WmmL
Lu...WWW\v~tk.~.

gnu” “mm bun:
.LMNWHWWWQQ
: memfla-

an...

. aft-Lain...
L , I

. 2? 59...!
‘ 5‘-

 

 

12' ,L,
"1.

3.

u I'n‘ ”3347 -
L 63:31.4; gig
m" 832E:

’91' . .

fi~~ w/é ‘
_ V’W‘piiwa‘fif '2?
Ififv’vfifi’?‘ A'JLM .

4

ti

31

."nnplnll
tadnfinpldfl‘h hhci .!
4 vulll' , V I'l

all 'L‘l l I
HWWMWIN u LOW

 

g .

 

L . . . . .. o ‘ n||
Anviqtrl . |.. .‘ .Jlil‘WuWLJvoh‘lnwllHnmf‘i‘v.v{t!
Lfir‘f .014, ... 30.5! u‘. 0.. 0.04" I "Ll, J. . .1' u 1 cl».

2333 llllllllllllllllllNW!IllH(lllfllllllllllllllllllllllllll

(WP) 1293 01716 4207

This is to certify that the

thesis entitled

Theoretical Predictions and Observations of
Peak Deceleration Levels for Perfect Edge and
Corner DrOps

presented by

John Dominic Jackson

has been accepted towards fulfillment
of the requirements for

Master degree in Packaging

 

aged! flange/LU

ajor professor

Date December 18, 1292

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

‘Iw '— v “—

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINE return on or before date due.

DATE DUE I DATE DUE DATE our
N 1

r771“:§"1‘-nt1.3‘..-
19....‘ua'u \1

a.
r.“ 1
> :.\..-v - ' d

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

me mm.“

THEORETICAL PREDICTIONS AND OBSERVATIONS OF PEAK
DECELERATION LEVELS FOR PERFECT EDGE AND CORNER DROPS
By

John Dominic Jackson

A THESIS

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

MASTER OF SCIENCE

School of Packaging
1 997

ABSTRACT

THEORETICAL PREDICTIONS AND OBSERVATIONS OF PEAK
DECELERATION LEVELS FOR PERFECT EDGE AND CORNER DROPS

BY

John Dominic Jackson

Statistically. combinations of edge, corner and flat drops occur in the distribution
environment as opposed to that of flat drops alone. To date, little work has been done
to study peak deceleration levels of either condition. This reason is two-fold; firstly, it is
difficult to identify the “true” impact geometry of either condition, and secondly, because
it is assumed that passing the flat drop test (which produces the highest G’s) provides a
built-in protection factor against all other impacts.

This study examines the performance of 2 pcf LDPE foam inside a corrugated
box, under simulated perfect edge and corner drop conditions. The experiments
followed ASTM procedures using five repeated drops from two drop heights of 24 and
36 inches. Two package weights of approximately 20 and 40 lbs per box were used. A
similar study was conducted, however, the drop heights ranged from 18", 24", 30", 36"
and 42'. Unlike ASTM standards. 8-10 minute intervals were given between each drop
for maximum cushion recovery.

A theoretical equation showed a very close correlation between the “predicted”
and 'actual' G levels to within :1: 10%. Shock pulse data suggests that the first impact is
absorbed by the corrugated board box. while the 2-5 impacts were absorbed mainly by
the cushionas the box gradually softens. The model confirmed that theoretical G levels

can also be calculated for varying edge lengths. Two lengths of 4.5 and 9 inches were

compared.

DEDICATIONS

To my mother
A woman who looked after me well and never discouraged me from doing anything I wanted too.

Most of the time I

In loving memory of my uncle: Thomas Doyle. Esquire.

Living proof that Santa Claus did exist Start the show and take it easy!

To Dr. Stuart and Dr. Jodi Jackson

For all of their spiritual, emotional and (on occasion) financial support.

To my wonderful remaining family:

Dr. and Mrs. Mark and Joanne Jackson
Colette, Graham, Michael, Martin, Jamie and Daniel
Chris, Lynne, Jessica and Fiona
Mr. Frank ‘Rocket Man' Owens

To the ‘late’ Catherine Richards.

Dr. David McCurdy, Darren and Warren, Dave and Sheila. the Zuc-Machine, Stuart Jennett (for

keeping me inspired for all those years) and the remaining brothers Jennett ( John and Jim),

Chris J.. Gerard and Theresa J., Sharon 8., Douglas J., Adam R., and Richard B.

And finally to Dhruti. What can I say about this person without making her blush.

This is me signing off. Till next time.

ACKNOWLEDGMENTS

There are only three (not two or four, but three) things that spring to mind when I

hear the name Gary Burgess:

(1 ) Genius!
(2) Simplicity!
(3) Patience (alot of it)!
(4) Humor!

And other twisted and bizarre things also come to mind, however, I don’t think

we should mention this stuff here (HA,HA)!

I would also like to thank my remaining committee members: Dr. S. Paul Singh &
Dr. Brian (he’s that nice ‘Chaos’ guy) Feeny. I would also like to thank Jason
Chaneske and Stephane Fayoux for their friendship and advice and patience

during my program and research.

A final thanks goes to the people in the background who made my academic

study more enjoyable:

Mr. Scott Kibler (my fourth brother);
To Dr. Robert LaMoreaux (my second dad) and his wonderful family.
Dr. Diana Twede and Dr. Bruce Harte.
Mr. Donald Abbott; and

Ms. Beveny Underwood.

iv

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

CHAPTER 1. INTRODUCTION AND LITERATURE REVIEW

1.1
1.2
1.3

Introduction

Literature Review

Choosing The Right Cushioning Material
1.3.1 Implications of Choosing Foam Types
1.3.2 Modelling Cushion Behavior

1.3.3 Pilot Study

1.3.4 Relation Between Edge and Flat Drops

CHAPTER 2 - EXPERIMENTAL DESIGN FOR

2.1

2.2
2.3
2.4

EDGE AND CORNER DROPS

Test Setup

2.1.1 Test Condition 1
2.1.2 Test Condition 2
2.1.3 Test Condition 3
Edge Drop Test
Corner Drop Test
Equipment

CHAPTER 3 - RESULTS AND THEORETICAL DEVELOPMENT

3.1

Results

viii

xii

10
11
17
19
20

23
27
28
28
29
29
31

33

3.2

3.3
3.4

TABLE OF CONTENTS - CONTINUED

Theoretical Development

3.2.1 Edge Drop Situation

3.2.2 Corner Drop Situation
Limitations of the Theoretical Model
Limitations of Curve Fit Software

CHAPTER 4- DISCUSSION I CONCLUSIONS

4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8

4.9

Discussion

The Affects of Weight and Cushion Thickness
Edge Drop Results

Comer Drop Results

Comparison of Repeated vs Incremental Drop Tests
Behavior of Shock Pulses During Impact
Conclusions

Experimental Errors

4.8.1 Test Method

4.8.2 Machine Error

4.8.3 Foam Fabrication

4.8.4 Corrugated Board Fluting

4.8.5 Corrugated Board: Box Assembly
4.8.6 Jig Design

4.8.7 Perfect vs Non-Perfect Drops

4.8.8 Filtering

Recommendations and Future Work

vi

33
33
50
51
51

63

65
66

68
72

73
74
75
75
76
78
79
80
84

TABLE OF CONTENTS - CONTINUED

APPENDICES

Appendix A - Figures (7-10) - Edge Drop Cushion Curves
1-5th Impacts

Appendix B — Figures (11-14) - Corner Drop Cushion Curves
1-5th Impacts

Appendix C — Figtures (16-20) - Edge Drop Shock Pulses
1- Impacts

Appendix D - Figures (21-26) - Comer Drop Shock Pulses
1-5 " Impacts

Appendix E - Tables (37-42) - Edge Drop Experimental Data

Appendix F — Tables (37-42) - Corner Drop Experimental Data

BIBLIOGRAPHY

vii

86

91

96

106

118

126

133

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table 1 1.

Table 12.

Table 13.

LIST OF TABLES

Handling Environments
Properties of Bulk Cushioning Materials

24” and 36" Edge Drop Results for 2" Cushion
Thickness - 1" Impact. Base Dimensions of Box Edge
Measures 4.5lnches

24” and 36” Edge Drop Results for 2" and 3" Cushion
Thickness - 2-5 Impacts. Base Dimensions of Box Edge
Measures 4.5lnches

24” Edge Drop Results for 2” and 3" Cushion Thickness
11“ Impact. Base Dimensions of Box Edge Measures
9.0 Inches

24” Edge Drop Results for 2” and 3" Cushion Thickness
2-5 Impacts. Base Dimensions of Box Edge Measures
9.0 Inches

36” Edge Drop Results for 2” and 3” Cushion Thickness
1" Impact. Base Dimensions of Box Edge Measures
9.0 Inches

36” Edge Drop Results for 2” and 3" Cushion Thickness
2-5 Impacts. Base Dimensions of Box Edge Measures
9.0 Inches

18” Edge Drop Results for 2" and 3” Cushion Thickness
1" Impact. Base Dimensions of Box Edge Measures
9.0 Inches

24", 30", 36’ and 42" Edge Drop Results for 2" and 3" Cushion
Thickness - 2-5 Impacts. Base Dimensions of Box Edge
Measures 9.0 Inches

24" Comer Drop Results for 2” and 3” Cushion Thickness
1st Impact

24” Corner Drop Results for 2” and 3” Cushion Thickness
2-5 Impacts

36" Corner Drop Results for 2” and 3” Cushion Thickness
1’t Impact

viii

15

35

36

37

38

39

40

41

42

43

Table 14.
Table 15.
Table 16.

Table 17.
Table 18.
Table 19.
Table 20.
Table 21.
Table 22.
Table 23.

Table 24.

LIST OF TABLES - CONTINUED

36”Comer Drop Results for 2” and 3" Cushion Thickness
2-5 Impacts

18" Corner Drop Results for 2” and 3” Cushion Thickness
1“ Impact

24", 30”, 36’ and 42” Corner Drop Results for 2" and 3”
Cushion Thickness - 2-5 Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 24” Edge Drop for 2" and 3" Cushion
9" Edge Length — 1" Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 24” Edge Drop for 2” Cushion — Comparing
4.5” and 9" Edge Length — 18t Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 36” Edge Drop for 2” and 3" Cushion
9" Edge Length - 1“ Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 36” Edge Drop for 2” Cushion - Comparing
4.5” and 9” Edge Length - 1"t Impacts

Percent Difference Between Curve Fit-and Experimental
G levels. 24" Edge Drop for 2” and 3” Cushion
9” Edge Length — 2-5 Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 24” Edge Drop for 2" Cushion - Comparing
4.5” and 9” Edge Length - 2-5 Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 36" Edge Drop for 2" and 3" Cushion
9" Edge Length - 2-5 Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 36” Edge Drop for 2” Cushion - Comparing
4.5” and 9" Edge Length - 2-5 Impacts

23

24

25

55

55

56

57

58

59

Table 25.

Table 26.

Table 27.

Table 28.

Table 29.

Table 30.

Table 31.
Table 32.
Table 33.
Table 34.
Table 35.
Table 36.

Table 37.

LIST OF TABLES - CONTINUED

Appendix D

Percent Difference Between Curve Fit and Experimental
G levels. 24" Corner Drop for 2” and 3" Cushion

No Edge Length - 1st Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 36" Corner Drop for 2” only

No Edge Length - 1st Impacts

Percent Difference Between Curve Fit and Experimental
G levels. 24" Corner Drop for 2” and 3” Cushion

No Edge Length — 2-5 Impacts

Percent Difference Between Curve Fit and Experimental

G levels. 36” Corner Drop for 2” and 3” Cushion
No Edge Length - 2-5 Impacts

Appendix E

24 Inch Edge Drops on 2" Cushion WIth Box Edge Dimensions
Measuring 4.5” Along Edge

36 InchEdge Drops for 2" Cushion WIth Box Edge Dimensions
Measuring 4.5” Along Edge

Edge Drops on 2" Cushion With Box. 9 Inch Edge Length
Edge Drops on 3" Cushion WIth Box. 9 Inch Edge Length
Edge Drops on 2” Cushion With Box. 9 Inch Edge Length
Edge Drops on 3" Cushion With Box. 9 Inch Edge Length
Edge Drops on 2" Cushion With Box. 9 Inch Edge Length
Edge Drops on 3" Cushion With Box. 9 Inch Edge Length

Appendix F

24" Corner Drops on 2" Cushion With Box. No Edge Length
x

60

60

61

62

119

119

120
121
122
123
124

125

127

LIST OF TABLES - CONTINUED

Appendix F

Table 38. 24“ Comer Drops on 3” Cushion With Box. No Edge Length 128
Table 39. 36” Corner Drops on 2” Cushion With Box. No Edge Length 129
Table 40. 36” Corner Drops on 3” Cushion WIth Box. No Edge Length 130

Table 41. 18", 24”, 30”, 36" and 42" Corner Drops on 2" Cushion With Box 131
No Edge Length

Table 41. 18", 24”, 30”, 36” and 42” Comer Drops on 3” Cushion With Box 132
No Edge Length

xi

Figure 1.
Figure 2.
Figure 3.
Figure 4.

Figure 5.

Figure 6.

LIST OF FIGURES

‘Vector’ Relationship Associated With Oblique Shocks
Cushion Tester With Test Apparatus (not to scale)

Edge Drop Jig Using Two Edge Lengths

Corner Drop Jig With Undefined Impact Geometry

Set of Typical Cushion Curves For a 24 Inch Drop
First, and Second Through Fifth Impacts

Model of a Linear Spring Mass System

Appendix A (Figure 7-1 1)

Figure 7.
Figure 8.

Figure 9.

Figure 10.

24” Edge Drop onto 2” and 3” Cushion - 1st Impacts
36” Edge Drop onto 2” and 3” Cushion - 1st Impacts
24” Edge Drop onto 2” and 3” Cushion - 2-5 Impacts

36” Edge Drop onto 2” and 3” Cushion - 2-5 Impacts

Appendix B (Figure 11-14)

Figure 11.
Figure 12.
Figure 13.
Figure 14.

Figure 15.

24” Corner Drop onto 2” and 3” Cushion - 1st Impacts
36” Corner Drop onto 2” and 3” Cushion - 1st Impacts
24" Corner Drop onto 2" and 3” Cushion - 2-5 Impacts
36" Corner Drop onto 2” and 3” Cushion - 2-5 Impacts

Shock Pulse For 24” Edge Drop Using 3” Cushion
1st-5th Impacts

xii

21
24
25
26

32

49

87

88

89

90

92
93
94
95

72

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.

Figure 26.

Figure 27.

LIST OF FIGURES - CONTINUED

Appendix C (Figure 16-20)
Shock Pulse For 24” Edge Drop Using 2” Cushion
1st-5th Impacts

Shock Pulse For 36” Edge Drop Using 2” Cushion
1 st-5th Impacts

Shock Pulse For 36” Edge Drop Using 3” Cushion
1st-5th Impacts

Shock Pulse For 18”, 24”, 30”, 36”, and 42“ Edge Drop
Using 2” Cushion - 1st-5th Impacts

Shock Pulse For 18”, 24”, 30”, 36”, and 42“ Edge Drop
Using 3” Cushion - 1st-5th Impacts

Appendix D (Figure 21-26)
Shock Pulse For 24” Corner Drop Using 2” Cushion
1st-5th Impacts

Shock Pulse For 24” Corner Drop Using 3” Cushion
1 st-5th Impacts

Shock Pulse For 36” Corner Drop Using 2” Cushion
1st-5th Impacts

Shock Pulse For 36” Corner Drop Using 3” Cushion
1st-5th Impacts

Shock Pulse For 18”, 24”, 30”, 36”, and 42" Comer Drop
Using 2” Cushion - 1st-5th Impacts

Shock Pulse For 18”, 24”, 30”, 36”, and 42” Corner Drop
Using 3” Cushion - 1st-5th Impacts

Perfect Edge and Corner Drop Conditions (located in text)

xiii

97

98

99

100

101

103

104

105

106

107

108

81

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

1.1 Background

In packaging, there is the option to either intentionally under or over
package products. Traditionally, companies have tended to underpackage a
product in order to reduce packaging costs at the expense of increasing product
damage. This runs the risk of damaging consumer loyalty and in the long run
may result in much higher packaging costs. Eventually these costs are passed
onto the consumer. The other option is to overpackage a product in order to
reduce damage levels considerably, but at the expense of increasing packaging
costs. At some point this will affect the consumer as they are faced with the

added costs of shipping and unnecessary excess packaging.

It has been estimated that damage to consumer products is as high as 10
billion dollars per year in the US. alone (1). This amount implies that products
a3 generally underpackaged, despite views within the shock and vibration world
that products are on the contrary, overpackaged (this will be discussed later in
the chapter). At some point there has to be some kind of trade-off between
damage and packaging, but at what cost? Another question that should be
asked is whether manufacturers are willing to accept a certain amount of damage
and live with the resulting financial loss versus the costs saved from

underpackaging.

2

The packaging industry, as a whole, fails to realize that in a real
distribution environment, much lower G’s are experienced by a package
compared to those encountered in testing laboratories. Tests used for designing
and assessing packaging performance are ASTMD-4169: Standard Practice for
Performing of Shipping Containers and Systems (2), the Hazmat / UN DOT and
International Safe Transit Association (ISTA) standards. All of these tests use
worst case scenario flat drop situations, in spite of the observed handling

environments shown in Table (1).

Table (1), tells the designer/engineer how to define the expected drop
height and environment. Based on certain parameters such as the package
weight and given dimensions, you should generally expect to encounter the drop
heights shown, impacting in a certain direction and as a result of being handled
in a certain way. “ It is generally agreed that, regardless of the transportation
mode, the severest shocks likely to be encountered in shipping result from
handling operations” (3). These Shocks can occur as a result of dropping the
package onto a floor, dock or platform. The information in Table (1) suggests
that conditions experienced in the ‘real’ distribution environment are not like
those recreated in the test lab. If a more realistic situation combines side, corner
and even edge impacts, then the question we need to ask ourselves is why is

this not accounted for in lab simulations?

To date, little attempt has been made to observe or characterize the
physical behavior and the peak G deceleration levels that occur in the less
severe, but more frequent edge or corner impacts under dynamic extremes. The
ASTM-D 1596 standardzlest Methodeor nggmicShock _C_ushioning
Characteristics of Packaging Material has not yet accounted for these types of

 

 

 

 

 

 

 

 

 

 

_8_c2oos_ a: to =2 .93 353 .9583. a. BEE: 88A
39:28.2 2. 3 =2 .98 355 .9580”. 3 pegs—=5 8868
39:235. a... ..o =8 .28 65.0 .9580". an an 898w
E06 :9: 95 558 3 e2» >c< 3 cm cowém
E8 cmE 0:0 .658 6 02m >c< «N 8 8709
E8 :9: 0.5 ono 8 09m >c< N me 873
E8 :9: one 8E8 to 02» E4. on mm 3.8
:25 5E one 858 to 02m >c< we we owv
305:: 305:: an:
22»: 8.88.5 2%;
9593: .0 E0“. new 3 0%... no.0 $93.0 monsoon.
8.68m.

 

 

 

 

 

«EoEco.._>:m 9:55: ._. o3!—

 

4

scenarios. This reason is two-fold: firstly, it is difficult to identify the true impact
geometry in an edge and a corner and secondly, because it is assumed that
passing the flat drop test (which produce the highest G’s) provides a package

with a built-in protection factor against all combinations of impacts.

It is considered a more rigorous test of the cushioning to consider
primarily flat drops, as the greatest Shock deceleration reaches the product when
dropped only on the flat surface, when compared to a drop on an edge or corner.
It is usually the case that perfect flat drops cause the most internal damage to
the product. Based on the results of the flat drop test, it will be important to
select the correct primary package or secondary package and an appropriate
cushion with particular characteristic(s) that will prevent these high deceleration

G levels being transmitted directly to the product.

Packages that ‘pass’ the flat surface test are very likely to pass edge,
comer or combination surface drops. When the package design passes the flat
drop test and package materials meet specifications, shipping shock damage of
a product is rarely a legitimate issue. It is for this reason that the established
procedure for cushion curve design is to protect against fiat drops. However,
“we never see perfect flat drops in a real distribution environment” (3). Packages
are rarely dropped precisely on one flat surface, except in the test lab (with a
shock machine). Another point worth noting is that of the differences in drop
heights conducted by test facilities and those encountered in the ‘real’
environment. “Currently, test drop heights are considerably more stringent than

general transportation standards” (4).

5

In all distribution environments, non-perfect flat drops occur much more
frequently than perfect flat drops. This is because in a drop situation, the base
experiences an initial impact followed by a combination of secondary impacts
either over an edge and flat surface or a comer and flat surface. Rather than
compressing the cushion, a large portion of the impact energy is dissipated
through rotation of the whole package. This is also due to both normal and
frictional surface forces acting on the product which prevents maximum

deceleration being transmitted directly to the product.

The American Society of Testing and Materials test method: ASTM D-
4169 has three assurance levels. All three specify the level of test intensity - one
of which must be specified for your particular product [package system prior to
testing. Choosing which test depends on several factors such as product value,
knowledge of the shipping environment, the number of units being shipped and
the desired level of damage that can be anticipated and tolerated. Choosing an
“assurance level” of three provides the least severe test, while assurance ‘level
one’ provides the most severe test. The ‘level two’ test is a medium intensity
level test that is generally specified unless any of the above criteria suggest

otherwise.

In terms of the aesthetics of a package, there are also two types of
acceptance criterion used in ASTM D-4169. “Criterion 1’ specifies that the
product is damage-free, while “criterion 2’ specifies that the package is intact (as
well as the product). Under ‘criterion 2’, the effect of edge and corner drops
would be the most harmful and influential as they tend to ‘beat up’ the outside of
the package. Under this test, such impacts would classify a package as a

‘failure'. These types of damage also cause severe crushing of the package and

6

possible damage to the contained product. In a situation were package physical
stability is required (especially in a palletized stack situation) edge and/or corner
damage would increase the Chances of damage. This would not only affect the
product, but could also greatly increase the risk of collapse of the palletized

stack and further damage to the products not directly inside the affected area.

Organizations such as the American Society for Testing and Materials
(ASTM), the lntemational Safe Transit Association (ISTA) and those involved in
the development of UNlDoT and HazMat regulations are heavily involved in the
development of testing standards for industry practice (5). It is known that these
organizations/committees have spent a great deal of time developing these

procedures and are used as a guideline. Several test methods such as ASTM

D-1596 :Test Method for Dynamic Shock Cushioning Characteristics of

Packaging Material and A_STM I_D-4169: Standard Practice for Performing of
Shipping Containers and Systems were developed for the sole purpose of

 

developing better package designs that would be able to withstand the hazards
of the distribution environment. Test procedures for hazardous materials fall
under the HAZMAT or UN Deparrnent of Transport regulations. It is in these
tests that under testing is being conducted on most packaging. Corner drop
tests in particular, should be conducted much more than the currently specified
one drop on the corner of the manufacturers joint. Edge drops are not as much
of a concern in comparison as the impact area is much larger than that of a
comer” (6). It is not known from my sources/findings whether this is a procedure
carried out by other facilities, but there are recommendations that more severe
testing should be placed on comer impacts for specific product and package

combinations.

7

Today, many companies in the US conduct tests primarily in the flat
orientation, than onto edges and corners. The test standard requires that only
the top, bottom, one long Side, one short side and one corner be tested (on the
corner of the manufacturers joint). Considering that a typical rectangular shipper
has 26 independent impact orientations to consider (12 edges, 8 comers, and
only 6 flat surfaces) it seems unusual that such a limited number of tests are
being conducted. However, there is no fixed method for testing a product I
package system on a specific orientation. AS a rule, either of these tests are
followed with modification, depending on the nature of the item being packaged
and its final destination. From this point a test can be quickly developed unique

to this application given a specific set of goals.

The basic goal of ISTA testing is to get a certification label on a box
signifying that a package has passed all aspects of ISTA laboratory testing
procedure. As a result, any damage that the package experiences in distribution
should be the fault of the carrier. Many people within the shock and vibration
testing world believe that certain companies in the industry seem to be making a
point of exploiting this fact. “Considering the minimal amount of testing that is
generally being conducted on the edges and corners of a package - It has been
found that certain companies still take advantage of the ISTA specified test
conditions. This involves designing the packaging purposely so that it passes
the test - even though in the real distribution environment it may fail. This way,
extra packaging costs can be avoided and if the package passes the ISTA and is
damaged during distribution then the carrier will suffer the consequences of

having to pay for lost shipment” (6).

8

In light of this information, it would be possible to speculate that it is a
more opportunistic and inexpensive option to underpackage a product by
‘boosting’ the structure of the test packages in certain areas expected for testing
in that particular orientation. As a result, if only these ‘boosted’ areas pass the
test and guarantee ISTA certification, then money has been saved. Any other
type of damage to the package and product not accounted for in the test will be
covered by the carrier. Based on these experiences and practices, one well
known testing laboratory has begun to adopt their own tests on a_ll of the

orientations not specified in the test - in addition to those that are specified (6).

In general, testing for most heavy containers under the UN and
Hazardous materials regulations produce failure on either the shortest or the top
side of the package (6). The first is due to the largest pressure distributed over
the smallest area (next to a corner area). The latter is because the product
inside the package may tend to leak after conducting a flat drop on the top Side -
which causes failure; not only of the bottle but also the corrugated box. The
other area of damage is at the corner of the manufacturers joint. Effective
packaging involves many other consideratons such as the shape of the
packaged object (important because it influences load transfer during impact);
the capacity of the foam to dissipate the energy (which controls rebound); and
the time for which the packaged object can tolerate the given deceleration (which

influences the choice of foam thickness and density).

There are four major criticisms with ASTM standard D-1596, firstly, that
the G levels predicted for the cushion curves are too high and secondly the tests
are conducted for perfect flat drops only. The third factor, as previously

mentioned, is that simulated tests use more stringent drop heights. The final

9

factor is that conventional cushion curves are constructed using free standing
cushions blocks, and in no way take into account the contribution of the
corrugated box. All of these factors induce erroneous results that lead to over-
specification and over-packaging. With the first example, the time interval
between drops onto the cushion are too short as only one minute of recovery
time is allowed for the cushion to regain as much of its original thickness as
possible before conducting the following drop. This short time interval between

drops does not allow the foam to recover.

1.2 Literature Review

Testing cushions for the development of cushion curves involves repeated
impacts on a particular type of cushion. This and the time factor have a
profound influence on cushion stiffness, and as a result, increases G levels.
Over a period of time this will eventually lead to a premature breakdown of the
material due to rupturing of the cells, which again results in an increase in peak
deceleration G level. In the light of these facts, the experiments in this thesis will

allow approximately 10 minutes between drops.

One way to handle edge and corner drop predictions is to equate the
situation to an ‘equivalent flat drop’. If the ‘true bearing area” were to be found,
then it would be possible to deduce edge and comer drop G levels from standard
cushion curves. If this was possible, then the results of this work could be further
adapted with the research conducted by Granthan (7) who developed a method
for predicting the shock transmission characteristics for ribbed polypropylene
cushioning material. This work was also based on the calculation of bearing
areas but looked at converting the ribbed cushion into an equivalent flat plank

cushions to determine G level using standard cushion curves.

10

Kuang-nan Taw (7) has developed an empirical model that observes the
isolated compression region of blocks of foam alone in a 45 degree edge drop.
This procedure, like that of the conventional cushion curves, does not consider
the role that corrugated board plays in deformation behavior and its effect on G
level. The work of Chen looked at predicting peak deceleration levels for ribbed
and flat EPS cushion(8). The results suggested that at low drop heights and
static stresses, the peak deceleration levels were quite similar. However, at
higher drop heights and static stresses the G levels were significantly different.
Apart from looking at drop heights and static stresses, consideration must also

be given to cushion density.

This research will develop a theoretical model for predicting G levels for
perfect edge and corner drops for various thicknesses of cushion. However, the
biggest problem in a non-flat drop situation is that cushion curves for the material
cannot be used because unlike flat drop impacts, it is difficult to identify the true
static loading. Given similar relationships of weight and drop height, the main
question that this thesis will attempt to answer is what are the differences in G
levels when comparing edge and corner impacts? What are the reasons for
these differences (or similarities)? Which drop angle and material gives the
greatest protection? Based on these G values will these experimental results
highlight the importance of each material used in the making of the package? If
we find that one material plays more of a significant role than another then it may
be possible to increase or reduce the amount of material(s) used, reduce

shipping costs and still provide sufficient protection?

11

1.3 Choosing The Right Cushioning Material

As well as defining the drop height the designer must define the product
fragility. This can be easily determined using the shock table to develop a
Damage Boundary Curve (DBC). Choosing the largest expected drop height and
type of handling environments will help select a cushion that will lower the
products critical deceleration and move the Shock experienced by the product
out of the “damage region” of the DEC (10). Choosing the highest drop height
would also lead to over-specification of material as the drop height is largely
dependent on the product weight. It is generally considered that the lighter the
package - the greater the drop height from which it is likely to be dropped from
(11).

An innapropriate choice of cushioning may give a scenario were the
cushioned product would be damaged in a situation were, normally, the
uncushioned product would not see any damage. Protection against certain
Shocks will lower the natural frequencies of the product and package, possibly
forcing them into one of the main forcing frequencies that causes resonance and
ultimately, vibration damage (12). Therefore, choosing the most economical
cushion that guarantees the most adequate protection for both shock and

vibration is important.

1.3.1 Implications of Choosing Foam Types

“The essence of any cushioning material and any type of packaging is its
ability to absorb the kinetic energy of the packaged object while keep the peak
force (acceleration or deceleration) on the packaged object below the limit which
will cause damage or injury” (13). The ‘ideal’ foam must be able to absorb
energy at constant deceleration. Cellular materials like foams, are especially

good at this as they often generate a lower peak force. In order to achieve good

12

cushion performance, it is important to understand how cellular materials
behave. The properties of cellular foam can be characterized by analysing the
properties of the chosen material itself. The second and most important property
of the material, above all else, is its relative density (or porosity). Choosing the

correct density is difficult because there are many factors involved.

Selection of the correct cell wall material must be chosen for the foam,
which is relatively simple. Choosing the foam depends on whether the
packaging material will carry repeated loading or whether it will be subjected to
severe environmental conditions such as high temperatures. For example, an
elastomeric cell wall material is needed for packaging which will be subjected to
repeated loading. If the protection is needed only once, a plastic or brittle

material is better because such cellular materials are more efficient.

Packaging sytems employing cellular materials are traditonally designed
with an experimental database, requiring a large number of impacts (14). The
“stress vs energy table” used in combination with existing cushion curves will
allow empiricism to be combined with physical modelling (13). If used properly,
the number of experiments needed in the design process can be significantly
reduced. The data will be used for both rough calculations of embodiment
design and detailed calculations, where it may be necessary to conduct a few

selected experiments.

Another Characteristic to consider is whether the material is an open or
closed cell structure and the dimensions of the corresponding mean cell
diameter (15). An open cell structure consists of a three dimensional network of

linked cells. On impact, energy is absorbed by the cushion by allowing the air to

13

move freely through and out of the cushion. In order to predict perfect edge and
corner drop G’s, a mathematical equation will be developed that will in some way
have to account for these characteristics of the material. A high degree of
consistency is required when accounting for material behavior and using an
open-cell structure may not be appropriate for predicting G levels as the degree
of confinement in the package alters the air flow and will cause unpredictability in

the results.

Based on these facts, the choice of material in this experiment is the very
popular and commercially available 2 pcf closed-cell Low Density Polyethylene
(LDPE). The primary reason for using this material is not only is it one of most
recognized foams on the market, but it is probably one of the most used for
product protection. A closed-cell foam consists of individual pockets of air or gas
trapped inside a thin unbroken plastic membrane. The choice of gases are
either carbon dioxide or nitrogen, depending on the type of manufacturing
process of the foam. The fact that closed cell foam has gas is trapped within the
cells makes it ideal for predicting material behavior as the conditions are
somewhat consistent. The fact consistency is accounted for compared to open-
cell makes closed-cell the most appropriate choice. Additional to this is that
closed-cell foam is much stiffer than open-cell and more suitable for heavier

products.

Despite this, there is also the added uncertainty in the actual amount of
contact bearing area and foam volume involved in the dynamic compression
process for both edge and corner drop situations. The process of calculating this

type of behavior is more complex than the methods used for flat drop analysis.

14

In addition to this the corrugated outer case also adds further complications to

the calculation.

Almost all man-made and natural foams such as coral, stalks, leaves and
many woods, are not isotropic (not regular) as their cells vary in shape and
length, however, the structure can be thought of as one of regular units (15).
However, no matter how anlstropic (irregular) a foam cushioning material can be
tailored in terms of refinement of the manufacturing process; few are completely
isotropic. This is because their structure is completely damaged by the initial
impact, which results in the progressive increase in the maximum deceleration
levels recorded at each successive drop. Polyethylene is a good example of this
and is one the reasons why deceleration G levels found in cushion curves are so

high.

Another important consideration when comparing foams of different
densities and types is ‘compressive creep resistance': the ability of a material to
resist progressive and permanent thickness loss over time under a static load.
As density decreases, so does creep resistance (11). Although most varieties of
resilient materials vary considerably, it is generally considered that many suffer
little from thickness loss given sufficient recovery time and fatigue resistance is
good. Thickness loss usually decreases logarithmically with time. “In practice,
thickness loss due to creep will probably be smaller than the estimate derived

from such curves” (11, 16).

Looking at Table (2), we can see the properties of many materials used in
consumer and industrial packaging applications. Looking at the main

characteristics of expanded polyethylene, we can see that the cushion factor is

15

24' DROP

figure 5
24' Drop, ist Impact Density = 2.0 PCF

 

1

cl

Decelerallon. 6's

.5

533$

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.5 1.0 1.5 2.0 2.5 3.0
Static Stress. psi
24" DROP
figure 6
24' Drop, 2-5 Impact Density = 2.0 PCF
12 m;
we \_/
'15“
.a so-
”. /
C
’2 N 2.0" ‘ °
3 so- ’
2
8 _/l/ D 2.5-
5 /
‘ \\ I! / w
- §_ 4°“
20L "‘ so-
0
0.5 1.0 1.5 2.0 2.5 3.0

 

 

 

 

 

Static Stress, psi

Set of Typical Cushion Curves for a 24 Inch Drop
First and Second Through Fifth Impacts

16

moderately good. This number refers to the efficiency of a cushioning material.
The lower the cushion factor, the higher the efficiency. Thickness recovery is
also good and given sufficient recovery time - thickness can be maintained for a
sufficient amount of time. However, foams (especially at higher densities)
cannot gain back their initial thickness. As the amount of impacts increase, this
produces a stiffer cushion which will inevitably increases G level. This becomes
more of a problem for the product the more the cushion is subjected to further

impacts.

Creep resistance is also good. The only drawback of polyethylene is the
fact that its fatigue resistance is poor (lack of structural resilience). Temperature
limits are reasonable, although below -20 degrees celsius the material can
become brittle and cause potential failure given a severe enough impact. Above
+ 60 degrees celsius; the material can become to soft and not provide sufficient
shock absorbency. In both cases, G levels under safe conditions can be
magnified significantly under these conditions. Water absorption is usually small
in closed cell materials as compared to open structures. Corrosive and mold
resistance also is not a problem in closed cell materials. The effects of dust on

polyethylene is also not a major factor.

In terms of cost, a lower density material contains less total raw plastic
resin than a higher density, therefore, it would seem logical that it would cost less
to make. However, this is not necessarily true as the manufacturing rate, cost of

blowing agent, the amount and price of the base resin all influence the cost (16).

An important factor affecting the performance of any closed cell

cushioning material is a change in the manufacturing process. This will result in

17

variations in terms of its eventual cell Size, structure, or composition of the
polymer that forms the cell walls, in which case could change the cushion’s
density and the results of the cushion curves significantly and the entire test
procedure would have to be repeated (17,15). It is therefore necessary to assign
grades of a particular foam material based on overall density, which is a function
of cell size, basis weights used and percent resin impregnation (8). Models for
these properties concern themselves with the microscopic struts and plates that
make up the cell edges and faces, and the way they respond to load, or transmit

heat, or dissipate energy (15).

1.3.2 Modelling Cushion Behavior

Several models have been developed for predicting the behavior of
polymeric cushions. “Burgess (18) derived a model based on his study of the
thermodynamic processes involved in closed cell cushions when partially trapped
air is compressed rapidly in an elastic network of interconnected membranes.
It was determined that the net effect of heat transfer is to dissipate energy
continuously over the duration of the impact. “Throne and Progelhof” (19), have
also researched the static and dynamic stress vs. strain behavior of closed cell
foams and also a similar study for dynamic loading of a cushion. In the latter
study, they determined that there is an energy balance between the maximum
potential energy available in the drop is converted into energy stored in the foam
at maximum cushion compression. At maximum compression, they determined
that the energy stored per unit volume of material is the area under the static
stress-strain curve, while the potential energy is simply the weight of the object
times the drop height. None of these methods however are immediately

applicable to edge or corner drops.

18

One study was undertaken were a mathematical model was developed for
a 45 degree edge drop in order to predict the dynamic behavior of a low density
closed-cell polyethylene foam. This was predicted by observing the stress-strain
behavior of an edge of a cushion under static compression and identifying an
isolated compression region (8). The predictions of the research were in error by
as much as 50% of the actual value, thus making it impossible to accurately

predict behavior theoretically.

Granthen (7), successfully predicted peak G levels by calculating the “true
bearing area” of cushions in a flat drop situation. Using this area to calculate the
static stress, he determined that ribbed cushions could be used with existing
cushion curves meant for flat plank cushions. In both cases the studies have not
investigated the influence of the box in either impact situation. This is an
important factor to consider as the box has an extreme affect on the package

behavior during an impact.

To date, little research has really attempted to study the prediction of G
levels for perfect edge and comer drops. ASTM D-1596 has never made any
attempt to specify tests for either edge and comer impacts or assess their
contributions in terms of G levels. No work has been conducted that
incorporates edge and corner drop G levels for use in combination with cushion
curves. No attempt has been to develop new cushion curves specifically for
these conditions. The reason for this is threefold. The first, because the ‘true’
bearing area of edge and corner impact geometry cannot be determined. This
makes prediction of peak deceleration levels difficult. The second reason is
because cushion curve data already exists for flat drop data. The final and most

accepted reason for using flat drop data is that in terms of the greatest G levels;

19

both edge and comer impacts are less severe. This is because most of the

impact goes into rotating the package which causes dissipation of energy.

1.3.3 Pilot Study

A pilot study was first conducted looking at cushion performance in flat
drop situations. Traditionally, it is assumed that the cushion area Should be
based on the dimensions of the product area. In the case of products that do
make full contact with the cushion, this is true. If the cushion does not make full
contact then how can we determine the ‘true bearing area’ involved in cushion
deformation of closed-cell Low Density Polyethylene cushions. The hypothesis
was that using four arrangements (each representative of a product base) with
the same weights and drop heights while sharing similar dimensional
relationships (but spread out), would we get the same resultant G levels for each
situation and if so, why? This is likely to be absorbing the force would be that
directly underneath the impacting object. If the values were similar, then the

distribution would not affect Peak G.

The study was split into two parts. In each part looked at two block legs
with similar relationships. The relationship was similar in that they all comprised
of a block with a fixed edge length of 9”. Each block was individual in that the
width dimension ‘x’ was divided into either one, two or three parts (x12 or xl3),
while still occupying the same dimensions. Four weights and three drop heights
were conducted on each arrangement and dropped onto a cushion measuring
9”x 9" x 2”. This gave a total of forty-eight drops. The G level data for both

studies was compared.

20

It was assumed that the area likely to be absorbing the force would be
that directly underneath the impacting object. The effect of cushion “drawdown”
in closed-cell foam suggests that contrary to belief, the area of cushion deflection
would be much greater than the area directly underneath the product base.

The results, however, showed that in some cases, there was a Significant
difference between certain arrangements, but this was minimal. Roughly 90% of
the results had G levels within i 1-4 G within each others range, proving that
there is no real correlation between the width of product base area, G level and
the corresponding area of cushion collapse. From here, it was decided that
“drawdown” was not as significant as expected, therefore eliminating the need to

do further work.

1.3.4 Relation Between Edge and Flat Drops

Technology is available that on impact will record three individual shock
pulses for the x, y, and z direction over a period of time. Individually, these
pulses are present in no particular relationship. All three waveforms can be
combined into a single Shock pulse, which assumes that the “resultant” shock,
which occurs in the vertical direction. This can be done by calculating

the’resultant’ G at every instant, using the equation:

 

G=jof+ef+0£

Using this combined ‘resultant’ G in combination with the individual x, y,
and z pulses, we can calculate the impact angles of each pulse over a given
duration. If the values on each pulse show a change over time, then it is
possible to determine that the package is rotating. Therefore, it is apparent that

a non-perfect edge or comer drop is taking place. Using this ‘vector’ relationship

21

it is possible to determine the actual G associated with an oblique shock. The
conclusion is that a vertical G on an edge or comer is always equivalent to two
(or even three) simultaneous, but much less intense G levels expected on the
sides of a package (see Figure 1). Therefore, designing for protection against
edge drops by performing only flat drops requires that shocks on two or three
faces be applied simultaneously, which is difficult to do. A shock to only one
face such as in a perfect flat drop, does not guarantee that components will
respond the same way. The downside of this is that over-packaging happens

because of this fact.

It is envisioned that if it is possible to predict peak deceleration G values
for perfect edge and corner drops, then the method should in some way have
applications for use by the packaging engineer, in a similar procedure to that
used by ASTM in the use of their cushion curves. It may be possible to predict
the response of the first impact, and collect the average G response for 2-5

drops using the same procedure.

22

 

 

 

 

Box experiences rotation

 

 

 

Figure 1. ‘Vector’ Relationship Associated With Oblique
Shocks

CHAPTER 2

EXPERIMENTAL DESIGN FOR EDGE AND CORNER DROPS

2.1 Test Setup

As mentioned in the introductory section, the original pilot study suggested
that there was no relationship between the distibution of product area, the
contact area under deformation and the corresponding G level. Further
investigation looked at situations in which the contact area involved in an impact
was smaller than that of flat drops: edge and comer drops. The behavior of foam
cushioning material and corresponding G levels in the corner drop mode has
never been conducted. Work has been done on edge drop impacts, but like
corner drops, the influence of the corrugated board and foam together has never
been studied. In both situations, it is impossible to determine the actual bearing
area. In this mode, the cushion encompasses greater and greater bearing area

as the cushion continues to deform on impact.

A specially constructed jig was built with the facilty to attach the edge and
corner block arrangements (each representative of the base edge or corner of a
product). This was then clamped to the cushion tester which would be dropped
onto a base measuring 9”x 9". The edge cushion/box system had fixed edge
lengths of 4.5 “ and 9” respectively. The choice of cushioning material used in
this experiment was the commercially available low density polyethylene foam
with a density of 2 pcf made by Dow Chemical. In Figure (2), the test setup
utilizes the basic cushion tester. In Figures (3) and (4), we can see that two jigs

were developed for both edge and corner drop tests, respectively.
23

00

2,5

 

 

 

14

 

11

15

 

12

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1] Electrical Power-in

2] Junction Box

3] Hoist

4] Top Plate

5] Hoist microprocessor

6] Chain Container

7] Guide Rods (3)

8] Release Member

9] Lifting Latch

10] Brake/Bearing Housings (2)
11] Test Platen

12] Optional Ballast Weights

13] Brake Trigger Switch

14] Instrumentation Trigger Switch
15] Accelerometer

16] Test Cushion (area not used)
17] Baseplate

18] Seismic Mass

19] Piezoelectric Charge Amplifier
20] Test Profiles (edge and corner)

21] Test Partner Software

 

 

 

 

 

Figure 2. Cushion Tester with Test Apparatus (not to scale)

 

25

4.5 inch Edge Length with 2 inch Cushion Thickness.

 

 

P‘\\\\ I

 

 

 

 

 

 

 

 

 

_/ , -
‘ -- ,
A A /
Foamm
\\ 7/
/ Corrugated Board A

 

 

 

Figure 3. Edge Drop Jig Using Two Edge Lengths

 

26

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4. Corner Drop Jig With Undefined Impact Geometry

27

This would hold the box/cushion combination in the correct orientation for these
drops. The experiment looked at two conditions for both perfect edge and corner

drops. A third condition was be studied for edge drops only.

2.1.1 Test Condition 1:

This was the main area of investigation for the thesis and involved
dropping the jig, with cushions measuring 2” and 3” in thickness and covered
with C-flute corrugated to simulate in-the-box performance. The system was
dropped from two heights of 24” and 36” using two weights at approximately 20.4
and 40.4 lbs (for 2”cushion) and 20.6 and 40.6 lbs for (3” cushion). This was
done for both edge and comer drops. The length of the edge in the edge drop
tests was 9” in order to fit inside the cushion tester. Five repeated drops were
conducted for each phase. This was considered a more representative model of
what happens to a package system over a period of time and repitition as
specified in ASTM D-1596: Es; Method for Dynamic Shock Cushioning
Characteristics of Packaging Material (20). Each box had the same weight, but
was dropped from the five heights repeatedly. The extra weight of the three inch

cushion had no significant affect on the peak deceleration values.

The reason for using the aforementioned weights is that based on the
values in Table (1) we can see that the types of drops on the sides or corners of
a package usually occur for packages weighing between 20 and 150 pounds.
Weights of around 20 and 40 pounds were chosen because they are known to
be dropped from greater heights. The drop heights chosen did not coincide with
the values of greatest box dimension as there was no dimension in either

condition that was near to these values. For this reason it was decided that two

28

randomly picked drop heights of 24 and 36 inches were more representative for

the lighter packages.

2.1.2 Test Condition 2:

This is a second experiment that involved dropping the jig, cushion and
box system using the same two cushion thicknesses of 2 and 3 inches. The
same two drop heights and weights were also used in this study (dependent on
whether an edge or corner impact). The variation here is that five incremental
drop heights of 18, 24, 30, 36 and 42 inches will be conducted for each phase.
This would possibly show some unusual findings in terms of shock pulse shape
and their associated G levels. Although this type of test is not conducted in any
of the test standards, it was useful to study the behaviour of the product, cushion
and box system over greater drop heights. This is possibly a more realistic
situation compared to that of the conditions specified in ASTM D-4169: Standard
Practice for Performing of Shipping Containers and Sfitems. This test also
covers most of the drop heights typically encountered in the distribution
environment (see Table 1). However, this cannot be proven. It was decided that
one box should be used for each weight and dropped from the five consecutively

increasing drop height measurements.

2.1.3 Test Condition 3:

In the case of the edge drop tests only, a third test was done which
involved dropping the jig, cushion (2 inch thickness only) and box system, but
this time using an edge length of 4.5 inches (see Figure 3), and again dropped
from the same two drop heights and two weights (due to the size of the system).

As in condition 1- five repeated drops were conducted for each phase.

29

The reason for this was to see if the model would predict theoretical peak
deceleration levels for varying edge lengths. This would provide useful
information for the packaging designer when predicting G values for prototype

package development.

Fewer drops were conducted for both conditions compared to the amount
in the flat drop phase. This is because, in the case of corner impacts - too much
weight would possibly damage the accelerometer on the cushion tester.

Relationships can also be derived with a minimal amount of information.

2.2 Edge Drop Test

The first experiment looked at establishing a correlation between
predicted and actual G’s for perfect edge drops. A perfect edge drop is where
the box sides make 45 degree angles with the ground. The study will attempt to
successfully predict G levels. The variables are the cushion thickness, weight,
drop height and edge length. Again, the drops would be conducted using the
experimental setup in Figure (3). The test involved measuring the peak G levels
associated with dropping the jig, fabricated cushion (LDPE Arcel 512) and a
corrugated outer box. A total of four boxes was used. The shock pulses
generated from the cushion tester were captured after filtered over a series of
drop height and weight conditions for the jig, cushion and box using the two
cushion thicknesses of 2 and 3 inches. As already mentioned, edge lengths of 9

inches and 4.5 inches were used to test varying edge lengths.

2.3 Corner Drop Test
The same procedure as used for the edge drops was used for the corner

drops in terms of weights and drop heights over both drop sequences. Edge

30

length and material behavior are no longer variables but the drop height and
weight will vary. The same method of capturing the shock pulses generated
from the cushion tester and filtered, for each of the conditions of jig, cushion and
corrugated box using the same two cushion thicknesses (2 and 3 inches) was
used.

Shock pulses were collected both before filtering and after filtering for
both edge and comer drops. Through dropping each test box repeatedly from
the same height, it was expected that the shape of the shock pulse (at specific
points in the sequence), would show different contribution of both materials to
the overall shock absorbtion during the entire duration of the test. The box for
example is likely to absorb most of the impact energy in the first drop because it
is fresh. Over time, the box gradually gets beat up with repeated impact and so
the cushion is likely to to take over absorbing most of the impact energy as drops
go on. At the same point there may be equal contributions from both materials.
At what point and time in the sequence of tests this will happen will be difficult to
determine. And since drop height and weight are likely to affect the relative
contributions it would not be possible to assume whether the box or the cushion

in particular dominates another in an impact situation.

ASTM 01596-91: Test Method for Dynamic Shock Cushioning
Characteristics of Packaging Material, will be used as a guideline in this

experiment. Some modification of the established ASTM D -1596 test procedure
was necessary, but in general, testing was conducted similar to the standard in
that five pieces of data were obtained. The theoretical model would then predict
peak G values for the first and second through fifth drops. If the theoretical

predictions are consistent and valid, then it may be possible to develop this data

31

in the form of a series of cushion curves like that seen for flat plank cushion

curves (see Figure 5), using the appropriate curve fitting/graphing software.

2.4 Equipment ,

The Lansmont Corporation Model 23 cushion tester with a flat dropping
platen head (see Figure 2) was used. Weights were added to the dropping
platen at different intervals. A Dytran piezolectric accelerometer having a
sensitivity of 10mV/g was mounted onto a free falling platen. A Dytran Model
4110 AC piezotron charge amplifier was also used to magnify the accelerometer

output. Hardware filtering frequency was approximately 5000Hz.

A Lansmont Corporation Test Partner Version #2 data acquisition
software system was used to record shock pulse waveforms from the
accelerometer mounted on the cushion testers’s platen as it impacted the base.
The weighted platen was instrumented with a piezoelectric accelerometer and
linked to Test Partner software. The complete history of the shock pulse was

recorded and later filtered and later analyzed (discussed later).

The waveforms were filtered at some specific frequency in order to
remove the high frequency components associated with the ringing of the test
fixture becoming superimposed onto the underlying shock pulses. A trigger
level of 20 6’5 has been specified in order to prevent small accelerations not
originating from the actual impact from being recorded. Lubricant was applied in
order to minimize frictional forces between the guide rods and the falling platen

during testing. Unfortunately, these forces cannot be completely eliminated.

24" DROP

figure 5
24' Drop, 1st Impact

32

Density = 2.0 PCF

 

1

Deceleration. G‘s

0.5

24' DROP

figure 6
24" Drop, 2-5 Impact

1.0 1.5
Static Stress. psi

 

2.5

2.0

Density = 2.0 PCF

 

120‘

 

T1

1.0”

/

 

\./

1 .5"
r

 

i

f

2.0”

 

Oeceleration. 6’3
0:
o

\

 

V

 

/
s:

 

 

 

 

 

 

 

 

 

 

0.5

1.0 1.5
Static Stress, psi

, 2.5"
11/ 10"
/
r; A— 4.0'
5.0”
2.0 2.5 3.0

Figure 5. Set of Typical Cushion Curves for a 24 Inch Drop
First and Second Through Fifth Impacts

CHAPTER 3

RESULTS AND THEORETICAL DEVELOPMENT

3.1 Results

This research combined physical testing and the development of a
mathematical model that attempted to accurately predict peak deceleration levels
in a dynamic situation. The first phase of this thesis was to develop a theoretical
model that identifies the ‘key’ variables used to predict G levels for impacts that
have non-defined dimensions absorbing most of the impact (edge and corner
drops). In a corner drop, the impact geometry is not is not really known. It is
assumed that if a 2 or 3 inch thickness cushion is specified, this does not
necessarily mean that the whole thickness is absorbing the shock. A more
realistic assumption is that only a portion of the corner edge is contributing to
shock absorbtion. In Tables (3—16), respectively, we can see the results for all
the actual peak G’s obtained for edge and corner drop tests over all three

conditions. The “theoretical” G’s in these tables were obtained in the following.

3.2 Theoretical Development
3.2.1 Perfect Edge Drop Situation

In a perfect edge drop, there are four variables which control the G level in
an impact. They are drop height, package weight, edge length and cushion
thickness. Of course, the type of foam and corrugated board making up the box
also play significant roles, but these are assumed fixed in the experiment. This
then means that the results of the curve fit will be valid only for this foam and box

arrangement. There is a good reason to believe that the fit will be reasonably
33

 

 

    

 

 

 

 

 

 

 

 

 

3.2.... . on use . a.
on 9an
BELOW.
asses s as ., s. . .
552303.”... .......8..o. H.._............,..~e.e..,r. . an? 8. use ._.o.
mm 033
0:30”.
..,...,.msu§ss; ,. 0......“ 3o .-..._._.mo..e~,_ em 92 a...
a S 3v av
we:_m> «52:00.90. cocototfi 0 O 3050:: an;
ascecoexm s 8:98,: 532 390185 £9.25 82.26

 

mono... m6 u 598.. eacm xom
«02:5 «3 . 0005.02... c0330 ..~ .8 3:53. no.5 ouum so” use sew .n 033

 

35

 

 

800 ES 20. we...” . . 5.8 on

    

 

 

     

  

 

     

 

 

 

 

 

 

9.65.050 vmd- Nvém vwhm om
mcficEmSO 5m «v.3 «v.8 mm
0000.. vmde «v.3. no.5 mm
«WWW .s . .w.
9.6: .330 5d- mmdm . nm on
0000 b0> 3;- 3.3 mean on
mEucflwSO mud- mm.mm ow.mm mm om came
own . 09mm . 2.5me .
mm o- nwdm . .
059.9050 NT Rdm
800 35> 58
3.8
_ omen .
- .l 9.65.98 8.5
5.? u N 962.9930 omen mmdm em m? an same
- Bib? 88 8.; . sea
$0.0: H m ., ..1.:......._ s i... 1 a; .. .1. .
wfimnE. w
I w0:_m> E0Eom.m< 00:29.5 0 2.. 6 $0205. Amos
_m_Eoconxw £0 6020.005 omega. .033. £90: no.0 £903 n coEmao

 

 

 

 

 

 

 

 

 

mono... m6 n 59.5.. ouum xom
Beans: m.u . 305.02... 520:0 .24. .8 8.30m no.0 caum son nausea .v 2am...

36

 

No 0.00...
9 .301

 

 

 

 

 

 

 

 

.. as...

.. Lush;

 

 

 

 

MNdN u N
med n a
mud. n m

 

   

 

 

   

 

m020> E0E00.m< 00:00....0 $050.5
_0_.c0conxm so 620.005.? .030... £90... no.0

 

5.530

 

 

 

 

 

 

 

 

00:0:— .0 n 59.0.. 03m .80
«owns: «0.. . 0005.02... 520:0 an 2.0 ..~ .2 8.500”. no.0 03m sea .0 030...

37

 

 

 

 

 

 

   

 

 

 

 

 

 

 

080 F5 on. 0 E8
080 5> we 0 E00
0000 No. we $.on
. vndm
5.00
. 5.00
.080 5> 2.9 2.8
. tuna
050w
whdw
wndm
,. $.00
96:90.30 nmdm
1l- 1|-II 0529030 new New
00.0.. n 0520.050 wvd magma
mod n 0000 . .
N001 n
m020> .:0E00.m< 00:0.0t_0 o
_0_.:0:onxm. as 6030.00,...

 

 

9009:.
00
000.03.

 

09mm
mmém
nmdm
mime

mm. 5
0060
v0.00
VNdN

 

 

 

 

 

 

 

.mococ:

£90: 85

 

as.

3925

 

NM 0.00.:
o. .80”.

 

_.m 050...
o. 0.0m

..,. MN

 

8.20:0

 

02.0... .0 n 59.0.. euum x00

88...... 3 . 0852.: 520:0 .3. .25 an .2 0.50am :2: seem =3 .0 e3...—

38

 

 

 

 

 

 

 

 

 

   

 

o. 00.0”.

 

        

vm 0.00...

  

 

 

08003. m. -
5505030... m. .. firwwrwxflfi _, .....
.
7
3.5 "N 8 2an
mud up . r .f . SEEM .
8.0. u m , 800:0... 0....» ....._u..._,.,,,
N.
w020> E0E000W< 00c0._0t_o O 0 $0205.
_0_E0coaxm axe 6030.002... 630.. .290: 005 £902. 8.530

 

 

 

 

 

 

 

 

 

 

02.0... .0 n 50:0... gum xom
.0008. «up . «005.03.. .3225 an 0:0 ..N .8 8.30m. 00.0 000m :00 .n 030...

39

 

 

800 00> E... 8.8 . 9.00 00 0.8

     

      

vaN 00 mdv
vmdm 00 wdv
..N.0N 00 odv
0N0 00 mdN
00.00 00 mdN
undo 00 odN «0 030%
00.00 00 wdw . 2.0.01

 

 

 

  

 

 

 

 

00.00 00
5.00 00
5.00 00
00.3.. . on
. . dev 00
«in 00
fl v .. .0. 00.00 00 V00 00 030..
. mm... $.00 . on . .- v.8 2......
.3- . . .0... . .0. _
208.... N. a a.
0020> E0E00.0< 00:99.5 0 0.0 0 60:05. .00..
_0_.c0coaxm o\o 6050.005... 000.22 _030< 20.01090 £0.03 8.20:0

 

 

 

 

 

 

 

 

 

 

 

 

00:05 .0 u 50:3 000m .80
0000...... 0.0 . 0005.02... :02030 :0 0:0 =0 .8 3.300”. 00.0 000m :00 .0 030...

40

 

 

 

 

 

 

 

 

00.0N u 00 0.00..
and H 0. 090$
00.0- u
.. 00.30%. .. 038.0906? -0000001051 - 1400,0000... -. ..

_0_.:0:oaxm o0 00:050.... _030< 2.0.0... 005 E0_0>> :oEwso

 

 

 

 

     
      

 

 

00 0.000.
o. .000”.

 

 

 

0000:... «0.. . 0005.02... :02000

00:05 .0 n 50:04 000m xom
:0 0:0 an .8 830$. 00.5 000m :2. .0 0.00...

41

 

 

 

 

 

_ 00.0 "N
. 00.0 n0
W8- .0

 

0.90.3050
vooO b0>
0000 00>

 

 

 

T 0020>

 

F0_E0coaxw

E08099.

 

 

080.0005
.0

 

 

 

 

0-0
000.03..

 

.8390ch

 

 

0-0
000.0><

 

   

 

 

 

mm 0.00...
o. 0000”.

  

 

 

 

00050:;
290: 85

 

 

 

£902.

 

:oEwao

 

00:05 .0 n 50:0... 000m 080 . 8000:... 0.0
. 035.05 520:0 =0 0:0 an 3. 33000 35 0000 :00 as. .00 .50 can .2 0300

42

 

 

 

 

 

 

 

    

 

 

 

  

 

 

8.8 "N 3 $5
8.0 up _ ,. V . ; - .,
5.0- u m .me , infigfiu H,
m
82% 8:995 o o .885. .8:
ascmcoaxm 0.. @3285 639°. 2%: 85 £925 8280

 

 

 

 

 

 

 

 

 

«own—:— «or . 0005.25. coin—.0 an 95 =0 .3 3:33. 320 .5500 ..VN .3 gash

43

 

0:65.050

 

00.0 n N

 

00.0 n n

 

 

00.0- n m

 

$5.050

.800
000
000

08.0 50>
080 5>
059.9050
080
cam
000
080

080 50>
0000
080 50>
, .0...

  
 

     
     

 

mmem

                           

£086.

 

 

mo:_m>
.mzcmcoaxm

 

E08099.

 

00:92.5
o\o

 

60.00.59:

0r.v0
00.00
#090
00.0r

    

    

.3.

 

    

.fifigg

mm 2an
o. 00.0”.

..

 

0-0
00903.

 

 

 

.855
£90: 85

 

.3;
£925

 

 

c2530

 

3009:. 0-0 - 3050...... 002030

:0 .25 .0 .8 3.33. 85 .280 .3 .3 22¢

 

 

 

 

 

 

 

      

      

0v 0.00...
2 a. my.

0.... .

 

 

 

 

 

 

 

 

 

 

 

 

 

:0 u N 8 033
00.0 n n . . 2 00.0m
00.0- n m 000...... . 0...... a
.3
m02m> E06099. 00:90.05 0 0 .00;
@8890 e. 88.85 832. £90: 85 £925 8280
003E. .2 - 805.02... 820:0 =0 0:0 =0 .8 3508.. 85 8500 :00 .9 280

 

45

 

O

0......”

 

00.0
0—20-

 

 

II
1000

800

800

800
80

.0000
0000 50>
.080
000

00m
080
080
0530.050

080

080

080
05.53050

 

0020>

 

 

.mzc0coaxw

E08006...

 

09.0.9.5
0..

 

60:050.:

 

0.0006.
0.0
000.02

 

 

 

V00

V00
V00
V00

  

o. 0.0...

 

0v 0.00..-

  

rm“...

....0 _..

00 0.00..-
o. 0.00.
.=.N

 

 

.039».

 

.885.
£90: 85

 

.8.
£902.

 

8.50:0

 

 

00009:. 0.0 - 0005.03... 002030

:0 0:0 a

N .2 3380 85 .830 .8 .2 28..

46

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

800 98... . .0
0.. 05$
... . . :1. . 3 . . . Ow hOme
.82.... 20> 8.0 mm: r. .w mm...» .08 .0
.
88 .. .8... $2. .0... we. a
Dwain. m- s. 2an
IE. . Wm - . , o. .08”.
8.0- . u m 88 00> .00. _ . 00$ . 0...... _. . , ,2 . v.00 .0
.N. a. .0. i -
00205 E0E005< 00:00.50 0 o .0055. .00..
5.8890 0.. 88.8,: 639.. £90: 85 2902. c2580 _.
.08.... .2 - 805.02... 820:0 :0 0:0 =0 .8 3.38”. 85 .0500 :2 .3 03....

47

 

0520.050

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  
 

 

 

   

 

 

 

 

 

 

 

 

 

0v 0.00...
800 9.0.8.,
, 00.00 . . ..
0520.050
080 50>
8.9.0.930 8.0
x .........0...
8.9.9230 .5 . . - , .
8.9.9030 8.0
00.0 “0 8.9.9.50 8... 3.0 s. 0.an
.0. "a .80 00.0.- v.00 9.0.0”.
8... n0 . .3... r .0
gouge. 0.00.5. m.
00:_0> E0E00.m< 00:92.5 0-0 0 0-0 .0052... .00..
_0_.c0coaxm. o\o 80.03. 60.0802... 08.0.3 £901.85. 2.0.0; :oEmzo
000008.00

«005—0....- coEmzo :0 new

:0 .50 3.300”. 09.0 .0500 =0¢ 0:0 £00 .aon ....v0 6.. 0.30...

48

accurate for many types of foam for two reasons. First, most closed-cell foams
are very similar in performance as the published cushion curves show and so
changing over to a different foam should not drastically alter the results.

Second, the foam only absorbs part of the energy: the box absorbs the rest. And
since C-flute corrugated boxes make up the majority of board used, the

contribution of the box to the G level is considered fixed.
The form of the curve fit to the experimental data was taken to be:

G=Zh‘W"L°t" (1)

where: h = drop height (inches) w = weight (lbs).
L = egde length (inches) t = cushion thickness (inches)
Z = unknown constant

a, b, c, d are unknown exponents

The choice of this form over any other fit such as a linear one (G = a + bh
+ cW + dL + et ) for example, is motivated by the prediction for G using the linear

spring mass model (21) in Figure (6),

G = 2hEA (2)
Wt

where h is the drop height, W is the weight, A is the impact area, and t is the
cushion thickness. The modulus of elasticity (E), depends on the type of foam,
and so embodies the unknown Z in the fitted equation (1). If this linear model

were to fit edge drop impacts, then the powers “a” and “c” in equation (1) would

be 1/2. The powers of “b” and “d” would be -112.

49

e =/ th =/ 2hEA ~Z(h)b(L)°(t)"(W)'(edge)
w Wt ~z (h)'° a)“ my (corner)

 

Weight

 

 

 

 

' ' " ' ' ”A
Spring k = _E_A
Constant t
00
Drop
Height

0‘)

/////////////7//i

Figure 6. Model of a Linear Spring Mass System

 

 

 

 

 

50

The choice of fit in equation (1) is also motivated by the fact that if either
W or h were zero, then G should be zero ( and is in this formula). Only a
‘product form’ where the variables are multiplied by each other accomplishes
this; a linear fit would still give a non-zero G even when h = 0 and therefore
makes no sense. Based on the linear model and general observations on the
cushion curves, it is expected that the powers “a” and “c” will be positive and less

than one. Only the fit to the experimental data will confirm this.

3.2.2 Perfect Corner Drop Situation

In a perfect corner drop, we have a similar situation except there are only
three variables which control the G level in an impact. They are drop height,
package weight and cushion thickness. Remember, there are three radiating
edges with no defined impact geometry, therefore, the same equation can be
used without the “edge length” variable. The type of foam and corrugated box
are again fixed, however, as mentioned previously the curve fit for this particular
setup may also applicable for other varieties of cushioning materials used in
combination with C-flute. The form of the curve fit in this situation looks slightly
different in that the edge length is not included in the formula. The corner drop

orientation does not have a defined edge length. Therefore:

G=Zh"Wbt° (3)

where: h = drop height W = weight
t = cushion thickness
Z = unknown constant

a, b, c, are unknown exponents.

51

3.3 Limitations of the Theoretical Model

For the remainder of the experiment, equations for both edge and corner
drops will only be calculated for test conditions [1] and [3]. Using condition [2] for
drop heights from 18”, 24”, 30”, 36” and 42" will not be studied in any further
detail. The fact that this is not the primary objective of the thesis adds to the fact
that these results should not be included in the following analysis, as they might
affect the outcome of the generated curve fitting information. Even if the
incremental results fitted the spring mass model it would not be possible to
construct a cushion curve in this form. The advantage of using these cushion
curves is that hypothetical weights can be used for a given thickness and drop

height.

3.4 Limitations of Curve Fit Software

Due to the limitations of commercial curve fitting software used, it was not
possible to use the power fit equation. The alternative method was to use a
polynomial fit equation. A slightly different method was used in which the
logarithm was taken for each of the five variables. This gave two sets of “Z” and
power values for “a” “b” “c” and “d” constants for both edges and corners
dropped from 24 and 36 inches only. The polynomial equation is similar to the
power fit equation except we take the logarithms of both equations (1) and (3).

This gives us the the appropriate logarithmic equation in following form:

In (G) = ln(Z)+ a ln(W) + b ln(h) + c ln(L) + d |n(t) (4)

This generates our power values. Two equations for the first and second
through fifth impacts, respectively for both edge and corner drop with the

variables having their corresponding power coefficients. Below, we can see that

52

calculating the power coefficient values (highlighted) for each of the variables
based on equations 2 and 3 we find that each value lies in the region of :l:1/2 (
:t 0.5) suggesting that the linear model does in fact apply to the behavior of edge

and corner impacts also.

Final Edge Drop Equations

G1 = 29.16 h +26 L +.16 t-.“ W -.32 (5)
62's = 27.33 h 4530 L £13 t 55‘ w -.19
Final Corner Drop Equations
G1 = 45.8 h +.14 t -.60 W -.30
62.5 = 24.5 h +.31 t-.5ii W-.12 (6)

The predicitions from the “least squares fit “ model were then entered into
“Mathematica“® curve fitting software in order to generate an equivalent set of
values for the “Z” and the power values for “a” “b" “c” and “d” constants (see
above). Using these values, the software will also calculate the corresponding
theoretical G values for both edge and corner drop conditions. The above
results show that the negative and positive values assigned to each variable

followed the prediction stated earlier in this chapter.

In Tables (3-16), we can see that in the “agreement” section of the
spreadsheet, there are several comments that precede the percent difference

value. Out of four possible comments anything over 16% error is considered

53

“bad”. A value of around 10% error (ASTM maximum percent error) is
considered “good”. Anything below 10% or 5% are considered “very good” or “
outstanding”, respectively. As already mentioned, the values of the experimental

G are also in error by a certain percentage.

The results of the predicted and actual G’s for both edge and comer drops
are graphed in the form of cushion curves, using Microsoft Excel® software. The
properties of each graph for both edge (see Appendix A (Figures 7-10)) and
comer drops (see Appendix B (Figures 11-14)), will be similar to the cushion
curves used for flat plank cushions in that the ‘Y’ axes will be the G level. The ‘X’
axis for edge drop cushion curves will be in the form of Lc I W”. This was to
make the curves analagous to standard cushion curves where static stress is
used (weight/area) for flat drops. The fact that the comer area has no defined
edge length, makes it impossible to calculate an equivalent static stress.
Therefore, the ‘X’ axis will represent the product “weight", as this is the only

remaining factor that G level is dependent upon.

A total of 25 “theoretical” data points were used to construct each cushion
curve. Using the “new” equations - G levels were predicted for hypothetical
product weights ranging from 2 - 50 pounds. This was applied to each condition
of drop height, thickness and cushion length. This method was used due to the
insufficient amount of experimental data necessary to construct each curve. The
“experimental” datapoints for each condition were then superimposed over the
“theoretical” cushion curves. In Tables 17-24 (edge drops) and 25-28 (corner
drops), we can see that the percent difference between the predicted

“theoretical” and “actual” values were within 2-10% for most conditions.

54

Table 17. Percent Differences Between Curve Fit and Experimental
G Levels - 24" Edge Drop for 2" and 3" Cushion - 9" Edge
Length - 1st Impacts

Actual

Cushion Thickness = 2" = 9 inches

24 20 .7
24 40 20.57

Cushion Thickness = 3" = 9 inches
24 20 21.52
24 17.22

 

Table 18. Percent Differences Between Curve Fit and Experimental
G Levels - 24" Edge Drop for 2" Cushion
Comparing 4.5" and 9" Edge Length -1stlmpacts

Curve Fitted

G

Cushion Thickness = 2" = 4.5 inches

24 . 23.01
. 1 .4

Cushion Thickness = 2" = 9 inches
24 20 25.7
40 20.57

 

55

Table 19. Percent Difference Between Curve Fit and Experimental G Levels
36" Edge Drop for 2" and 3" Cushion - 9" Edge Length
1st Impacts

Cushion Thickness = 2" = 9 inches

20.4 43
40 40.4 22.75

Cushion Thickness = 3" = 9 inches
20

 

1

Table 20. Percent Difference Between Curve Fit and Experimental G Levels
36" Edge Drop for 2" Cushion - Comparing 4.5" and 9" Edge
Length - 1st Impacts

Cushion Thickness = 2" = 4.5 inches

39.8

Cushion Thickness = 2" = 9 inches
20 20.4
40.4 75

 

56

 

88%. 3
59.3 euum =0 . .8226 .a .2; =0 .8 85 83 :3.
205... w 3:08:0an use 0E 02:0 :eezaom noose-5:5 «520m ._.0 sink

57

 

38...... 90
59:: BE .a 2:. =3. antennae . .5226 ..0 .8 35 seem .30
«.05.. G .5ceEtenxm 0.5 “E 9.50 .3058 eeocfieEo «520n— .NN saw...

58

 

38...... m0
593 anew .a - .6326 en .25 =0 .5 85 omum =8
«.05.. 0 .ScoEtonxm use «E 9:30 cooiom 02.2050 0:35.“. .3 03a...

59

 

88...... 3
59.3 oaem .a use :3. 2:388 - .3226 ..0 .8 85 anew :8
£05.. 0 .8ceEtenxm tea 0.... 02:0 season woo—.28....”— uceEem .3. 23¢...

60

Table 25. Percent Difference Between Curve Fit and Experimental G Levels
24" Corner Drop for 2" and 3" Cushion - No Edge Length
1st Impacts

Cushion Thickness = 2"

4O

Cushion Thickness = 3"

 

40

Table 26. Percent Difference Between Curve Fit and Experimental G Levels
36" Corner Drop for 2" Cushion only - No Edge Length
1st Impacts

Thickness = 2"
20
4O

Thickness = 2"

 

40

61

 

8038.
226 52.3 ouum oz - 5.5.6 =0 .28 ..0 .8 e89 358 WM
.. o _5=eE..oexm .2... E 856 535m 8.3.885 .528. .00 28¢

62

 

88...... 90
59.3 oaum oz - .3296 =0 85 =0 .8 e80 388 =8
e.o>e.. 0 EcoEteaxm new 0.". also cesium 3:255 «neocon— du use...

CHAPTER 4

DISCUSSION I CONCLUSIONS

4.1 Discussion

The results confirmed a very close correlation between the predicted and
actual G levels obtained in tests using a ‘power law’ equation. The model
predicted values in the same way that the cushion curves are determined for flat
planks in ASTM D-1596, in that the first drop will be predicted and the average of
the second through fifth drops. The prediction of the theoretical G was very
important in that if any of the results were in error of more than 10% of the actual
G value, then the method of prediction was incorrect and a different approach
should be taken. Fortunately, most of the theoretical predictions were within i 2-
10% of the actual G value given the parameters of weight and drop height and
edge length (the latter applies to edge drops only). It should be noted that the
actual values generated (experimentally) through Testpartner®will possibly be

subject to error.

Although this theory is based on an ASTM test standard, I am unsure
about using this approach as there can (and generally is) alot of deviation
between these last four drops. Looking at the results in Table 4, 6, 8, 10, 12, 14
and 16, we can see that there is significant difference between the second and
the third through fifth drops. The method of predicting theoretical G values for
the second through fifth drops relies on an algorithm that produces a “Z" constant
along with the appropriate number of power values. Having such a large

deviation in the second drop produces a fixed theoretical G that is not
63

64

representative of what is actually happening during the last four impacts. A
modification of the spreadsheet would allow calculation of the first, second and
third through fifth values. This would be a more accuarate and representative

model.

4.2 The Affects of Weight and Cushion Thickness

The weight had an extreme influence on the G levels. The lighter weight
conditions produced higher G levels. This is because there was not enough
weight behind the impact to crush the box easily. This was more prevalent in
the edge drop tests due to the resistance of the edge to deformation. In general,
this results in a much higher peak stress and corresponding G level. The
heavier box conditions, however, where generating lower G levels as the weight
behind the cushion and box was sufficient enough to easily crumble the test

boxes on impact, producing lower G’s.

Comparing edge and corner drop conditions we also have to consider the
contact area of each box/cushion system as both have different available
compression boundaries. The comer drop, for example, has three radiating
edges which provides a greater amount of material available for compression
(especially with the 3 inch cushion). This is reflected in the results which show

that corner impacts do absorb a greater amount of energy than an edge.

Although the G levels were not significantly different, in general, It was
found that the three inch foam, being a softer and more flexible cushion,
absorbed more energy than the two inch foam. Given the identical conditions of
weight and drop height, the thicker foam undergoes smaller cell compression.

As pointed out by Kuang (8), the thicker foam allows more cells to move and

65

absorb the impact energy. The 3” foam accomodates a larger dynamic
deflection compared to the 2” cushion under the same conditions because of the
larger available compression boundary radius. Looking at equation 5 and 6, we
see that the power values for thickness for both edge and comer drops are

consistently high and do not change a great over time.

4.3 Edge Drop Results

For all three conditions - the first drop predictions were generally
excellent. Out of the sixteen first drop predictions - only two showed 21.5 and
26.94% error. A third value showed 16.55% error, while the remainder showed
less than 14% error. The 10% error values are in accordance with ASTM as the
maximum acceptable limit for error precision. For the second through fifth data
the results were in similar, if not better agreement with the first impacts. Looking
at the drops from 24 and 36 inches for the 4.5 and 9 inch edge lengths 29 out of
48 were much less than 10% in error. The remaining 16 drops for the 18-42 inch
incremental drops had values that were very inconsistent. The reliability of the
theoretical model confirmed that theoretical deceleration levels can be calculated
for a varying edge lengths. Two edge lengths of 9” and 4.5” were also
compared. The results for the 4.5 inch edge length came out better than
expected as all of the first impact values were less than 1% ; thirteen of the
sixteen values for the 2-5 impacts were less than :t 8% of the experimental G.
The fact that many of the high percent error values lay slightly above the :t 10%
range is not a major concern as the actual results are also subject to some

degree of error.

The theoretical comparisons for the 4.5 vs 9 inch edge lengths were very

good and showed that reducing the edge length produces lower G levels. This

66

would make sense as conventional wisdom suggests that the longer the a longer
edge length the more material there is - which makes the cushion alot stiffer. As
a result more impact energy can be absorbed - providing a bigger resistance that
produces G levels that should be significantly higher too. However, looking at
power values for edge length (equations 5 and 6), we should expect to see a
power value that is very high (close to 0.5), yet the resultant value is very low.
The value still show that as edge length increases-so does G level, but the
difference is not really significant. This is because with a longer edge the
material deforms non-unifonnly. At maximum deformation, the cushion has not
deformed as much expected producing a narrower contact area. The result is a
trade -off between the stiffness aspects and the non-uniform deformation

characteristics associated with longer edge lengths.

Despite the fact that we were getting higher G values for the edge drops
compared to corner drop values, the G levels lacked consistency. This could be
attributed to the box splitting which would allow the cushion to break free from
the confines of the box. If this was the case then the cushion would be aflgwfl
to deform naturally. In both cases, the cushioned products center of gravity may
have not been centered directly over an edge or a corner of a package assuming

a perfect drop situation.

4.4 Corner Drop Results

For the corner tests only the first two conditions were tested. This is
because a comer has three radiating edges and no defined dimensions. For
both conditions - the first drop predictions were generally excellent and much
more accurate than the edge values. Out of the 12 first drop predictions - two

showed errors of 2.24% and 2.38%. One value borderlined at 15.77% error,

67

while the remainder showed less than 15% error. Again the 10% error value is
based on ASTM standard as the maximum acceptable limit for error (this will be
discussed later). For the second through fifth data on average the results were
similar in agreement with the second through fifth impacts as the edge drop test.
Out of the sixty-four values, only twelve values were in error by more than 15%,
nine values were less than 20% error and the remaining 43 values were less

than 15% in error.

Lower G’s were experienced than those in the edge drop phase. Also, the
rate at which the G levels increased over the five drop sequences, was more
progressive and consistent. This can be attributed to the fact that the impact
area was not affected by susceptable areas (such as a manufacturers joint). In
addition to this, although deformation was more contolled, it was also more
severe as neither of the three radiating edges (of non-specific dimensions) would
dominate during the entire duration of the impact which lead to a general lack of
resistance. Also, the impact area involved in a corner drop was much smaller
compared to the edge drop - making it an easier target. As the comers of the
test boxes began to soften the cushion to dominate impact absorbtion, but less

effectively than the corrugated board.

4.5 Comparison of Repeated Impacts vs Incremental Drop Tests
Comparing the theoretical model results with the 5 repeated drop
conditions (Tables 9-10 and 1516), we can see that the drops from 5 increasing
heights showed few values that were close to the experimental G. Unlike the
edge drops from repeated heights the power values for the comer impacts
suggest the behavior in this condition to be representative of a non-linear spring

mass system. The first impact values generate the correct power values, which

68

suggest a linear spring mass model system and a s a result predict nicely.
However, these numbers represent the box impacting against the ground.
Looking at both 2-5 corner and edge impacts, the numbers suggest that the
incremented drop heights are causing too much variation in the results. As a
result the logarithms calculate power values that are not representative of a
linear spring mass system. Like the drops from the repeated height; we see that
the experimental values show that as weight increases, G level reduces. This is
correct. Unfortunately, the theoretical predictions for the corner 2-5 impacts

suggest the opposite.

4.6 Behavior of Shock Pulses During Impact

Looking at Figures (15), (16-20 (Appendix C)) and Figures (21-26
(Appendix D)) we see that, in general, the filtered shock pulses for both edge
and comer drops behave in exactly the same way during the third through fifth
drop. This indicates that no matter how you drop the box, it will always show a
characteristic process of deformation. The first and second drops, however, vary
depending on the weight, drop height and thickness. The shape of the pulse is a
combination of either a very full half-sine wave or a short duration square wave
each having some amount of surrounding noise. At this point the peak G is
sometimes not clearly defined. In the case of the 3" cushion thickness -this is

more significant as more energy is being absorbed over a longer duration.

Out of the five impacts conducted on each specimen the first and second
drops (which are the most important), produced considerably less peak
deceleration G values than those obtained on the 4-5th drops by as much as

50% in some cases. The first impact was absorbed by the corrugated board

69

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R \\f
I l l _ |, l l L l l
Drop 1. Drop 2.
G level = 17.69 G level = 18.48
Drop 3. _
G level = 23.87
Drop 4. __ Drop 5.
G level = 24.37 G level = 25.06
L L
\____,_\ \’
l - _L_ l l l l

 

 

 

Figure 15. Shock Pulses For 24” Edge Drop Using 3” Cushion -1st-5th Impacts

 

70

while the second through fifth impacts were absorbed mainly by the cushion as
the box gradually softened. This showed that the box was, initially, the better
absorber of shock than its foam counterpart. The strangest result in all of the
tests was that, initially, the G level was low. By the third drop, the G level had
peaked. The fourth and fifth drop would show a gradual reduction in G - even
though the edges of the test boxes were softening and allowing the cushion to

dominate absorbtion; we were experiencing fluctuations in values.

Looking at the first impact we can see that there are many pulses that
contribute to this very full sine-wavelshort duration square wave. Generally, the
first pulse shows the impact of the box against the impacting surface. The
second peak is the impact of the exterior plane of the cushion wall against the
interior of the corrugated box. This happens naturally inside any box because
the manufacturers joint prevents full contact between the cushion and the
corrugated board. With the addition because extra folds are produced when
gluing the half-box section together, an increased thickness contribution provides
extra strength and rigidity. On impact, the box absorbed most of the impact
while the cushion repositioned itself inside the box in order to eliminate this air
space. As a result, it was found that the outer box edges absorbed the most

severe part of the impact before the cushion began to absorb shock.

The shape of the shock pulse is now alot more reminiscent of a sine
wave with either a sharp or rounded peak as both the outer box and cushion
acting in unison. In some cases the peak G is clearly defined -while in other
cases we find the opposite. As the drops progress, the shape of the pulse

becomes narrower producing a more perfect half-sine wave with a very defined

71

peak G. This is true whether we are dropping from repeated or incremental drop

heights.

By the third drop, the box is now very flexible and is contributing less to

the overall performance. As the number of drops are increased so did the G
level. Additional to this we see that the edges and corners of the test boxes
begin to soften -gradually contributing less and less to absorbing shock. It is at
this point that the cushion starts to play a dominant role in absorbing the shocks
and contribute to the overall cushioning performance. There was quite a steep
increase in the G level. There was also a very large increase in G level as this
transition progressed. You could also hear the differences in the impacts as the
tests progress. The first drop on a solid edge produces a loud “thud” sound,
whereas the third and fourth drops produce a softer sound more indicative of a

soft cushion impact.

A dramatic transition in terms of the shock pulse shape demonstrated a
shift from an initial square wave to a more concave half sine wave. If we look at
Figure (15), we can see that the shape of the shock pulse for the last two drops
indicates that the cushion is going through what is known as material
“hardening”. This is a result of repeated impacts which cause the material to
compress until it starts to act more like a solid block. This makes sense if we
pay more attention to the coeeficient values. If the coefficient value for any of
the variables is less than 1 0.5, then the cushion undergoes material “softening”,
therefore, the spring/mass system (cushion) is deforming non-linearly. The
shape of the shock pulse, resulting from this condition is also similar to that of a
square wave. If the coefficient value is very close to :l: 0.5, then the cushion is

deforming exactly like a linear spring/mass model. On the other extreme, if any

72

of the variable values are greater than 10.5 (like that of the “thickness” variable
for both edge and corner drops), then the cushion will undergo material
“hardening” and we will see the shape of the shock pulse looking like a concave

sine wave. This also produces a cushion that is deforming non-linearly.

It is not clear as to what degree the corrugated board influenced the
behavior of both edge and corner drop conditions, however, from my
observations of the experimental drops it was clearly a significant contribution.
This behaviour further reinforces my opinion that the ASTM procedure for
developing flat drop cushion curves is incorrect test method not only because it
does not represent the ‘real’ distribution environment in terms of the oncorrect of
impact, but it also does not take into account the effect of the corrugated board

box.

4.7 Conclusions

Looking at the results in Tables (3-16), we see that out of the five tests
conducted on each boxes, the first drop and second drops (the most crucial )
produced considerably less peak deceleration G values compared to the third
through fifth drop for that particular box under those certain test conditions.
When comparing the actual G values we also notice that the corrugated board
was a more effective material for absorbing the initial shock compared to the
foam cushion. This is mainly because lower G values experienced during those
crucial and most damaging first and second impacts (compared to the higher 3-
5th values absorbed by the cushion ). This is assuming that the box is subjected
to a limited number of drops. The efficiency of the board will depend on the
moisture content. However, this was not part of the experiment but it is expected

that this will play a large role in predicting G values using this theoretical model.

73

Because of the large abundance and availability of corrugated board it is
also very competitive with low density foams. It is also justifiable to say that
more consideration should be given to its performance when designing transport
packaging. It is usual for any packaging designer to place more emphasis on
selection of the right type and thickness of foam rather than consider the
contribution of the corrugated board. If more attention were given to corrugated
material in package development along with the theoretical model defined in
theis thesis - it is possible that cushioning can be reduced considerably along

with material costs, without loss of protection.

4.8 Experimental Errors
4.8.1 Test Method

The experiment and the model predict values in accordance with ASTM
D-1596 - in that the first drop and also the average of the second through fifth
drops will be predicted for each condition. Because this is a standard test and l
have experienced these fluctuations in G values - I am unsure about using this
approach as there can be (and generally is) alot of deviation between these last
four drops. Therefore, an average value should not be considered a value that is
representative of what is actually going on during the last four impacts. Despite
this fact, it is important that the results of these experiments have some
application. Applications can be found for use in the testing laboratory and in the
design of prototype packages. To correct this problem, I have adapted the
original theoretical model to account for these large deviations. The model
calculates three phases of a drop individually. This is because there are large
deviations between the first, second and third to fifth impacts (see Tables 4, 6, 8,
10, 12, 14 and 16). Therefore, the new model can account for these differences

by calculating and comparing the percent differences between all three phases.

74

The result is that the first value uses equation (4) divided by a percent error
constant. The second drop is calculated using the standard equation (4), while
the third through fifth drops use equation four multiplied by another percent error

constant. The results are far more consistent.

4.8.2 Machine Error

Dropping the jig, cushion and box showed that for the first drop, very low
G’s were reported using the Test Partner software. The trigger level was 20 6’3
and anything below this the software could not detect. Using the oscilloscope
was difficult because the reading of the shock pulses was not accurate enough
(on a personal note), also, it was not possible to print out shock pulses. A more
sensitive version of Test Partner was used ( courtesy of Lansmont Corporation),
which introduced its own set of problems. Although the machine was more
sensitive to lower G levels, the difference in platen apparatus made it difficult to
clamp the jig fixture as securely as that on the previous platen. This may have
been the reason for the certain high frequency noise superimposed onto my
shock pulses. It is one of the errors of the experiments that the Testpartner

hardware inside both machines may have varying automatic filter frequencies.

A simple explanation of this is that once an impact transmits a voltage
output through the amplifier converting it into a shock pulse; the Testpartner
hardware puts this through an automatic filtering process to eliminate extraneous
noise, before it is directed to the software for further manual filtering. It is at the
point were the pulse reaches the software were the differences in the pulses are
obviously different. From the differences in shock pulses, it would seem that the

first Test Partner had a lower automatic frequency than the latter. This means

75

that the lower the filtering frequency, the more noise is removed and the cleaner

the shock pulse.

4.8.3 Foam Fabrication

The foam material was manually constructed around both jigs for the edge
and corner test setup. Both involved sculptured fabrication and may have
possibly recieved some damage in this process. This may have also introduced

errors into the system.

4.8.4 Corrugated Board Fluting

There were many other possible errors involved in testing that may have
affected the results. A major cause of variability would be the corrugated board
itself. The box samples were made using ‘C’-flute corrugated board sheets -
many of which may have suffered from slight flute crushing or some other type of
damaged. It is not known whether the fluting has been damaged internally in
anyway, despite close inspection. The fluting will be positioned so that they will
be vertically oriented like that of an actual box. This will play a vital role in the
amount of contribution the cushion plays in an impact situation. If the fluting is in
the vertical direction, then we could see a lot more contribution from the box
because of the much stiffer nature of the board (still assuming that there is also
air space between the box and cushion surface) rather than the cushion. This
would assume that if the fluting was switched to horizontal, we would possibly
see the cushion absorb more of the impact as a result of a greater amount of
collapse from the corrugated medium because the material collapses easier in

this direction.

76

During both the edge and corner drop studies it was found that the flute
direction did play a vital role in absorbing the impact. The fluting in a regular box
is always used in the vertical direction, then we saw a lot more contribution from
the box (still assuming that there is also air space between the box and cushion
surface) rather than the cushion. In the test positon for the comer drop, the

fluting direction was switched to the horizontal plane.

On impact, the cushion absorbed more of energy as a result of easier and
a greater amount of collapse from the corrugated medium. This was further
complicated by the influence of temperature, relative humidity and resulting
moisture content. These factors were never calculated during the experiment,
however, on impact, it was obvious that certain boxes deformed and sounded
differently in comparison to other test boxes. It is not known when the cushion
begins to absorb the impact, but we can speculate that it would happen once the
box has begun to crush and deform. Despite these factors, all box edges
performed really well and stayed intact - during both conditions of repeated and

increasing drop heights.

4.8.5 Corrugated Board: Box Assembly

The construction of the boxes was important because it was a represent
one edge and one comer of a box and had to be as realistic a box construction
as possible (given the confines of the 9”x 9” available space underneath the
platen). It was obvious that the corrugated board did dominate the initial impact
absorbtion more effectively than the cushion. However, the way in which the jig
was constructed could have possibly influenced the behavior of the box slightly,
in corner impacts especially as we will have extra-toughened corners. The half -

split nature of the box meant that extra glue points were needed to keep it

77

together. This provided extra reinforcement as gluing the corners and edges

was necessary in order to keep it together.

Another source of error could have been the inconsistency in the amount
of glue used. These factors could possibly influence the stiffness behaviour of
the box slightly, but not a great deal. A contribution from a more stronger than
usual edge or corner support in box form will have an extreme affect on the
results (especially at the manufacturers joint). Their are differences in each
situation as the G values for dropping on cushions only are slightly less than
versus cushion in a box, because the cushion is allowed to compress/collapse in
free form whereas with a cushion in a box, the cushion is restrained by the grip
of the box walls and as a result tends to act a lot stiffer than if freestanding,

therefore producing higher G values.

The most severe damage caused throughout the testing was from drops
conducted on the edge. A more controlled resistance to deformation resulted
but at the sake of ‘bulging’ the remaining areas of the test box. This was due to
the two susceptable manufacturers joints some of which began to split as the
drops increased. The nature of the test warranted a box design that needed two
joints, which under certain extreme conditions would cause this to happen.
However, this was generally a bigger problem for the incremented drop
sequence. The differences in deforming naturally compared to simultaneous
deformation may have also caused fluctuations in the response of the cushion.
This only happened on certain boxes - so some type of cushion relaxation may

also have been involved.

78

Although this sequence of tests were not the main thrust of the research,
this type of damage was exhibited during the repeated drop sequences. In both
cases, if the package and product’s center of gravity was corrected maybe this

‘splitting’ could have been avoided.

4.8.6 Jig-Design

Affixing the box to the jig was difficult as we did not want the box to slip off
the jig and yet we needed the snugness associated with a cushion tightly fitted
inside a box. It was necessary to screw the box onto the jig in order to create the

impression of a tight fit inside a closed box.

It was not known how both box types would deform. It is expected that
when comparing both box/cushion systems, the edge length provides
considerably more durability during both drop sequences. This seems obvious
because the corner configuration has three radiating edges of non-specific
dimensions (when talking about impact geometry) exposed to an impact and no
one edge would dominate during an impact. The amount of deformation is not
easy to predict in both situations, but it is clear that the corner will be more

susceptable to more severe deformation.

There were many drawbacks associated with the method of attaching the
cushion and box to the jig and maintaining a ‘tight’ connection. This was difficult
as we did not want the box to slip off the jig and yet we needed the snugness
associated with a cushion tightly fitted inside a box. It was also important not to
fix the components together with a substrate that would act as a spring/mass

system, i.e. velcro, tape, and adhesive pads, etc. The most reasonable choice

79

was to screw the box onto the jig in order to create the impression of a ‘tight’ fit

inside a closed box.

On impact, the box would tend to push itself up the side of the jig despite
being screwed into position which caused the box to split. This could have
slightly influenced G level as the jig, cushion and box are supposed to act as one
body during freefall and impact rather than independent systems. but
considering that the system was not completely in position under the platen like
that of perfect flat drops - this was the most sensible method. It is possible that
the screw fixture could have destoyed the box at the point were it made contact
with the jig. This would possibly allow the cushion to loosen on its travel to the

top of the platen before the next drop test.

4.8.7 Perfect vs Non-Perfect Drops

There are many possible errors involved in testing both corrugated board
and foam together under edge and corner drop conditions. The main area of
concern is the unpredictability of the box material as this influences the behavior
of peak deceleration by deadening the effect of the impact. In a distribution
environment, this affect may be lost through rotation of the box. The fact that the
cushion tester was recreating a ‘perfect’ drop situation, prevented the realistic
compression of the edge or the corner through rotation of the package.
Therefore, the energy will not be dissipated between the edge or the corner as
well causing the G to be slightly higher in the test lab than compared to values
obtained in a non-perfect edge or corner drop situation. We can therefore
assume that producing an average value for G over several impacts would not

be a true reflection of what is actually happening.

80

With both drop conditions the impact angles may not have been exact. In
Figure (27), we see that in the case of the edge drops - any deviation from two of
the 45 degree angles would have produced a non-perfect edge drop. In the
corner drop situation - if one of the three angles were greater or less than 35.3
degrees, then this experiment would also be non-perfect. As a result both
conditions would not give representative G levels. In the case of an edge
impact, if the box is tilted down slightly, then the initial reported shock would be
absorbed by one of the end points of the edge length before the other. We now
know that the coefficient values in equations (5) and (6), suggest deformation
that is non-linear. This probably resulted in a lower than expected G value ,
however, this is not conclusive. The influence of drop angle will be critical to the
corresponding G level. However, the influence of weight shift/repositioning and
its influence on box shifting was not a major factor in this theoretical model
because it could not be entirely controlled. The major problem was that unlike
perfect flat drops, the jig was not compressing the entire surface of the material.
This meant that the box/cushion system would reposition itself slightly each time

due to the initial change in direction of shifting.

4.8.8 Filtering

Filtering refers to the elimination of certain false information from a shock
pulse. It is difficult to know how much should you filter before you lose vital
information about the original underlying pulse. The lower the filter frequency, the
less ripples in the shock pulse and the smoother it becomes, however, this can
be a drawback as you could begin to lose the important characteristics of the
shock pulse that identify the calculated values of peak G, drop height, coefficient
of restitution, average G, RMS G, and faired G’s.

81

Resultant Vector Force

 

 

45°

(--------------. ..

45°

 

 

Perfect edge drop with 2 angles at 45° to each other

Resultant Vector Force

 

 

 

 

 

 

 

Perfect corner drop with 3 angles at 353° to each other

Figure 27. Perfect Edge and Comer Drop Conditions

82

It is possible that a shock pulse can contain a large amount of noise
superimposed on a smooth underlying pulse. This phenomena is known as
“ringing” which refers to outside vibrational noise created from either the test
equipment, the test speciman or from less conspicuous sources such as loose
cable connections between the accelerometer and the coupler, electromagnetic
interference or triboelectric charging. All of which can be working alongside the
original shock pulse and superimpose themselves to produce false peaks and
sometimes an unreconizable shock pulse. This leads to incorrect values being
given in terms of peak G(affected most), duration and velocity change least (in

that order).

The question several times during the filtering process was how much
should I filter without losing vital information about the original pulse. Many say
that you should use the very minimum frequency equal to that of the original
unfiltered shock pulse, while others suggest an optimum filter frequency equal to
three or five times that value. There was no correct method except for trial and
error experimentation, however, consistency was required throughout the whole
experiment. I decided to use the accepted industry standard of five times the

original unfiltered shock pulse, as this was the most effective during trial runs.

Early on in the experiment the shock pulses produced through Test
Partner contained a large proportion of “ringing” which could have come from
either the test equipment, the test speciman or other sourecs mentioned in
chapter 2 - Alot of false peaks were apparent early on. At one point it was
necessary to change cushion testers because the machine at the school of

packaging was not sensitive enough to pick up G levels lower than 20 G’s. The

83

oscillosc0pe was used to determine peak G levels, however, the machine was

fairly primitive and tended to give results that were not as precise.

A second tester was used to finish the experiments because of this
innaccuracy - courtesy of the Lansmont Corporation tesing laboratories. This
machine was more sensitive in terms of picking up lower G levels, however a
different in setup in the platen construction would not allow as much of a secure
clamping of the jig as necessary, compared to the previous machine. This
created a small space between the platen base and the top of the wooden jig.
Unfortunately some of the shock pulses were quite noisy. On realizing this, I
bridged the space with a small piece of LDPE foam in a hope that this would
soften the impact. Despite this, alot of ringing was appearing on the shock
pulses probably because of the jig impacting the platen. Although this may have
been the case, this did not interfere with identifying the true peak G of the shock
pulse. Seeking advice from certain parties experienced in this area considered
that the noise on the shock pulses did not hinder the analysis of the impact

performance.

In general, filtering these types of shock pulses were not as easy. When
looking at a shock pulse it was evident that the second peak was more useful as
that was the point at which the cushion impacted the interior surface of the box
and was at this point a combination of both box and cushion acting as the shock
absorbers. On most of the pulses the filtering and analysis of G level, velocity
change, duration, drop height and coefficient of restitution was calculated using
the latter of the two peaks fifth drop would result in the shock pulse acting more

like a traditional half sign pulse.

34

It is possible that errors in predicting the filtering frequency may have also
contributed to errors in filtering and determination of the true peak G. ASTM
3332-93 (10) specifies the duration to be the time width that corresponds to 10%
of the peak acceleration. As a result, it was necessary to visually inspect each
shock pulse to determine actual duration rather than to rely on Testpartner. This

could also have resulted in incorrect filtering frequencies and peak G levels.

4.9 Recommendations and Future Work

Another area of investigation would be to determine a theoretical model
for predicting G levels for a tumbling package. This will be useful information
when you consider that in the real distribution environment we never see a
perfect flat, edge or corner drop. What we actually see is a combination of all
impact types. The results of this research suggest a way to predict with a good
degree of certainty the type of G levels found in perfect drop situations. With this
method of prediction in mind it would be beneficial if we could calculate

predictions for any non- perfect drop situation.

This test method could be achieved by using the Environmental Data
Recorder (EDR) and placing it inside a weighted box, and subjecting it to random
non-perfect drop situations. This way it would be possible to obtain the triaxial
shock pulses from the EDR memory and find a way of calcualting a predicting
overall G contribution from each of the three given shock pulses. It would be
feasible to estimate that you would get much lower G values than those obtained
in perfect drop situations. This is mainly because a lot of the shock is lost in the

rotation and tumbling of the box and not through the packaging material.

85

Using the same test conditions, but, varying the moisture content we can
study the amount of contribution the box now has compared to the foam. We
should see that the board has less impact on G level and the foam playing a
more important role in absorbing impact forces. I found that this will be important

in particularly for more severe environments.

APPENDICES

APPENDIX (A)

FIGURES (7-10)

EDGE DROP SHOCK PULSES
1- 5 IMPACTS

87

655898 .3. a .8 .0 c

88988.8 .3. .9 .8 .0 -

8205889 .3. .3. .8 .0 o
.8588... .3. .m .8 are:
.8885 .388 .8 .0I-i
8.8.85 .3. .3 .8 .01

.88 5533:8888 5?...

00 on ow on ON 0.. o

 

 

0..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.89... .8- 8.5.6 .n 95 ..0 85 85 8cm 5.0 .0 8:9“.

< 88.8.3

88

.3.. antimofig. £93..
8 8 8 8 cm 2 o

 

 

or

 

 

 

 

 

3.55593. .3. .o .8 a o

 

623886 .3. .m .8 .N I

 

 

3.58886 .3. .3 .8 ..N o

.8585 .3. .m .8 .m 1?

 

.8828... .3. .m .8 .~|.T

 

.8828... .3. .3. .8 .Nlol

 

 

 

 

 

 

 

 

 

 

an

889... .2 . .3330 .n .23 .u 2.5 35 3am .8 .o 2:9".

< 5.2.8.2

89

5555585 .3. a .8 .n
555598 .3. .m .8 .m
5.55.595 .3. .m... .8 .m
.95 .95 .3. .m .8 .m

.m>m .96 .3.. .m .8 ..~

.9... .95 .3. .m... .8 .N
85.5... .3. .m .8 .0 I?
85.5... .3. .m .8 .Nnuu
.855... 3...... .8 5+

0.0...

00

88.552.18.55: .55..
8 8 8 ow o.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.885. o... . 5.5.6 .5 a... .N. 85 55 83. .3. ... 2.3...

< 5233

0..

0m

90

5.555me .3. .m .8 .n
.505585 .3. .m .8 .m
555.58.... .3. .3 .8 .N
.96 .98 .3. .m .8 .n

.95 .95 .3. .m .8 .N

.96 .98 .3. .3 .8 .~
.8825... .3. .m .8 .n :0:
.855... .3. .m .8 .~+
.855... .3. .3 .8 .N +

0.0...

.2... 5955.555. 595..
mm on mm on m. 0.. m

 

 

0w

 

 

 

 

 

1......

ON

 

 

 

.14
H

 

 

J/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.585. m." - 5.5.6 ..... .5 ..~ 0.5 55 59.“... .5 .2 2:9“.

< 55......

05

APPENDIX (B)

FIGURES (11-14)

CORNER DROP SHOCK PULSES
1- 5 IMPACTS

92

85559.... .8 .n ..

5555...... .8 .N ..
.855... .8 .m l...
.855... .8 .8 IT

.2... .552.
8 9. 8 ow o. o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mv

5...... .2 . 5.5.6 .n .2... .N 85 no... 5.50 .8 .3 2:9...

m 52%

93

555585 .8 .n ..

.85....398 .0. .N o
.855... .8 .m In:
.855... .8 .mlou

00

.3.. .5925
o... 9. 8

ON

0..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MK,

 

V

 

 

 

 

 

 

 

 

 

.285. .3 . 5.5.... .n .5 .N. 0.5 no... 55.6 .8 .8. 259..

m 5882

0..

mw

0N

mm

0v

0v

om

94

m5 .96 .8 .n I
53 .96 .8 ..~ o
.35....85 .8 a -
.aEmsnoaxm .8 ..~ o
.838... .8 .n +

.8828... .8 hi?

.8... .533

 

 

 

 

 

 

 

 

 

on Co. on ON or
III]... I III-I‘ll; I
Y1 I
1+; w'IHlT/I:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3009:. o..N . 8.530 an we. :N 8:0 :05 .3500 :VN 6—. 0.52“—

m 528......

or

9

ON

mu

mm

I 0?

mv

95

o>m .96 .8 ..n I
m>m .96 .8 ..N 0
3:68:86 .8 ..n I
32655.5 .8 ..~ .
60.6.06... .8 gm 1...
.8828... .8 ..~ I?

.8: £985

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E
£1411
14.9%ij... alurflrf.
/J

 

 

 

503E. mun . coin—.0 :0 93 :N 3:0 :20 :0» .3. 8:2".

6 5.2.8.2

or

2

:1 ON

mm

mm

ow

,mv

APPENDIX (C)
FIGURES (16-20)

EDGE DROP SHOCK PULSES
1-5 IMPACTS

97

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.. M
Drop 1. Drop 2.
G level = 21.05 _ p G level = 29.58
Dmp3 _
G level = 32.69
Drop 4. _ Drop 5.
G level = 33.28 G level = 33.7
I l I I J l
l

— l—

l 4 l - 4 l I ,_l .-_-.__l-

 

 

 

 

 

 

 

 

 

Figure 16. Shock Pulses For 24“ Edge Drop Using 2" Cushion - 1st-5th Impacts

APPENDIX (C)
FIGURES (16-20)

EDGE DROP SHOCK PULSES
1-5 IMPACTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

99

 

 

 

 

 

 

 

 

\.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Drop 1.
G level = 21.19
Drop 3.
G level = 35.61
Dmp4
G level = 39.17

 

 

 

 

 

 

 

 

Dmp2
G level = 34.88
Dmp5
G level = 39.09

 

 

 

 

 

 

 

/ \

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 17. Shock Pulses For 36” Edge Drop Using 2” Cushion - 1st-5th Impacts

 

 

APPENDIX (C)
FIGURES (16-20)

EDGE DROP SHOCK PULSES
1-5 IMPACTS

101

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

If \\
__/ \‘ _,/ \ -
Drop 1. Drop 2.
G level = 19.46 A G level = 23.21

 

 

 

7
Drop 3. - / \
G level = 25.84 / \

Drop 4. Drop 5.
G level = 27.54 T .~ G level = 28.13

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l / /\

/\ /\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 18. Shock Pulses For 36” Edge Drop Using 3” Cushion - 1st-5th Impacts

APPENDIX (C)
FIGURES (16-20)

EDGE DROP SHOCK PULSES
1-5 IMPACTS

103

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\ \
I l l l ,__l__ I I l_.

Drop 1. Drop 2.

G level = 16.39 - G level = 26.65
F
r

Dmp3 -

G level = 32.42

Drop 4. _ Drop 5.

G level = 42.51 1 I l I G level = 53.61

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l _L_ l l l l

 

 

 

Figure 19. 18, 24, 30, 36, and 42” Edge Drop Using 2” Cushion - 1st-5th Impacts

APPENDIX (C)
FIGURES (16-20)

EDGE DROP SHOCK PULSES
1-5 IMPACTS

105

 

 

 

/\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\r/ I\/

I I __l__._L_ I I - _l I
Drop 1. Drop 2.
G level = 10.11 G level = 13.41
Dmp3
G level = 24.58
Dmp4 Dmp5
G level = 27.69 G level = 32.85

 

 

 

 

 

 

 

 

 

l ‘ l

l

 

l

 

 

 

 

 

 

l

l

l

 

 

l l l

 

 

Figure 20. 18, 24, 30, 36, and 42” Edge Drop Using 3” Cushion -1st-5th Impacts

 

 

APPENDIX (D)

 

107

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I I I

Drop 1. Drop 2.

G level = 17.27 G level = 25.38

Dmp3

G level = 29.45

Dmp4 Dmp5

G level = 31.18 G level = 30.78
I L I I

,- I l I I I I I I I I I I

 

 

 

 

Figure 21. Shock Pulses For 24” Corner Drop Using 2” Cushion -1st-5th

Impacts

 

 

APPENDIX (D)
FIGURES (21 ~26)

CORNER DROP SHOCK PULSES
1-5 IMPACTS

109

 

 

 

 

 

 

 

 

 

 

 

 

Drop 1.
G level = 12.84
Dmp3
G level = 21.04
Dmp4
G level = 22.25

 

l |

 

 

 

 

 

'—

 

 

 

 

Dmpz
G level = 19.92
Dmp5
G level = 24.12

 

 

 

 

 

 

 

l

 

 

l

l

l

l l l

 

 

 

Figure 22. Shock Pulses For 24” Corner Drop Using 3” Cushion - 1st-5th

Impacts

 

 

APPENDIX (D)
FIGURES (21 -26)

CORNER DROP SHOCK PULSES
1-5 IMPACTS

111

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

WN/ \
\
Drop 1. Drop 2.
G level = 16.85 G level = 28.14
Drop 3.
G level = 31.24 /
//
J
Drop 4. Drop 5.
G level = 32.00 G level = 33.62

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 23. Shock Pulses For 36” Corner Drop Using 2” Cushion - 1st-5th

Impacts

 

 

APPENDIX (D)
FIGURES (21-26)

CORNER DROP SHOCK PULSES
1-5 IMPACTS

113

 

 

 

 

 

 

 

 

 

 

 

 

 

 

/ \N- ,/ \

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Drop 1. Drop 2.
G level = 14.93 G level = 23.39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

/
Drop 3. / \
G level = 24.30 / \\
” \
./ \_
Drop 4. Drop 5.
G level = 24.66 G level = 23.55

 

 

 

 

 

 

fl) ‘ /\

 

 

 

 

 

 

 

 

 

-/ \ / \

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 24. Shock Pulses For 36” Corner Drop Using 3” Cushion -1st-5th
Impacts

APPENDIX (D)
FIGURES (21-26)

CORNER DROP SHOCK PULSES
1-5 IMPACTS

 

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I l I I I I I

Drop 1. Drop 2.

G level = 11.41 G level = 20.14

Drop 3.

G level = 28.58

Drop 4. Drop 5.

G level = 34.72 1 I G level = 44.72
I I I I I I I I

 

 

 

 

 

Figure 25. Shock Pulses For 18, 24, 30, 36, and 42” Corner Drop Using 2”
Cushion - 1st-5th Impacts

 

APPENDIX (D)
FIGURES (21-26)

CORNER DROP SHOCK PULSES
1-5 IMPACTS

 

 

 

 

I I I

 

 

 

117

 

 

 

 

 

l I I

 

 

Drop 1.

G level = 9.80
Dmp3

G level = 21.35
Dmp4

G level = 27.61

 

 

 

 

 

 

 

 

 

Dmp2
G level = 17.77
Dmp5
G level = 31.59

 

 

 

 

I

 

I

I

 

 

 

 

 

I

I

I

 

 

 

 

 

 

Figure 26. Shock Pulses For 18, 24, 30, 36, and 42 Corner Drop Using 3”

Cushion - 1st-5th Impacts

APPENDIX (E)

TABLES (29-36)

EDGE DROP EXPERIMENTAL DATA

119

 

 

 

 

 

 

 

   

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

81 88 E8 , 8 w ism 3.3 can 8 ~
. 81 . as... a 1m .. a a 81.83 . 1.3 a m
81 ,. 8 : a .81 88 was en, a
81 .. on. , _~ in _ «EN 0.8 .vu N
.. .101 8.9, 93 , . 8. .N. 81... . and. ad» .8 N
.81 85 . 6.1. , 8 u .81 8.8 as. N N
.81 9.8L. 1.2 , 8., ., N, . ,81 . .88 92 8. u
,... .81 88 . 9,2 ,8 . ,. .. a 81 :88 , 6.2 . 3 m
81 ., 8.8 . , as. ,8: a , 81 H .811..." .. ,3. a N
3.8 . 38 ad. . 8 . ,. ,w . 81 8.8 . 3. . .8 N
912...... .88.... .88.... 812.9. .88.... .86....
.o .81. o .3.. £91... 88.35 E .81. o .3.. 8.1: 888.5
.589... 821...“. 21.1.5 120 8.55 EoE:9< 821...". £1.12. 85 8.88
ouum use? smé 55.503! 2.0.2.250 ouum Rom ovum 9.03 :mi 9:530: 2.0.2.055 ovum Rom
5.5.. 8.126 ..~ .5 1195 «sum so... on 5.3 .3226 a co 829 1.8m so... a
.on 63¢... .au 28...

m x_nzmn_n_<

120

   

      

   

,1... .8 111.11.88.18

 

 

   

 

 

 

 

 

 

 

 

 

22.2.2 a... ct c>. E. .5. 88...... 3.. fix. .3..

13.31. 593381 .0 539.5 2.00.; .6850 .9629 we. 0 3:03.21 3:03.21 .25

Even? 310530 65301 ucaonom 811.5 .035. 5:830 3.8:”. .051 551 138:0 8.2.ch 2925
8.9.... 8.5111 82 3.811

 

 

 

 

 

 

 

 

 

 

 

59.... 12.1 so... o .61 5.3 .5226 ..~ .3 2.21 Sum 18.3 .3 as...»

m Xazmn—fi

 

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m x52wma<

392.2 E ct ea AE A2 fiscasea 3: E: 33.9; an: 3295
3 x33 coaafimom .0 529.30 Eoo_o> c3830 §o> A95 0 3:262”. 55:02.... AwEv 0 £90...
205:3. .8380 ,uSBom 958mm suns. 835. 8:28 3.2:”. 55”. SE :32...“ 8.2.5.: 2925 85
SEE 8:83. 32 Begum
595.. ouum so... a .xom 55> 5230 .h :0 on.95 auum so... cu .Nn asap

122

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

323.2 E at ca ct 3c 3.3225 3.: 3.: 3.3.53 an; 3295

.6 €an c8381 .o 8:25 £020) :8an 3.8.; 3.5 0 3:032". 6:259“. Amos 0 £92..

«55:3. 23580 2:801, 2:80: .895 682. 8:88 8.2:". as". .85. :823 8.3%: 2925 85
BEE 3:83. 32 8:83

 

 

 

59.... 83 so... a .xom 5.; .5226 a no 2.29 ouum cu... an .8 2...;

m xazmmnz

123

“$15.5.fimaibfl .
. :3: .5 r}. “WA”:

mama

ggfifi? .1..

m;

 

 

 

 

 

 

fizfifis
.Hdfifiwm... ggfi
992.2. 3. E. .5. .E .5. 3.3286 3.. 3... 38.2.5 .3.. .86....
.0 x33 couauwom .0 5.5.5 3.8.0) c3250 33o> A25 0 Eco—.3.". .6533."— AnE. 0 £92..
52552 58580 2:33., 258$. .08.... .03.... 8.650 8.2... .2... .3... c825 .5555 £925 85
33...". 3.83. 252 3.83.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

59.01. ouum :0...— a .xom :2; coin—.0 ..n :0 «no.5 ouum :0...— on in 030...

m Xazmfiz

 

124

.. .35.... fix 1 , 1. at. $5,. ,... :3 n g 1. .11. . . . 1 ......:...¢. . €70.95 Mg»
mewwfimfi..1 1...... 5... n. . , . . .,. ...w. .5 . . [Lu T ,1 .. .1; .. . . 152mwrrr§rai ..« n

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35.5.. E .E .5. .E .5. . 35.286 3... E... ,381525 .2... .86....

.6 x38 5.55313 5.55 b.020> confine b.0025 WE. 0 3:362“. 3:262". WE. 0 £90...

.5553 520580 2.6%: 9583. 62.5. .89... 852.5 8.2.... .8... .2... 5.5.3 8.2.55 2%.... 85
use... vocaom 3oz 3.8%

 

 

 

 

 

 

 

59.04 ouum :05 9 ion 55> :03030 ..N :0 2.0.5 ouum :0:— Nv new an .on .QN .2. .nn 03'».

m 502mg“:

125

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8.9. . 8.. 3mm , 3.3 . we a...
is. , H. .58. 9...: , 2.8 we. .8
a £6 , a... . .. 8m. 3...... 8.3. mo... 8
., a . m3 . 3.9, may» 83%.? 8.8, .3.“... $.35... .80 2.9.. 8.: 99 «N
.U.......Hu... _ 3.... 36..., 8.8 3.3 ...........u. 84.... .29 88. . _ 8m and» I 8.9 0.9. ., m.
. j
a. _ . .. 8... 8.3 8.“... . 8...... . 3.: 8.8. :3» 8E , 8w 9.8 38 gm . «v
. 4... , . 8.... 2.21 madngwmmfi..- 3.6.. 8.8 .58 2.2 .8... 8.8 9.8 a...” , 8
a, . 5... 8.8 - 3.3. , 8.3.. 33 8.8 28,: «NE. .80 8am 8am .- new 8
a 3... 8...... 9+... 8.3, 3.3 8.8 9.8 9.2 . , So 2.3 8.8 wow. an
.
a 3... 8 E 8.3 . 8.5 8.8 8.3. 32 2.5. . .o8 - 8.9. . 8.: can .9
[93.2- . ...........1.- .E ..>. .E I ..>..\F--.i M11335 \.~:. .N... .. 1.1-. muflfigm. .3.. .85....[
.o‘xmoq. . co_S.._.mmm.o c2550 3.02.5 c2650 zwomW>.. .mE. . 0 5:33.“. .6533...“ \\..mE. .1110 £92.
McwEamR. Eyoutwoo 9.333.. 28an SEE. .. .womaE. .1. 15.125... 3.9:“. w 5...". 5...”. -1 .ko..m.l...ro.|% 8.25:: 2925 no.0
1W pf. \. . 3.2... 1 8:33. I 52 41mg... m‘ .

 

 

 

5?... ouum so... a .3 5.2. 8.5.5 ..n .3 2.2a 33 go... 3. .2.“ on .8 in .2 .8 2.3

m x_n_zmn_n_<

126

APPENDIX (F)

TABLES (3742)

CORNER DROP EXPERIMENTAL DATA

 

127

4w fidméfi % 5 ‘
“fiamfififii

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

"— x_nzmn_n_<

322.2 E at ca AE Go 35323 3: 3: 3.3.85 :3: $22..
a :33 cessuoms 8:28 2.8.9 5.2.5 3.8.; 3.5 o 6:33.”. 8:38.". 3.5 0 29¢:
€253 EonEooo 2:33. 2:83. USE. tag 8:28 85...“. 5:". 5...”. 8:23 85.95 2935 8.5
at 8:83. 302 8:83.
new 5.3 .3226 ..~ co 820 3E8 5:. Va Km 033.

128

«a: 1&3)...» 1W7.

    

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

922.2 E at ca 3.: E: g as: .885
a x83 coazaszs c828 5323 £528 38.; we. 0 8:88.“. 6:38.". A95 0 29a:
.8235 23580 2:33. 2:33. 835. 632. .6323 33m SE .25. 5.85 39.5.: 2%; 85
3.2:". 8:83. 32 8:83.
.25 55> .5230 .h :o 396 5E3 :2: E .8 2:3

n. X_Ozmmm<

 

129

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

922.2 3 at :2 E E "3.5.825 3: fix. 38.2.55 :3: 39.2.:

a .303 533850 c828 §o> 8.5.5 avg; 35 0 8832". 3:83.“. A25 o 2%:

.8235 28580 6583. 288m 882. 685. 8:28 8.2:“. as". as“. 5:28 8.3%: 293$ .35
BEE 388. 302 8:88

 

 

 

"— x_azmmn<

.3 5.; .8226 :u no £35 3&8 so... an .3 03¢

130

:wi i:

I. 1.4%. #

ngftfix i??? 3.15:.

 

 

308::

 

c3950

$8.9

5:88

2%...

 

,ucaonam ,

952.01

88E.

c8950

82.55

5925

 

no.5

 

 

 

 

 

 

“5.3:”.

 

 

umtoaum

 

 

3w:

 

 

Begum

 

 

 

 

u. X_Dzmn_n_<

xom 55> 53.330 ..n =0 2.20 3500 :0:— on .8 033.

 

 

mg .w

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

922.2 E c: :2 RE .5 3283855 3.: it an: 32.9.:

.0 x33 c3333: *0 5:230 3.8.; 5:23 3.02; A25 0 35:52”. 8:252”. 3.5 290...

22:82 220580 238% 238mm .086. 882. 8:28 BEE .25. 22E 8:23 8.23:: 2925 85
8.2:... 8:83. 262 8:83.

 

 

x00 5.; c0225 ..N :0 80.5 .5500 :0:— Nv uca an .on .VN .3 .3 0.35...

"— x_nzmn_m<

 

132

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

$23.2 E at at .E :2 3.3283 3: 3.3.855 A3: A855
.0 :33 c3358”. .0 c883 Eoo_o> c3830 E02; .25 0 8:262“. 55:69.“. 35 0 290:
22:82 28580 assay. 8:83. .082. 682. 8:28 SEE as... as“. 5:28 8.3%: 2%; 85
“.225. 8:83. 32 8:83.
E85000 885. .3530 oz .xom 5:5 5230 an :o 820 3E8 so... at u:- un .3 .3. .2. .3 03-...

u. van—ZN}?

10.

11.

12.

BIBLIOGRAPHY

Burgess. G.J.. Lecture notes from Pkg 310 "Iaghnigal Princigles of
Pagkaging”. Michigan State University. pg 99. 1997.

ASTM D-4169: “Standard Practice for Performing of §higging Containers
and Systems.” Selected ASTM Standards on Packaging. 4th Edition.
1994.

Author unknown, 5 Steg Packaging Davelggment. date unknown.

Erickson. Roger. Comments taken from a fax message on package
testing related subjects, February. 1997.

Young. Dennis E.. "lntrodugion 19 Package Perfgrmance Teating” loPP
Taghnigl Jggmal. A paper presented at the Pack Expo 92 Conference.

Based on Conversations with William Hughes. Laboratory Manager. and
Eric Joneson. Director, Lansmont Corporation, 1997.

Granthen. Gary Carlton.. Thesis entitled “Predigting Shack Transmission
Characteristics for Ribbed Expanded Polygrggylana Cushions Using

Standard Cushion Cgrvea far Flat Plank Quahignaf. Michigan State
University, 1991.

Kuang-Nan Taw., Thesis entitled “Daveloging a Mathematical Model for a
45 Degree Edge Drog to Predict the Dynamic Behavior of a Low Densig
Closed-Cell Foam.”, 1988, Michigan State University.

Chen, George Kuo-Hsin.. Thesis entitled ‘Tha Effagt gf Rigbing an Shock

Transmission Thrgggh gnaggag Eglyamraga Quahign Materia .", 1986,
Michigan State University.

ASTM D-3332—93.. “Standard Tast Methfia [gr Maghanical - Shock
Fragility of ProductsI Using Shock Maghinaa.’ Selected ASTM Standards
on Packaging. 4th Edition. 1994

Allen. D.C.. “W2 Chapter 5. pp 547 - 5.67.

date unknown.

Singh. S.P., Lecture notes from Pkg 460 “Diatrigution Packaging”,
Michigan State University, 1996.

133

13.

14.

15.

16.
17.

18.

19.

20.

Zhang, J.. and Ashby, M. F., Mechanigal selaction of foams and

honeycombs used for packaging and enargy abaoggtion”, Journal of
Materials Science 29, pp 157 - 163, 1994

Mustin. G.S., Theogg and Pragig of Cushion Deaign. The Shock and
Vibration Information Center, United States Department of Defense. 1968

Gibson. Lorna J 8 Ashby, Michael. F., “Cellular Solids - Structure Progem
Relationshigs” pp 34, 37. 39, 42, 122, 127, 141-143. 147, 213 - 214,
Pergamon Press, First Edition, 1988

Author unknown, “Chggsing ma Best anhign'. date unknown.

Burgess, G.J.. Lecture notes from Pkg 805 “Advangg PackagingShock
and Dynamics”, Michigan State University, 1997.

Burgess, G.J.. " Some Th rm namic b rvations on the Mechanical
Pro erties of ushions'. Journal of Cellular Plastics. 1987.

Throne, J.l.. and Progelhof, R.C.. “ Clgaag Call Foam Behaviour Under

Dynamic Loading - ll. Lgaging Dynamig of Low Density Foam”. Journal
of Cellular Plastics, Jan/Feb. 1985.

Brandenberg, Richard K.. and Julian June.Ling Lee, “Shock in

DistributionI Product Fragility and Cushion Design.“ Fundamentals of
Packaging Dynamics. Skameateles: L.A.B..1991.

134

RRRIES

V. LIB
lllllllllllll
120377

ll

MICHIGAN sran UNI
llWWII“IIHIIWIIIIWH
31293017

 

1