be observed in the
vibration levels recorded at different speeds. This means
a relative narrowband vibration input around the dominant
frequencies of vehicle input would be found.

Low frequency acceleration levels quoted for the
vibration spectrum are a problem. On many charts the low-
est frequencies cited are among the highest in magnitude
across the entire range. It is important that the levels
be closely examined for the true extent of their impact.
Some accelerometers do not adequately measure the low
frequency ranges from 0-5 Hz. In Foley's report where the
chart begins at 0 Hz he admits that his equipment did not
measure below 2.5 Hz. This was not stated in the presen-
tation.

Analysis of low frequencies requires good resolu-
tion so that the amplitudes are not superimposed. The fine
Aresolution requires a very long analysis time due to the
settling rate for the small filters used. The exact impact
of low frequency input lies in the natural frequency of a
product and the transmissibility of the package. Although

typical fragility levels have been established for different

40

product types it remains to be seen what are typical fn

of products. Bishop states that natural frequencies of
whole structures are generally less that.50 Hz and rarely
more than 500 Hz. This means that the excitation input of
the transportation environment will have considerable effect.

Vibration is supposed to be a steady state input.
Sandia reports that there is significant incidence of trans-
ients in the low frequencies. These higher levels are gen-
erally included in graphs as the normal steady state con-
dition. These transients are not any higher than the
steady state input of higher frequencies. More sensitive
instruments may fill these apparent voids. Sampling rates
may have had an effect on the percent of data examined and
the probability that these points are indeed transients.

As in shock evaluation the duration of these inputs is of
the most importance.

Presentation of data in chart form is good for en-
gineering requirements of design levels. It is important
that the values be correct and inclusive Of the anticipated
environment. Levels and weight factors are needed for the
develOpment of vibration simulation tests. Tape input of
an actual road test to a shaker would give an almost real—
istic input. This of course does not allow for interaction
with the vehicle structure.

Previous reports present a certain amount of
useful information. A survey of vibration measurements

reveals a wide array of levels and relevant ranges.

41

With a solid background of expected levels, common measure-
ment problems, and analysis techniques future reports can
be of a more conclusive nature. New data is necessary

because of the improvement in equipment and vehicles.

VII. REPORT COMPARISON

A review of literature defining the vibration en-
vironment shows that few studies have been compiled in
recent years. In general, older studies were concerned
with factors other than the environment experienced by car-
‘go as it moves through the distribution system. Vibration
analysis had been performed to measure the influence on the
integrity of a vehicle's structure. Transportation vibration
analysis has also been conducted with a variety of military
vehicles or for specialized carrier applications. Due to
the expense and time involved fora complete study few have
been attempted and the full extent of vibration's affect on
cargo has not been adequately defined.

Two of the most recently completed studies on the
vibration encountered in common carriers were conducted by
the School of Packaging. Separate reports examine the ex-
tent Of vibration levels in freight cars and over-the-road
trailers. The results of these reports are compared with the
latest publication of the vibration environment as established
in Sandia Laboratories Environmental Data Bank. These re-
ports should represent the state of the art of engineering
research on this subjeCt. Pertinent information from other
publications is also included.

42

43

This attempt at comparing vibration measurements
has proven to be very complicated. Two areas of possible
comparison are the methodology and actual values. As in
any engineering study the Objectives and parameters were
established according to intended use and analysis method
available. As a consequence the choice of equipment for
sensitivity and the level of presentation may reduce the
actual amount of information presented.

An account of the equipment, methods, and analysis
instrumentation is necessary to compare the accuracy and
extent of the findings. Parameters which should be examined
include:

1. the frequency range and sensitivity of instru-

ments

2. the use of filters in data analysis

3. the amount of actual data used in the analysis

Comparison of the actual data presented considerable
problems. Although the basic elements of frequency and
acceleration were the same the scales used were not the same.
Problems encountered were:

1. different overall frequency ranges

2. different frequency intervals analyzed

3. different acceleration increments

4. different bandwidths filters used for resolution

5. the size of the sample and the number of

averages employed

44

6. the statistical level of the presented data

For some of the presentations it is virtually im-
possible to manipulate the data to achieve a base for com-
parison. Some eliminate the very low frequencies due to
the inherent difficulties in obtaining and analyzing them.
Some do not carry out the analysis over a wide enough range
due to the belief that the levels diminish or do not occur.
The choice of the interval presented is dependent on indi-
vididual analyzing and measuring techniques. A significant
amount of overlap of intervals occurs in the different re-
ports and it is impossible to break down the interval to
individual frequency component levels. Use of standard
deviation presentations present similar problems. Since
the majority of occurrences of a frequency center around the
mean level the higher the acceleration level the less num-
ber of events that will be occurring. For instance, 1% of
the data may be responsible for 60% of the acceleration
range. The distribution of the occurrences under the peak
is not known and cannot be scaled down for a comparison.

The reports covered here will be individually sum-
marized highlighting the significant findings and conclusions
as drawn by the authors. A comparison of the reports will
be made in a table for their significant factors and levels
defined. Discrepancies in the given information and pre-

viously reported values will be noted.

45

Technical Report No. 20 published in November, 1972
was conducted to measure typical in-service rail vibration
and analyze it using modern methods. Information concern-
ing the equipment, methodology, and route factors were com-
pletely detailed. In general the vibration levels of a Hy
Cube, 70 foot 100 ton box car were recorded on magnetic
tape for three runs over the same track. The car contained
30 tons of scrap iron. The basic parameter changed was the
springing of the suspension.

A continuous voice record monitored the variable
events such as speed, crossings, and track type. Due to an
over estimation of the vibration levels expected, the sen-
sitivity of the recording system was set low. The result
was a poor signal to noise ratio with an error in dominant
peak readings of i10% and larger for smaller levels. Analysis
was conducted on segments of the data record where speed was
held constant for a minute or more. These segments were
selected from a strip chart read out of the RMS values.

This selection method was used to establish both the transient
and stationary periods of the runs chosen for analysis.

These minimum stationary segments were required by the anal-
yzer in order to allow proper settling time for the filter-
ing function.

Wave analysis was not used due to the relative
shortness of the record segments. The average RMS values

were determined by a random noise voltmeter for all runs

46

for speeds of 20, 30, and 45 mph over both welded and jointed

track. Resulting ranges for these values were:

vertical .079 to .3129
tOp lateral . .026 to .116g
bottom lateral .022 to .129g

In general the measurements showed that the lateral directions
were approximately the same while being three to ten times
lower than the vertical accelerations.

PSD analysis of the stationary segments was done by
a realtime wave analyzer to determine the frequency charac-
ter of the data. This machine is capable of fairly good
analysis of a very short signal depending on range. Due
to the nature of the instrument's resolution, smaller inter-
vals cannot be measured as closely over the short time. The
maximum value obtained at the predominant peak at 60 Hz was
.00082 GZ/Hz for a o-soo Hz range. This measurement fell
to .00038 Gz/Hz over a 0-200 Hz range. The computed RMS
value for the PSD curves ranged from .15 to .389. Other
prominent peaks of lesser intensity were at 20, 40, 80, 120
and 400 Hz.

The PSD curves for lateral directions measured the
top lateral value of .00015 Gz/Hz at 60 Hz over the 0-200 Hz
range. The bottom lateral was .0003 Gz/Hz at the same point.
Peak vibration intensity occurred at the same frequencies
irregardless of springing which implies that there are
resonances in the car and loading itself. An increase of

speed raised all the PSD curves slightly but only shifted a

47

single peak in the lateral direction. Welded rails scaled
down the expected acceleration levels.

The extremely long data records required for high
resolution of very low frequency data were unavailable.
Instead spectral analysis was performed over the 0-20 Hz
range. This revealed few predominant frequencies and the
highest acceleration levels were very low compared to the
average RMS measurements.

peak vertical .035g at 3.6 Hz
peak lateral .05g at .35 Hz

This was about the driving frequency of the rail joints.
Other peaks though of a lower magnitude occurred at:

vertical .2, .6, .9, 1.3 Hz
lateral .5, .85, 1.0 Hz

A peak acceleration analysis for the 0-20 Hz range of the
vertical data recorded the maximum value at .159 for 1 Hz.

An attempt at probability measurement was made
through an interface with a digital computer. The data was
found not to be a normal random distribution. The skewness
was to be expected because although thought to be random
from the large number of possible inputs, there are certain
driving frequencies in the system. These will interact with
load and speed variables to influence the dominant peaks
recorded.

Analysis of the transient occurrences of rail
crossings showed two different responses to the input. One

was the gradual build up and subsequent gradual decline in

48

acceleration levels across the time range. The second re-
sponse was a large displacement then a slow return to normal
levels. The maximum acceleration levels attained from

these inputs were:

Vertical 1.4g
Ilateral .42g

Generally these shocks at crossing introduced accelerations
of lg vertical and l/Bg lateral above the normal state.

The authors conclude that more work is necessary to
completely define the character of rail transportation
vibration. The largest components of vertical vibration
occurred at higher frequencies--60, 80, 120, 400 Hz--than
originally expected. There were no predominant low fre-
quencies. Presence of hydraulic snubbers was found to have
a large effect on the vibration levels not the springing.

Technical Report No. 22 was published by the School
of Packaging in September, 1973. The intent of the study
was to obtain and analyze vibration data on common carrier
trucks for use in package testing. Modern instrumentation
was employed. The general setup was an array of piezoelectric
and piezoresistive accelerometers placed in six spots in
the trailer. The tape recorder and power source were placed
in the cab and the data was monitored continuously. Three
different trucks with three different loads were used over
the same route. A fourth truck was recorded making local
deliveries. Stationary segments of data for speeds 40, 50, 55

and 60 mph were used in analysis.

49

The RMS values were measured by a voltmeter with
frequency response 2 to 20 KHz. The average range of

acceleration levels were:

rear vertical .33 to .689
mid vertical .15 to .349
tOp lateral .13 to .289
bottom lateral .13 to .309

The front vertical measurement usually fell between the mid
and rear levels. Comparison shows lateral measurements are
smaller by a value of two or greater. Also the tOp lateral
is usually rougher than the bottom lateral measurement.
Lightly loaded trucks had a somewhat lower level of vibra-
tion independent of speed. Speed increased the level of
vibration experienced.

PSD analysis was performed on the stationary seg-
ments of data. The predominant peaks for a curve with a.559
RMS level over 0-200 Hz are:
-3 :23: :3 13:3 3: H.
10'3 GZ/Hz at 140-200 Hz
GZ/Hz at 7,15,20-25 Hz

10-3 GZ/Hz at 15,25 Hz
10-4 G2/Hz at 7, 50 Hz

rear vertical

top lateral
bottom lateral

rdnawwooxm
xxxxxx
H
‘0
I
u»

Greater resolution of the lower vertical peaks over a 0-50
Hz range changed some values:

2 x 10:: Gg/Hz at 3.75 Hz

1 x 10 G /Hz at 15,25 Hz
Again the effect of speed was to raise the level of
the PSD curve slightly while maintaining the peaks, although
different peaks were more predominate at different speeds.

Rear vertical and lateral curves look much the same while

50

differing from those for forward and mid in frequency con-
tent. Heavy loaded runs exhibited larger peaks at 120 Hz.
In general dominant frequencies are found from 0-50, 60-100
and 120-200 Hz.

Peak acceleration levels were computed by the real-
time wave analyzer using 200 increments over 0-200 Hz range.
The highest levels attained were:

rear vertical .69 at 3.75 Hz
top lateral .159 at 20 Hz

The rear vertical curve also showed prominent peaks of .459
at 70 Hz and .49 at 170 Hz.

Probability density was determined from a computer
sampling of the signal at the rate of 500 samples per
second. Sensitivity was 10 mv over a maximum peak to peak
signal, 5 volts for a typical rear vertical signal. Results
show a slightly skewed distribution of accelerations ranging
from +29 to -29. The standard deviation was .469 or in other
words 68% of the data had less than that acceleration and 95%
had less than .929.

The VIBRAN graph was plotted to show the distribution
of the vibration occurrences at each frequency. The highest
9 levels attained for the smaller intervals measured were .259
at 3.75 Hz and .559 at 150 Hz. The highest acceleration
within the 95% confidence level was .19 across the entire
range. When compared with the above standard deviation these
figures reflect the effect of interval size on the total
signal amplitude expected. Smaller intervals do not allow

many superimposed amplitudes.

51

Analysis of transients was done using the real-
time wave analyzer. The records for a one second segment
can be adequately computed. This showed a large increase
in low level frequencies above the steady state level and
then a decay back to the steady state. Several transients
from all runs were measured for the peak acceleration ex-
perienced. The following are the recorded ranges of trans-

ient inputs:

rear vertical 1.7 to 2.79
top lateral .7 to 2.09
bottom lateral .3 to 1.19

Separate analysis of the local peddle run determined
the peak transient values attained to be from .3 to 19
with a more uniform distribution of peaks. Few low fre-
quencies predominated. This might have been the direct
result of a small 30' trailer with a very light load.
These values are well within the range of the over-the-road
transients.

The authors conclude that present standardized
tests are not representative of the dynamic environment as
indicated in the report. They advocate a test based on a
composite PSD plot which states more information than other
plot forms. Transients show a gradual increase in level
and could be included within the steady state tests.

The third article reviewed was used as a comparison
base. ”Current Predictive Models of the Dynamic Environment

of Transportation" by J.T. Foley was published in

52

TABLE 1. --MSU-Reported Values.

 

 

#20 - RAIL #22 - TRUCK
RMS Range vert. .079 - .3129 r. vert. .33 - .689
(2-10K) t. lat. .026 - .1169' an vert. .15 - .349
b. lat. .022 - .1299 t. lat. .13 - .289
b. lat. .13 - .309
PSD Gz/Hz Verticals ‘Verticals
max peaks 8.2 x 10-4 at 60Hz (0—500) 2 x 10% at 3.75 Hz (0-200)
3.8 x 10‘4I at 60 Hz (0-200) 6 x 10" at 15,25,70 Hz (0-200)
1 x 10"2 at 15,25 Hz (0-200)
Laterals . Laterals ‘
t. 1.5 x 10'3 at 60Hz t. 3 x 10-3 at 7, 15, 25 Hz
_4 (0—200) _3 (0-200)
b. .3xlO at mm b. 2x 10 at 15,25Hz
(0-200) _4 (0-200)
1 x 10 at 7, 50 Hz -
(0—200)
Other Peaks Other Peaks
20,40,80,120,400 0-50,60-100,120-200
Calculated RMS
.15-.389
PrObability o = .1069 o = .469
range = +.759 to -1.19 range = +29 to -29
Low'Frequency ‘Vertical
Analysis .0359 at 3.6 Hz
(0-20) Lateral
.059 at .35 Hz
PeakHMeasurememt
vert. .159 at 9 Hz
Peak Measure- vertical
ment .69 at 3.75 Hz
(0-200) .459 at 70 Hz
.49 at 150 Hz
Lateral
.159 at 20 Hz
Transients vertical .95 - 1.429 vertical 1.7 - 2.7g
Range t. lateral .14 - .409 t. lateral .7 - 2.09
1 19

:max. values

b. lateral .24 — .429

b. lateral .3

 

53

TABLE 2.--List of Variables and Instrumentation.

 

 

STUDY M80 #20 MSU #22
Objective in-service rail commercial truck
conditions shipments
SCOpe different spring different load
rates weights
different trucks
Vehicle Hy Cube 70 ton semi-trailers
100' boxcar 30',40',45' lengths
Road Type well traveled main state highway
trunkline local city road
Accelerometer range 0-19 range 0-19 and 0-109,
sensitivity sensitivity 100 mv/g
6.25 mv/g and 10 mv/g
Acc. position vertical 4 vertical

Time

Recorder

Reproducer

Amplifier

Filter

Voltmeter

Wave Analyzer

top and bottom
lateral

continuous record
1 hour cassettes

100 mv max. input
DC-400 Hz

1v max. output
100 mv/g

100 x gain
variable
.02 Hz bandpass

range:
2-10 KHz

DC-l MHz
1/400 resolution

tOp and bOttom
lateral

continuous record
1 hour cassettes

’100 mv max.

DC-400 Hz
1 RMS max.
until max. signal
5v peak to peak
.1 to 500 Hz bandpass
range:
2-10 KHz

DC-l MHz
1/500 resolution

 

54

The Journal of Environmental Sciences for January/February

 

1973. This report was a combination of selected tests

and previous studies run by the Environmental Criteria
Group at Sandia Laboratories. The fact that the same group
did all the measurements provides for a certain amount of
continuity in data reduction and presentation.

Two types of graphs are presented for each mode of
transportation. Where enough detailed data was available
to complete a statistically reliable presentation, a dis-
tributive model was given. This method attempted to use
only Gaussia random data by first removing any discrete
frequencies within. A bar chart of lo or 68% of the data
was presented as continuous broadband excitation. An envelOp
model was used to display peak accelerations obtained for
the various transient phenomena. While the envelop model
gives the maximum value obtained and is overly conservative,
the distributive model provides a degree of risk in accepting
the established levels.

Truck information was summarized over a 10-1200 Hz
frequency range for the continuous data and 0-100 Hz range
for the discrete transient data. Information was used from
seven different truck studies. The range of continuous
accelerations recorded were:

vertical .13 - .69

lateral .04 - .49
longitudinal .08 - .59

55

The lateral data is actually lower than both vertical and
longitudinal. There was not a PSD presentation given to
determine the frequency content. Predominant frequencies
were estimated from the center frequency of the bar inter-
val with the highest acceleration value. Based on this

approximation the values were:

vertical 15 Hz
lateral 60 Hz
longitudinal 110 Hz

Peak transient accelerations as measured over the l-lOO Hz

range were:

vertical 109 at 3.75 Hz and 10 Hz
lateral 1.29 at 30 Hz
longitudinal 1.29 at 15 Hz

Foley states that the major sources of excitation

input for a truck depending on speed are:

suspension system 3 - '6 Hz
tires 15 - 25 Hz
drive train ' 60 - 80 Hz
frame 100 -120 Hz

Comparison will show that these frequencies do dominate.

Rail transport was summarized over a 10-350 Hz fre-
quency range for a 10 distribution model and 1-100 Hz range
for the enveloped transient model. Information was used
from 22 different events. Foley states that data was only
available to 350 Hz and claims that the accelerations had
leveled Off by that point. The range of continuous accel-
erations recorded were:

vertical .03 - .lg

lateral .023 - .0559
longitudinal .018 - .0559

56

Predominant frequencies were again estimated from the

center frequency of the bar interval. The results are:

vertical 30 Hz
lateral 100 Hz
longitudinal 60 Hz

Peak transient occurrences not including switching were:

vertical 4.59 at 70 Hz and 3.75 Hz
lateral 59 at 30 Hz
longitudinal 59 at 30 Hz

Comparison with the major frequencies as stated by

Foley for rail vehicles again shows some relationship:

suspension system 5 - 10 Hz
frame 60 - 100 Hz
track input 10 - 30 Hz

TABLE 3.--Foley Reported Values.

 

 

Foley - RAIL Foley - TRUCK
RMS Range (0-350) (lo-1200)
Vert. .03 -.lg Vert. .13 - .69
lat. .023 -.0559 lat. .04 - .49
long. .018 -.0559 long. .08 — .59
Predominant vertical 30 Hz vertical 15 Hz
Frequencies lateral 100 Hz lateral 60 & 1100 Hz
longitudinal 60 Hz longitudinal 110 &
1000 Hz
Transients (0-100) (1-100)
vert. 4.59 at 70 Hz vert. 109 at 3.75,
10 &1100 Hz
lat. 5.09 at 3.75 & lat. 1.259 ‘at 30 Hz
30 Hz

long. 5.09 at 30 Hz long. 1.29 at 3.751&
15 Hz

 

57

Another study conducted in 1970 for Sandia by Gens
is used in conjunction with Foley's report. It covers the
vibration experienced by a flat car and the apparent weight
of its cargo. This study was presented as a detailed table
of the VIBRAN program. Approximate Jr! levels were ex-
tracted for inclusion with Foley's values. Predominant
frequencies were again centered within presented interval
and were compared with Foley's results. Individual levels
are not discussed here but are presented in Table 4.

Deciding which values to compare among the varied
presentations was difficult to do. A straight forward com—
parison of acceleration levels necessitated having the same
level of probability of occurrence. Foley's 10 levels were
taken to be nearly equivalent to MSU's RMS levels according
to statistical theory and sinusoid measurement. The values
for different vibration directions such as mid and aft
verticals and top and bottom laterals were combined to make
one value range for comparison.

A more difficult problem was the ranges presented
which do affect the values for acceleration levels and fre-
quencies. Due to the number of driving frequencies and
structural frequencies present in the vehicle, more than one
frequency may be of an equal intensity across the spectrum.
Indeed a true random situation would exhibit equal values.
Division at a driving frequency leaves a decision as to its

real value. The accuracy of the two groups' data methods

 

58

a:
N3

 

 

efi an ex... .83
ofi pm 3. .33
mm e pm 3 .935
859mm same
we com um Mmm. .93
3 gm mm. 52
58. .82 mm o . m m . we
$8.. .93 Human Hut
3. It"? mmmmm
8878 we
was. I am.
Em
a 92 @2955
mm omm I mS mm mm 8 mm mm mNImS
stood pm on an? Same mm. I me. we mg I Tm mm manm .83
mofiammm Same 8°78 52mm mam mfiozmoommm .28 mm 378 33
ad. I .8. mm meIom .5
mm SA a 3 38> mmo 9502mm mm um. I m. E
momma Va: magma
m8; I H. Rd. I S. .83
3. v .fime lemme 82am mm 2 a m mm TS x me am. I mm. Hm. I 8. 33
887: TE $0 $802 one gem Beam one. I 2. 38>
mam
ensue .988 gonna Hm Hmoxom .m.m 98me .m.m
m? 29.60 am. $98 - SEES mane mzmo

 

.mphommm QfiosmfimomflT . 9 gm.

59

and instrumentation could not be compared since Foley did
not explain his in the magazine.

Since Foley's report is a composite of different
individual studies, it would be expected to have a wider,
more extreme range of values. This is verified. Like-
wise, since trucking encounters a more diverse road bed
and input, these vibration levels would be expected, and
are larger than those experienced for rail transport.

The relevant figures are summarized in Table 5.

Rail vibration levels measured by MSU fall within
the range established by Sandia. The frequency values are
somewhat the same. A problem occurs where the out Off fre-
quency of the interval is the predominant frequency in
PSD analysis. These predominant frequencies are higher than
those recorded by Guins in 1953 of 2.5-7.5 Hz and 55-60 Hz.
Detailed low frequencies analysis over 0-20 Hz presented by
MSU reveal very low accelerations: .

vertical .0359 at 3.6 Hz
lateral .05 g at .35, 1.2 and 9 Hz

compared with the .29 to .59 given in Guins report. This
may be as a direct result of improvement in rail cars and
the data collection and analysis.

One other source for comparison was a report pre-
sented by Luebke. He suggests the rail environment has
dominant frequencies at 5 Hz and 70 Hz with a range of

.03 - .39. This is close to the MSU figures.

60

 

 

 

am on em mm; III we em um 86 III 553
um om um um; SS mm mas a omen 9m awe. B933
at 2 an. 52 ohm mm on em. mmé awe; Hons? meamzmme
III we com um mam. guests
at cm um mma. um ooH um mam. Hmumpma mOZHoamm
mm me.m em 8. mm Tm um 3 Huge mama
mm o: III mm 08.8 Hmfisfiamcoa
um om um om.m~.ma.> um ooa.om Hm om Hmumuma mMHozmsommm
um ma um on.ma.me.m um om mm ooe.o~a.om.om.oe.om Hmoauume Luasnumxu
am. I we. III 58. I as. IIII guesses
me. I 3. 58. I 2. Re. I m8. mama. I «no. H883 82mm
8. I ma. m8. I me. on... I 8. amen. I as. Huge ma
memo a meson ems memo a sweet am:
zooms Seam

 

.mcomflnmmeuo fiommmIfi mg

61

Transients measured were much higher in Foley's
report though still at dominant frequencies. The transverse
values are almost the same as the vertical values.

Truck vibration levels measured by MSU were almost
the same as those of Sandia. Foley used a wider analysis
range so that those values could be expected to be higher.
He also reports that the values for the longitudinal direction
exceeded those for the transverse direction. In general
any of the directions can record the highest level within
some part of the spectrum. The predominant frequencies in
the truck environment are lower than those for rail. There
is considerable activity between 0 Hz and 25 Hz. The trans-

ients experienced are larger than those recorded for rail.

VIII. CONCLUSIONS

The preceeding examination of vibration and its
measurement in transportation environment still leaves con-
siderable room for improvement. Until certain standards
are formed this area will remain an unwieldly conglomeration
of data. Vibration measurement, analysis, and presentation
should be done in an orderly and useful manner. Indiscrim-
inate use of numbers or unsatisfactory equipment cause a
misrepresentation of the phenomena as experienced by an
actual shipment. Many areas remain to be tackled in extensive
studies to give a more complete picture of the real environ-
ment.

Orderly and exact collection techniques are the
responsibility of the researcher. It is also his responsi-
bility to state the objectives and methods used when the
figures are presented. This will allow others to know what
has been done and allow a possible comparison. Statements
of omissions or mistakes are just as important for under-
standing.

Presentation of vibration data should be in a useful
and realistic form. Since the transportation environment

is a random phenomena, probability theory should be

62

63

incorporated to establish the chance of occurrence. Units
used for description should be easily understood and useable
for comparisons. Accurate conversion for inclusion in
mathematical calculations is necessary.

The engineer must have an adequate base of information
about the amplitudes and frequencies expected. Information
on cushioning and natural frequencies is needed to develOp
the resultant input to the product due to package inter-
action. A useful chart is one which will allow these values
to be extracted for use in package design. A package should
be able to withstand the input, maintaining its function of
protection and marketability throughout the distribution
phase. The decision of the extent of vibration protection
will involve a trade-off between package costs and cost of
possible loss from damage. .

The most conservative presentation design for a
mode is the composite envelOp method covering the peak
amplitudes across the fullest possible array of variable
parameters. When comparing these the truck vibration en-
vironment seems to be the most severe excluding the rail
switching impacts and airline's high frequencies. This form
would seem to be the single most useful presentation since
virtually all product travels by truck some of the time.

The most statistically useful presentation from the
standpoint of the magnitude of input is the VIBRAN format.

Comparison of the different levels of occurrence in a

64

particular environment allows an Option for a greater degree
of risk in lower protection levels as a cost consideration.

The method which yields the best frequency informa-
tion is the PSD curve. It shows the dominant frequencies
and has the average RMS acceleration level of the curve
as measured by the area under the curve. Use of the real-
time wave analyzer requires considerably less time and cost
than the computer program VIBRAN. This method is better
because of the continuous uninterrupted analysis across the
range which does not cause measurement problems at the inter-
val cuts.

As is generally the case, one chart cannot adequately
define the information required for the vibration experienced.
Combining innumerable variables will make it very difficult
to confidently generalize because there will always be
exceptions. One reason for the definition of vibration input
is to have realistic values for use in developing an ideal
package for any product. Analysis of the package and the
product reaction to the input is important. This analysis
can be conducted more economically in lab simulation than
in test shipments.

The second reason is a compilation of charts which
represent the entire vibration field and all its variables
is necessary for definition of test limits. Current methods
for testing packages are inadequate and incomplete.

Vibration tests are conducted to simulate the environment.

65

ASTM specifies a continuous one inch amplitude input to a
mechanical shaker which is a very large input for most
packages. The test is customarily conducted for one hour.
The result of this test method is acceleration in excess of
19 and either immediate failure or none at all.

Test equipment is becoming more sophisticated.
Exact input for frequency, amplitude, and force may be
closely regulated to approximate levels of the environment.
Vibration measurement is required to establish exact inputs
for simulation. Current controversy is over the constant
force input or constant amplitude input. Frequency may be
taken care of by a suitable sweep time which would allow
enough time for product to react. Established driving fre—
quencies could be singled out for a longer dwell time. New
vibration standards should be of general nature to allow
for both large and small product and different types of
equipment. Of course each engineer can set more rigorous
limits in line with his product, distribution system, and
test machine.

Other subjects which were previously mentioned in
earlier sections would be of further interest for clarifica-
tion and definition of the vibration environment. These
areas are necessary in defining the vibration limits and
repercussions so that all calculations would be done with

assurance .

66

A set of standards is needed for the measurement of
vibration. Previous studies have shown a tremendous varia-
tion in technique. For a detailed investigation instrumen-
tation ranges, placement, and time requirements should be
specified to allow more exacting compilation and comparison.
The specific parameters such as vehicle load, road type,
and Operations should be exactly stated. Measurement Of
the vibration effect on cargo in ships and planes needs to
be done and these must also have placement specifications.

Of more immediate concern is clarification of the
complex vibration experienced in piggyback and other con—
tainerized shipments. Palletized shipments should be exam-
ined to trace the effect of added weight and the dynamics of
stacking.

A more precise measurement of driving frequencies
and structure of vehicles is needed. Present statements on
the subject cover a significant range. The calculation could
be made to show the effect of speed and load variables.

A study of the typical natural frequencies of
various products would be informative. Fragility factors
for shock endurance have already received much attention
from both a how to define standpoint and common values for
different product types. A knowledge of the natural
frequency of a product and its interaction with various

cushioning materials would allow accurate designs.

67

Cushion materials should have curves relating
natural frequency to the static loading for different
cushion depths. Also a statement of the expected trans-
missibility curves for the cushions. The vibration input
from stacking arrangements is important.

Low frequency should be more clearly resolved to
determine exact magnitudes and the occurrence of transients
in this area. Their implication for design in relation to
typical natural frequencies and amplification factors should
be carefully examined to see if this area is really critical
in relation to its high cost of analysis.

Other aspects of environment affecting vibration
levels should also be examined. The effect of temperature
on the natural frequency of cushioning materials and the
effect of humidity on the strength and transmissibility of
corrugated containers is critical. The amount of dynamic
stress able to be handled by these package materials in
adverse conditions should be plotted.

Knowledge of these factors would allow a more real-
istic evaluation of transportation surveys. All shipments
of product which sustain damage should be carefully examined
for unusual circumstances before redesign. Interaction of
these other factors may influence the amplitude or dominant
frequencies registered for a specific incident.

In conclusion, the importance of the transportation

vibration environment in packaging must be stressed.

68

Although the 9 levels are not as high as those for shock

input from rough handling, the result of a steady state

input can be just as devastating. The exact damage value
attributed to vibration is unknown. However, as shown by

the examples vibration does cause considerable problems.

Any change in product, transportation mode or package design
should be suspect and checked out for possible repercussions.
PrOper understanding and emphasis is essential for eliminating
this type of damage. The engineer needs realistic and

clearly defined input information in order to thoroughly

design the package.

REFERENCES

69

REFERENCES

Anderson, J.J. "Real-time Spectrum Analysis in Vibration
Testing." ISA Transactions, No. 10, No. 3, 1971.

Bendat, J.S. Principles and Applications of Random Noise
Theory. 1958. Wiley and Sons, New York.

. "Discussion of PrOposed USASI Standard on Methods
for Analysis and Presentation." Shock and Vibration
Bulletin, Vol 39, Part 6, 1969.

Beraneck, Leo L. (ed.). Noise and Vibration Control. McGraw-

Hill, New York. 1 71.

Bozich, D.J. "Data Handling Methods for Large Vehicle Testing."
Shock and Vibration Bulletin, Vol. 37, Part 5, 1968.

Bishop, R.E.D. Vibration, 1962. Cambridge University Press,
London.

 

Cerni, R.H. Instrumentation for Engineering Measurement.
Wiley and Sons, New York, 1962.

Crede, C.E. Vibration and Shock Isolation. Wiley and Sons,
New York, 1951.

Foley, J.T. "Current Predictive Models of the Dynamic Environ-
ment of Transportation." Journal of Environmental

Sciences. January, 1973.

. "Normal and Abnormal Dynamic Environments En-
countered in Truck Transportation." Shock and
Vibration Bulletin, Vol. 39, Part 6, 1969.

Gens, M.D.~ "Rail Transport Environment." The Journal of
Environmental Science. July/August 1970:

Godshall, W.D. Effects of Vertical Dynamic Loading on

Corrugated Fiberboard Container. Forest Products
Laboratory, USDA. July 1968.

70

71

Harris, C.M. and Crede, C.E. (eds.). Shock and Vibration
Handbook. McGraw-Hill, New York, 1961.

 

Hunter, N.F. "Measurement of Mechanical Importance and Its
Use in Vibration Testing." Shock and Vibration
Bulletin, Vol. 42, Part 1, 1972.

Kazmierczak, F.F. "Objective Criteria for Comparison of
Vibration Environment." Shock and Vibration Bulletin,
Vol. 41, Part 3, 1970.

Luebke, R.W. "The Box Car Dynamic Environment." Shock and
Vibration Bulletin, Vol. 41, Part 4, 1970.

Malvino, A. Electronic Instrumentation Fundamentals. McGraw-
Hill, New York, 1967.

 

Marcus, A. Measurements for Technicians. Prentice-Hall.
Englewood Cliffs, N.J., 1971.

 

Murfin, W. "Dual Specifications in Vibration Testing."
Shock and Vibration Bulletin, Vol. 38, Part 1, 1968.

Ostrem, F.E. "A Survey of the Transportation Shock and
Vibration Input to Cargo." Shock and Vibration Bulletin,
Vol. 42, Part 1, 1972.

Pear, C.E. (ed.). Magnetic Recording in Science and Industry.
Rienhold Publishing Corp, New York, 1967.

 

Root, L. "Vibration Equivalence: Fact or Fiction." Shock
.and Vibration Bulletin, Vol. 39, Part 2, 1969.

Ryder, J. Electronic Fundamentals and Applications. Prentice-
Hall, Englewood Cliffs, N.J., 1964.

 

Schlue, J.W. "A New Look at Transportation Vibration Sta-
tistics." Shock and Vibration Bulletin, Vol. 37,
Part 7, 1968.

Sharpe, W.N. and Goff, J.W. Preliminary Investigation of
Freight Car Vibration. Michigan State University
School of Packaging. Technical Report No. 20, 1972.

Sharpe, W.N., Goff, J.W., and Kusza, T.J. Preliminary
Measurement and Analysis of the Vibration Environ-
ment of Common Motor Carriers. Michigan State
University School of Packaging. Technical Report
No. 22, 1973.

72

Thaller, R.W. "Narrow Band Time History Analysis of Trans-
port." Schock and Vibration Bulletin, Vol. 44,

Part 4, 1974.

 

Tustin, W. "Reducing In-Transit Damage to Fruit." Test.
December/January 1973-74.

"II III'IIIIIIIIIIIIIIII