THE USE OF A1/4SCALE IMPACT
YES? PLANT FQR SIMULATfiNG RAILROAD

SWITCHiNG OPERATIONS

The“: for the Degree of M. S.
MICHIGAN STATE UNWERSETY

Paile Jacobsen

1966

THESIS

   
     

  

LIBRARY

Michigan State
University

 

 

 

mam ”or ANAL
RUE/’33} DUE. um:

 

Ans'mAcr
run 031: or A ll‘ scar IMPACT

TEST PLAN! ’0! SIMULATIH; umom
SUITCHIW OPERATIONS

by Pelle Jecobeen

Hitb the objective of predicting denege potentiel ceueed by
helping end ewitching operetione performed in prectice on reilwey
freight cere. e leboretory ecele reilwey teet track with oer end
leding wee inveetigeted ee the model eyetcn.

Being intereeted in en output (reedinse concerned‘with damage)
fro-.tbe model efleiler to, or‘with e einple celcnletory reletionehip
to the reel eyetee during operetion. we tried to conpere the conponente
of the input to the two eyetele.

Inpect conditione, input-fore“ to the leding end eelf-excited
love-onto in the leding ell conetitute fectore, thet nuet be eeeumed
to be ilportent, with e leek of compereble infornetion.on the deteiled
behevior of the leding ee e function of the ewitching operetione elonc,
our etteept wee to dieonee end neke enggeetione in order to edjnet the
model fro- en «Viral-utel enelyeie-etendpoint, involving permtere
tech ee

- iepeot velocity

- negnitode. ebepe end duretion of impact forcee
-Iinpect quentnn (equivelent neee of etriked cere)
- rebound velocity

“'IDNIIOnt during end efter impect

-negnitude end duretion of input forcee.

Pelle Jecobeen

Bevin; reeched the input to the leding quite eucceeefully. we
propoeed direct end indirect einuletion proceduree using iteme or model
iteee iron which it ehould be poeeible to deecribe damage. Correletion
to prectice ie precticelly unknown. end peremetere euch ee

- euperinpoeed vibretioue on impact

- eeli-excited tendon vibretione in the
loed ceueed by internel eliding, etc.

eeeued inpoeeible to einulete in the leboretory. But we etill euggeet
the: proposed proceduree - eiter proper edeptetion of the teet treck -
ouch ee:

- eecuring etrictly horizontel impact end
'eowenent of ell outer componente

- rebuilding becketop to perforn ee 3 or 4
etriked cere in ectuel operatione.
preferebly with no rebound trevel
- reinforce conetruction end hauling devicee
to operete et leeet eeeeee of 1200 lbe.
eecurely
will eeke e welid prediction of demege poeeible for e wide renge of
connoditiee being treneported by the reilroed.

Air-cuehione included to divide leding into eectione were
inweetigeted end ehoved up to here e decreasing effect on the uegnitude
of input iorcee to the ledins. Further investigatione ere euggeeted
including other typee of loede then wooden blocks. preferebly with e
controlled type of internel friction, ueing unitired loede end enoother
eliding eurfecee. Alec the effect of unitirins, bracing end feetening
eeene euch ee eteel ecceeeoriee. non-ekid pletee, enchor pletee end

reterder pletee in connection with eteel etrepping end wooden type

dividere cen be invertigeted end competed to eir-cuehione.

Pelle Jeeobeen

An ettenpt to reproduce the deeege of cenned goode in fiberboard
boeee were not eucceeeful, but damage wee obteined by expoeing one

or two cene to the eexinun obteineble input-forces.

THE USE OF A 1/4 SCALE IIIPACT
TEST PLANT FOR SIMULATING RAILROAD

SWITCHING OPEATIOKS

BY

Pnlle Jacobeen

A TIIESI 8

Submitted to
Michigen Stete Univereity
in pertiel fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Depertuent of Foreet Products
School of Packaging

1966

ACilIflhlEm $311213: 3

The "School of Packaging," Michigan State Univereity, ie

epprecieted for the provieion of the neceeeary teet facilitiee end
inetrumentetion for the etudy. Alec the etaff of the echool is
epprecieted for the generel beckground in peckaging I received in
couree work and lecturee. end for the hints end advice, especially
from :7 principal adviser Dr. Jemee Coil. Very useful information wee
provided by the “American Aeeocietion of Railroads.” ”Reclemetion end
Deluge Drench" (Hr. Burt Willieme) end the Weseerch Depertment"
(Dr. L. L. Oleeon) end from verioue companies. A epeciel thank you to
Mre. Herilyn‘flolu for cleening my written language end for typing the
theeie. beet but not least. I want to thank my wife end children for
ell the help end patience with their husband and father, who really

epent too little time together with them during the period of etudy.

ii.

TAEE OF CONTENTS

ACKMEWMIKS..........
TAEBOFCONTENIS .........
LISTOFTAH.ES...........
LISIOFFIBURES ..........
LISTOFAYPENDIXES.........
1.0NECTIVES...........
2. GENERALBACEROUHD.......

PART A:
ENRONHENTAL ANALYSIS Am CHOICE OF

St‘tGmOU£ Of PrObIGa. e e e e e
HOChOdOIOE’ e e e e e e e e e e
Dreft Genre teeted end comnte
Choice Of ‘0‘! e e e e e e e e

PART B:
TRANSIW PULSES 1'0 IRE LOAD AND THE

Introduction e e e e e e e e e
SUEDer Of finding! e o e e e e
Theory of eir-euehione . . . .

Experimntel work end diecueeion

T‘bl.‘ .fld ICIUIt' e e e e e e

PART C:

REMARKS ON SIMULATION AND REPRODUCTION OF DAMAGE . e

Sinu1.t1°n at beCC. e e e e e
Simuletion of dmege . . . . .
The one of reeilient leding .

Experimentel work end celcnletione

TEST DRAFT CFAR, ETC.

EFFECT OF PNEUI‘IATIC

e e e e e
e e e e e
0 O O 0 O
0 e 0 e e
e e e e e
O O C I 0
e e e e e
e e 0 e e
o O o O O

I O O I
O O O O
O I . O
O O O O
O I O O
O I O O
O O O I

O
O
O O O
O

6. SUMMARY AID COHCLUSION. SIBGESTIONS . . . . e e . .

7. Bmxmmm O O O O O O O I O O

APPEmIXEs O O O O I O O O O O O O 0

iii

Page
ii
iii

iv

vi

ll
18

DUNNACE 23

C O O . O

26
25

32
3‘5
53
£4
£5
58
50
55
58

61

LIST OF TABLES

Table Page
1e “wetVClOC‘t1CUOOOOeeeeeeeeeeeeee62

2. Heaprene end Cambiuetione with Neoprene.
Dreitceere.....o................13

3. 0th.: Meriele 13 Dreft Curl e e e e e e e e e e e 1‘
4. Pulee Cherecterietiee for Selected Draft Genre . . . 15

5. Degredetion of Selected Heteriele need in
Dreftceer'reetlo................o8O

6e cmr‘CtCrl.‘£¢ O! G-Rubber-Geer e e e e e e e e e e e

r»:
C)

7. Inpnttouding,hont8ndofCeroo..o..... 33
8. PorceeAlomendUpeide-Dovnthe Cer .. . . . . . . 39

9. Pulee Characteristic for Pneumatic Dunnage -
BIIZIthoid...o...............(.0

10. Pulee Cherecterietiee for Pneumatic Dunnege -
VerloueConditione.................42

lle “fit. on M113 Including CIM e e e e e e e e e e 52

iv.

LIST OF FICUR

Pizure Page

1. feet tree): with on end loed. Including mmmonc
.ir'cn‘hion. ‘3‘ two blOCk IO‘d c.11. e e e e e e e e e 2

2. Inetrueentetion. Counter end reley for evitchee.
Aopliliere. oecilloecope end POLAROID cenere.
.ICk‘tOP O! ‘0': tICCk ‘3‘ light 50“ tr188.r e e e e e 2

3. moms-mm! Conpreeeion Heehine. Pulling eecheniee
0‘ th. tOIt track e e e e e e e e e e e e e e e e e e e 10

4. Heteriele need in experience (eee text). Pull eiee
v.90: b‘s IttCr 1NP.Ct. Poeitlon ‘ e e e e e e e e e e 12

5. Peek force venue inspect velocity for the eyeten
6-1’ubber-dreft gee: plue rigid lumber-Iced . . . . . . 21

6. Land (ector Verena impecting nee for the eyeten

350'. e e e e e e e e e e e e e e e e e e e e e e e e e 22

7. 6-fieoprene gear + Pelt damper. Poeition 3. “5 lb.
loed. Pulee end derived velocity. trevel-end etreee
"ICUC dIIIGCtiofl-CurVQC e e e o e e e e e e e e e e e 83

a. Gene efter expoeure to impact: from the eide.
diegonelly (3 etegee) end iron the end. Device for
dtlflODll t..t£fi‘ in 9106. e e e e e e e e e e e e‘e e e 54

II.

III.
IV.

V.

VI.

LIST OF APPENDIXES

Impact Velocity Vereue Poeition on Incline,
Theoretical end Heeeured, Uncerteintiee involved

Selected Photographic Seriee for Draft Geer
Puleee,TebleeZend3............

BmvaOthlftcflfirUeeeeeeeeeeee
Cherecterietice for the Test Treck System . . .
Sample Derivation of Velocity, Dieplecement end
Streee-Deflectioanurvee for the 6 Ring Rubber

cmuthICICDmmreeeeeeeeeeeee

Selected Photogrephic Seriee for Puleee to the
Miragr‘b1037gapg‘M10eeeeeeeee

vi.

Page

61

65

76

85

91

96

l. Obiectives

The purpose of the present study is to analyze and to sane extent
deve10p an approach in the field of package testing. namely simulation
of railroad hunping and switching impacts by means of a laboratory
scale test plant. The test plant as originally built by "The

Chesapeelaes Ohio Railway Company" (1)”

. which has donated it to the
"Peckag in; School" at Michigan State University with the expectation
that further develoment would take place. Figure 1 shows the built-up
of the test incline. backstop and car.

The present study in the first part is devoted to development of
a reasonable input to the test car with its load. Comparisons are
made to field cond mom”. Nat the transnitted pulses are measured
to the load of wooden blocks with and without air cushions (pneumatic
dunnage) in different where. locations and exerting different
preemres. Finally a few remarks are presented explaining which
procedures may be used in the simulation of common types of denege
observed duritg railroad transportation. One exanple on reproduction
of damage is given.

Some suggestions are made for improvement of the test device

itself, as well as preposals for further work with the railroad.

 

1)
timbers in parenthesis here and in the following refer to the
references or bibliography back in the thesis.

2
)lnfortsat ion given by courtesy of Association of American Railroads
(AAR), 59 East Van Buren Street, Chicago 60605, 111.

Figure 1

 

Test track with car and load. Including TRELLEBORC
air-cushions and two block load cells.

Figure 2

 

Instrumentation. Counter and relay for switches.
Amplifiers, oscilloscope and POLAROID camera.
Backstop of test track and light beam trigger.

2. General Background

2.1 Since 1963 the incline unpact test for shipping containers
(the Coolant-tester) adopted by American Society for Testing and Materials
(AS'Ei) (2) has had a widespread use in the U.S.A. as well as overseas.
The standard being, set first of all to secure uniformity of the
apparatus used. defines an impact angle of 10°. a rigid wooden hamper
perpendicular to the lovenent of a dolly. and a calibration in terms
of tapact'velocity when the empty dolly is hitting the rigid bumper.
Although the device may have some value in comparison of one container
or package to another. it is quite evident that it will not necessarily
‘gire pulses comparable to pulses generated in real switching operations.
As a matter of fact. measurements by use of an Endevco piezoelectric
accelerometer (3) have shown that the load-factors (CZ-factors) are much
higher and the durations of impact shorter at a given impact velocity
with the rigid wooden backstop than it is when cars are hammering each
other with buffers between them. On the other hand. the forces due to
the enormous masses behind an actual rail impact are much higher than
at the Conbur-tester. According to the concept, that (a) duration
(frequency), (h) magnitude (force, acceleration, displacement or stellar)
and (c) shape of the pulse (eventually expressed as the area) together
determine the potential risk of damage for the particular container
or package, it will be evident that the Conbur-tester has several
shortcomings.

Also in recent years. some companies built inclines with the
backstOp as a sandwich construction of steel plates on a core of hard

maple. This would give a more resistant surface and be similar to

.1.

cannon drop test operations made against steel plates. This, of course.
will again change the pulse developed, and control will be difficult to
obtain.

. The most interesting attempt presently being We might be supplement-
itg the dolly of the Conbur tester with hydraulic type shock absorbers”
that are able to generate almost any type of pulse as known from, for

example, the BYOB-shock-machine‘)

and the Honterey type drop tester. The
pressure in hydraulic cylinders and the size and shape of orifices between
cylinders determine the magnitude and shape of the pulse (4).

2.2 A true simulatim device undoubtedly is being provided by using
a full scale ramp with striking car and hammer car plus instrumentation. '
The only disadvantage here would be the space occupied and the price
for establishing such a device (about 1 million dollars (5).)

5) that are used not

Several companies do have full scale test ramps
only for direct meamrment and simulation, but also as "calibration
devices" for developing smaller scale-test ramps. This last mentioned
principle night he the way to go in the future, not only for calibratirg
laboratory scale devices according to the input in practical operations, but
also accordirg to damage observed. The only approach in that direction
published for normal freight conmodities seems to be du Pont de Nemour's

(7) dealirg with bag—damage.”

 

————— ‘—

3) Information obtained by the courtesy of Gaynes Testing Laboratories,

1642-105 West Fulton Street. Chicago 12.111inois.

0 Information obtained by the courtesy of Dept. of the Army, Picatinny
' Arsenal. Dover, New Jersey.

3) Signode Steel Strappdng Corp. Miner Corporation (5). Pullmann Standard(6)’
I. I. du Pont de Nemours E Co. (7). Aberdeen Proving Ground.
Sandie Corp. (8).

5) Also published in ASTI-i Special Technical Publication No. 32/. "Simulated
Service Testing of Packaging".

PART A

EWIROHMETIAL ANALYSIS AND CHOICE 0?

TEST DRAFT GEAR. ETC.

§_t_Latanent of problems

In the earlier work with the lib Scale Impact Tear. Plant (1). different
leatha of hard rubber (Neoprene, Duroaeter A-Hardneea 70) in the form of
O-rinaa 3/4 in. thick and with eoft rubber (Durometer-U eponge rinse 1/5 in.
thick between each two of the herd rubber tinge were need. Theee draft
gear couponente uere mounted on a hollow eteel cylinder eliding againat
Irate linime made of aabeatoe filled. hard textile material underneath the
teat car. It ia pouible to tighten the brake lininge, but not in a very
eatiafectatily defined way. The rubber componenta accounted for noet of the
energy abeorption capacity of the teat draft gear.

There are aeveral typee of draft geera in practical usage:
(a) standard geere characterized by a travel of 2 - 6 1/2 in. (often-

tiaea about t in.) between coupler and body in either direction.

Theee care are further epecified ae havim a nininun energy abeorption'

capacity and a narinua till-preaeure (for exonple, apecification

m 244013—62: hininun capacity 36; 000 ft.-lba., uteri-um reaction

force 300,000 lba .) by the nae of certain qualification teata (9) .
(b) Special ouehionim devicee. including alidim center eill. end of

car devicae or column-connected draft geare. Theae sure have

cabinet! travel of lore than 5 in. up to about 36 in. or even ‘8 in.

About 251 of 80.000 care purcheaed in 1965 were ouahionad care. but

atill the lain part of about 1.3 million railway care in the U .8.A.
eoneiat of the conventional type equipped with atandard draft geara.

A euehioeed ear nay coat twice ae Inch ae a conventional car. Several

fir-a ere lenufacturiq draft geare,” and the Hiner draft gear in

7) Miner Corp” 209 8. La Belle Street. Chicago; Waugh, 332 S. Michigan Ave.
Chicago: National Caetinaa. 22‘ 5. Michigan Ave.. Chicago; Cartvell
Iieetiuhouae. 332 8. Hichuen Ave» Chicago.

7

one of the most widely used. It is also used in some places in
Eur0pe. Typical characteristics for Miner type so. RF 6-29 for
example are a travel of &.33 in., capacity £2,000 ft.-lbs., and
sill pressure 380,000 lbs. The characteristic pulse under standard

impact conditions (hammer test) is described under (a) below.8)

It is clear that about 2/3 of the cherry absorption by the chosen
type of standard draft gear is due to friction between brake shoes, and
that only 1/3 is energy-storing by spring action. In our simulation we
are unable to reproduce such a high relative amount of friction; therefore,
the obtained pulses show a characteristically different 335333
(a) The standard rubber-friction gears show a moderate increase in the

coupler force in the beginning. then a steep increase until the

maximum force, and thereafter a gradual decrease with a considerable
drag out of the pulse. It is remarkable that not only Miner Corp..

but also research authors such as Baillie (10) and Fongevesn (ll)

indicate that type of pulse as typical. On the other hand, those

”clean" (noise-free) pulses are not taking into account the super-

imposed vibrationa that occur under impact conditions (I?) which may

represent a damage potential for certain types of commodities.

To simulate and describe also the superimposed vibrations,
instrumentation may be required such as oscillographs (whereas in
the present study, we have only oecilloscoPes), frou.which shock
spectre nay be derived by mans of a tape-programd electro-
dynanic shaker. (See reference (8) and tho Bibliography (13) ~ (17)

for shock spectra.)

 

8
) Information given by the courtesy of R. H. Miner, Inc.

(b) Despite the limitations touched upon , the pulse derived from
rubber draft gears (1) may be made acceptable by prOper
arrangement for use as inputs. In the following section. the
experimental results are discussed in detail for different types

of spring type draft gears.

The duration of a normal impact pulse is considered by most authors
to be 75-100 ease. and it was possible to obtain that duration fairly
accurately by using a 6 ring rubber test gear.

The EEEBEEBQE.1“ terms of load factor (G's) goes from 2-3 a in. normal
commercial operations, involving i~6 mph coupling speed (the allowable
speed in coupling is a mph, which is observed in nest automatic or semi-
automatic switching operations in switching yards) to 30 3's or more in
Military Operations. where a possible impact speed of 10 mph is taken into
account (18). A.requirement of 8 mph as design criteria and limitation of
the carrier operation to a maximum of 6 mph is proposed in the NASA Report
HR 1262 (l9).

It was decided to present a characteristic of the teat device forces
as a function of impact velocities and the weight of load (consisting of
wooden blocks rigidly mounted between the two bulk heads). This enables
us to choose the magnitude that is desired for the particular simulation.

Hith the available instrumentation. it is difficult to compare test
pulses with actual pulses as far as vibrations are concerned. So it was
decided to clean the pulse rather than have an undefined noise superimposed
on it. A felt damper freely mounted at the backstop was very effective in

that respect.

It is convenient to use e standard pulse in laboratory simulation
tether than the more complicated pulse shape occurring in the field,
eccording to the opinion of severel euthors (20) (21). However, the
lore ideel method for the purpose of direct simulation.vould be to
use the shock—spectrum derived es mentioned before from the sctual
impect conditions. Also extension of the shock-spectre concept
coverim e two degree of {radon-system has been suggested (19) . when
union the shock-spectre es the basic input. expensive instrumentation

end highly skilled personnel would be necessary.

Experimental work and calculations

Methodologl
The incline is eerked with Positions 0 - 6. They correspond

to verticsl heights shove the floor level es follows:

Position In2
0 9.4
lB.‘
17.5
21.5
25.5
.29.6
33.8
At e reletively eerly stege of the experiments. it heceee clear

Quit-tn”

thet Position 0 geve too lerse en uncertsinty in the impect

velocities V neesured et the end of the course. end Position 6 could

i
herdly be used heceuee of the leek of cspecity of the pulling nechenisn.
Position 6 corresponds to e 6.0 eph impect velocity. For the higher
loeds Position S was not used either.

Positioning the cer et the incline nee sccomplished by the use of
e stop. positioned et e 90°~etcle with the heck end of the cor. The

theoreticel inpect velocities then were celculeted end compared to the
velocities measured (Appendix i).

10

i t
v'fiy‘
J r

 

BALDWIN-EflERY Compression Machine

Figure 3

Pulling mechanism of the test track.

 

11

A Tektronix Oscilloscope. Type 502. Dual Beam with a camera
(POLAROID) was used to monitor the signals from strain gage load cells:

1 tubular shape load cell;
top capacity 30,000 lbs.

2 wooden block load cells;
top capacity 1,500 lbs. each.

vie amplifiers PH 153 X and PH 125. The load cells with respective
amplifiers were calibrated twice during the experiments at BALDWIN-
IMERY’Static Load Tester. Figures 2 and 3 show the apparatus arrangements.
The oscilloscope, triggered through a light beam being cut by the front-
wheels of the car, placed the impact pulse at a convenient position on the
oscilloscope screen.
Draft Gears tested
Tables 2 and 3 show the different types and sises of draft gears

tested, with the picture reference to the collection of pictures in
Appendix I! and Table i. Figure 4 shows from the left to the right:

On the table;

Asbestos-filled brake lining

Tufflex pad, after testing (above)

Glassfiber rings

Hard felt pads

Standard shock absorber

Neoprene rings

Silicone-rubber rings

Blue Polystyrene (closed cells)

Shite Polystyrene (open cells)

Polyethylene-foes

'“l’rue Hark" felt pads

Behind:

Rubber-steel sandwich pad.

12

 

Materials used in experiments (see text).

Figure 4

Full size paper bag after impact, position 4.

“

 

13

Table 2. Neoprene and Combinations with heoprene

Draft Gears

~w

 

 

 

em..-..V....-...,...,.-..m-.........n--..--. .-..-.,.-. Wm item... we-.-” PPicturefl
ho. Composition Reference
. x ) I.
1 12 Leoprene with 1/4 Sponge Rubber 1
between
2 Sane as No. l. thOugh 1/2" glass wool 2
felt in front, only 11 hecprene
3 3 Neoprene with Sponge Rubber between 3
4 Same as 3. though 1/2” glass wool felt d
in front .
5 ~ As 3. plus 3 x 1" Tufflex in front 5-6
6 As 3, plus 3 x 1" polystyrene xx) in front 7-9
7 As 3. plus 1" polyethylne foamxx‘) at 10-11
the beckstoo .
3 6 neoprene with Sponge Rubber between 12
9 Same as 8, plus 3 x 1" Tufflex ' 13
10 Same as 8, plus 3 x 1‘2" felt pad spot-
glued on backstOpxxxx 14
11 As 8, plus 3" felt (clothing type, soft), 15-17
plus 1/2" felt pad on backstop
12 As 11. chaugh without felt pad on backstop 13
'—.~‘—.

 

 

1:)

Xx)

Duromcter lurdness 70, durometer A. Thickness 3/4".
Open cell type, Dow. white. density as new: 0.043 g/cm3.

xxx) Dow "Ethnfoen", density: 0.060 g/cmg.
xxxx) "True‘Hark" pad for office machines. hard; Distributor:
University Typewriter 60.. 1912 East Michigan Ave.,
free to move horizontally except for the 3-4 spots of
glue to the front face of the backstop.

14

Table 3.

Other Materials in Draft Gears

 

Picture

 

 

 

 

No. Composition Reference

15 10 x 1/2" felt plus 3 x 1/2“ felt on 25~25, 27
backstop

l6 Sane ss 15. though 2" sandwich rubber 29
pad‘) on backstop

17 10 x 1" polystyrene“) 30, 32-31.

13 6 x 1" polystyrene (continued from 17) 36-37

19 6 x 1" polystyrene plus 3 x 1/2" felt 38-39
damper on bsckstop (continued from 18)

20 Sue ss 19, sfter 48h in compression 40
(continued from 19, half length)

..... - - - - on . - - ‘ - - - - I. .. ~ - - ~ - -% .I - - - - - -

21 8 x 1" "Ethafocu". plus 3 x 1/2" 61. 46
felt damper

22 3 x 1" "i-Ztlusfoam" plus felt damper 65-46
(continued from 21)

23 2 x 2 llfi" Silicone Rubber XXX) 47.48.59.50

26 Commercial types hydraulic shock honcxxxx)
absorbers

8) .

Used on arOp-tester. Ono random sanple of stooldrubber
sandwich-construction.

xx) "Styrofoam". blue, density: 0.039 g/cm3.

xxx) Dow "Silastic 583 RT?" (selfcuring compound) prepared after
24h hardening at roan temperature and 43h hardening at 130°?
ss recommended.

xxxx)

"Volkswagen" standard absorber, 1 1/2" diameter, and GABRIEL

Type 43C (were both bottoming very fast).

15

nu. l». Pulse Characteristics for Select-d Drsf: Guts

 

 

 

 

 

 

 

 

On: Picture st. Force Duration
Ms '0. Weird H.18ht) ‘ u” L 1)
Lbs. t I: Ranks
lo 0
l l 2300 155 160 Havcsine; noisy
3 3 2750 105 120 Close to linearmoisy
8 12 3100 85 100 Parabolic; noisy up
toR' so mac: drug”
10 1‘ 2750 110 120 tall. 60 asses
symmetrical; no miss
- - u a d ----- Id -------- .4 I. as q «- n- 1» ...... - - u - a
i l» 2330 110 125 Some noise, most
upwsrds
S 6 2700 100 113 Noisy, both up and
down
6 9 2750 105 120 toR' ‘5 mssc; sons
noiss
7 11 2400 125 MO Little miss
.- O - n q — - - s- - J ........ ad ~— — u c- - 1 ccccccc - - CI . -
ll 16 2600 100 120 Lou traqusncy noise;
. ' only up
12 18 2700 100 115 Noisy; me than
{ picture 17
o - - u u - o o s- - a1 ———————— cl - c- 1r - - 0 uuuuuu u u n - a
5 S 2200 130 180 Hsversinc, altost
symctricsl; low-ascnituds
noise .
6 7 1700 145 160 ton. 120 Inc. trian-
gular miss, noisy
in 2040 misc
1)

Tbs pulse chsrsctsristics sro defined in s SANDIA Corp. proposal
standard (25). In describim shapeAV would not suffice, when
maxim different materials like here. Therefors the attempt to
13:13ch as the rmrks show. t0R indicates tiss tine.

16

Comments to Table A
Effect of length

There was a change of shape from almost haversine to almost parabolic
with significant drag—out, when going from 12 to 6 Neoprenedrings. In
all cases, the noise in the first 15-5*nsec is of high frequency and
has a magnitude up to alnost the peak force.
Effect of dampers

The 1/2 inch glass wool had little influence. and the 3 inch Tufflex
and 3 inch polystyrene had pg influence, except in the very first impact.
The polyethylene were damping some noise in the beginning of the pulse.
The resilient type of rubberized felt was very effective. The clothing
type felt was effective only in the first couple of impacts. where it
cleaned for noise of higher frequency than that of about 100 cps.
Effect on magnitude_§nd duration

Polyethylene_drags out the duration the most. The glans-wool will do
some. The inverse relationship between pulse duration and magnitude holds.
Degradation

The normally accepted cushioning materials such as ruffle: and
polystyrene had an energy absorption effect (most polystyrene) and damping
effect (most Tufflex) in the first blow and showed the characteristic
shapes only in that first blow. ‘The effect of high mass plus low G-factor
(compared to drop) made these cushions useless. although the sire-order of

the force was the same as in drop-test operations.

l7

Comcnts

All draft gears were tested from position 2 on the incline with
an empty car (say Vi I 3.0 Iph, M2 . 573 lbs.). The thinking behind
this was that if the different types of relatively soft eateriels do
not act satisfactorily at these relatively small impact momentums,
they must be excluded. If one or more of the combinations acted good
enough at the stated mutton. they ought to be tested thromh a whole
range of Ionentums.

The draft gear originally supplied with the apparatus ads of

12 Neoprene rings gives a pulse duration of 160 nee. See table 6.

Accordim to the fomulses:

 

c_- \FZbkz/hz and III urW‘T"! k2
for linear elastic systems without damping, we get:

I" '12] 2-“: °WV2hIg II Constant

(independent of spring constant).

He therefore assume that no single elastic or nearly elastic ustuial

will be able to decrease the pulse duration without increasing the

min force, and the maxim force divided by load (is-factor) is
already fairly high towered to the actual enviroments.

(1) He tried softer materials having the possibility of high duping
which could act in a shorter time. cornered to the rubbc gear. by
the seas energy absorption. These included glassfiber, felt
materials, ruffle: and plastic-foams, and plastic materials such

3. I.“ (22) (23)e

18

certain
(2) The other possibility is to use non-linear springs. These are

able to produce square pulses or at least pulses showing
considerable hyperbolic tangent elasticity which means a
relatively low In, if bottoming does not occur. Hydraulic
devices, as mentioned before. nay be useful; and also, certain

11) could have the desired properties.

polyurethane constructions
The fact that Miner Corporation is experimenting with polyurethane-
steel sandwich construction for draft gears supports the desire to
try that material. Unfortunately, the time did not allow for the
manufacturing of Odrings from polyurethane. The use of specially
designed fluid springs. DIENE (Butadiene rubber with 30! oil from
Goodyear) confined into a convenient shape, would be still another
valuable attempt in the direction of square-type~pulses (26).
Choice of gear
Unfortunately, the softer materials and combinations tried showed
up to be somewhat degrading (see also Appendix 11!) or plastic. so
that they could be used only once, or a few times. The best material
was felt; although, it did not give sufficient decrease in duration
without increasing the laxinul force. It had a very valuable effect.
namely. the cleaninc of the pulse for the undefined noise especially in
the beginning. Three inch felt was counted by spotgluing to the

tubular cell on the backstop in all further work.

 

11)
For example, ESTANE 5701 Resin, manufactured by B. I. Goodrich

Chasicsl Company. Cleveland. Ohio.

19

After all. the negative result for simulating pulses just by
uain conventional spring-type materials was not surprising, when
considerizg, that the actual draft gear construction includes a
cons‘idersble amount of energy absorption by friction.

The best available draft gear chosen for use was a 6 ring
rubber draft gear with a travel of about 2.0 in. by e 2800 lb. load.
The travel was measured by using the light bean-triggering systen,
loving it from the 0-position (the point of impact), to the position
where it was not switched off at a certain impact. Table 6 shows
the final characteristic values for the test track using this draft
gear. and figures 3 and 6 are corresponding curves (see Appendix IV
for further discussion) .

Comparing measured and pulsfleigivcdéebound velocities and maximum
deflections

A final comparison of velocities (AV). derived from the photo-
graphed pulses. with rebound velocities (V

R
interesting weaknesses of the system. Especially the probable stand

.- Vt + AV) revealed some

forces is the systen due to backstop construction was rmrkable. It
seems desirable to stake the direction of forces and velocity truly
horizontal. in order to get better cowsrable results as far as rebound
velocities. (See Appendix V for further details.) This probably plays

an important role for the load-damage potential of the system.

 

 

Table 6. Characteristic of 6-Rubber-Gear
Peak Duration
Position Impact Rebound Force (Pm) of Pulse
on Total Velocity Velocity Measured lbs. fllsec.
Incline Load in/sec. in/aec. Vel.A V (Fairad (“f/2)
lbs. (V1) (Va) in/sec. Height) (101 Lb its)
1 575 42.5 21.0 65.5 1850 120
2 ” 52.5 27.5 86.5 2900 100
3 " 62.0 35.0 105.5 3700 90
4 " 71.5 39.5 119.5 4300 90
5 " 81.5 43.5 133.5 5300 85
II .- - Cr 0- n- - -— ooooo 1r - c- c- c- 4 a- u- v- a— r- ------------
1 795 40.5 23.0 63.5 2350 110
2 ” 52.5 30.0 82.5 3500 100
3 " 63.0 36.0 99.0 4550 100
4 " 72.5 42.5 115.0 5400 95
5 " 80.5 (48.5) (129.0) 6350 90
CI - D - 1|- - I- — ---------- II- o In va- nnnnn aa —————
1 1020 39.5 23.0 62.5 2850 120
2 " 51.5 31.5 83.0 3950 110
3 " 61.0 37.0 98.0 5100 110
4 " 70.0 44.0 114.0 6200 105
a- a— - ens (b e- a. e- ..... p u c- a— e- on. e- - e- - 1 ...........
1 1200 32.9 20.3 53.2 2900 ~125
2 " 43.2 27.4 70.6 4250 120
3 " 54.0 35.8 89.8 5750 115
4 ” 63.5 40.5 104.0 6800 110

 

 

 

 

 

 

 

huuoouebmm
.wemxcw

21

0"

 

 

b Ob. ' I "N." L “at. N? P m. N I "b. fl P n“. d b
4 a q q d 3 d 4 a e 4 a e
on me an me go mm am we on mm on mm ...
n.
x a
c
a.
..n~ man
.nxa news econ ucnaa— ocean +.uuoe ounce nuance e
.aa~ w
anon wnuafioacu
\t
.ebu non

w ownuuh

l
.. .
.e

m

moan

coma

G¢GN

ccmN

econ

ccmn

crew

acme

occn

ocmm

Qcoo

.ecq wowoh xeoh

 

22

Figure 6
Load Factor

an

  
   
   

   
 
 
    

7
'. (7O In/Sec.)
d. 6
(60 In/Sec .)
.. 5 e
(51 In/Sec.)
A e

V1 - 2.3 nph. (4O In/Sec.)
0 g)

4} 2 a
head Factor Vgrsug Igpagting Has; to; the fizgtgm

x
)Relatively high uncertainty

Total Mass n

1 \
7'

200 400 600 800 1000 1200 Lbs.

 

J-

PART B

EWJSIEIH' PULSES TO THE LOAD AND

THE EFFECT OF PNEUMATIC DUNNACE

23

24

Introduction

(1)

(2)

(3)

(4)

(5)

The following measurements were taken:

Transient pulses in the load of wooden blocks, horizontal direction.
Position 4. Load: 450-690 Lba.

(a) in front; lower and middle row

(b) in the middle: middle row

(c) in the back end; lower row.

Check on forces in‘vertical direction.

In the earlier report (1) the peak forces all the way along and

upside down the car were mapped. Rather than repeat such

nee-urementa of ltaited intereat aince they are only valid for

the wooden block type of load, we went a step further to measure:

Transient pulses in the load of wooden blocka, horizontal direction.

All positions and all loads used in Part A were also used in

Part B. Only in front; lower, middle or upper row.

Transient pulaea, when l, 2 or 3 pneumatic type cushions with

different air preaaurea in them were placed vertically between

wooden blocke along the load.

General check on the validity of measurements:

(a) inter-changing the wooden block load cells

(b) turning wooden block cells 180° around.

do change of amplifiera was made. The load cell marked (2) was

aluaya connected to amplifier PM 125 and the load cell marked (4) van

connected to amplifier PR 153 X.

25

Summary of finding;_

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(3)

There are slightly higher forces in row 1 (near the car floor
level) than above in rows 2 and 3.

No vertical forces of any significance compared to the horizontal
could be observed.

The forces decreased rapidly along the load from the front to

the back and;

In the back and, two impacts of the same magnitude size-order
occurred; first. the rear hulk head reaction, and after 100 uses
the Iain pulse appeared.

Theznagnitude of the transient pulse followed monotonously

the magnitude of the impact pulse. In the range of impact
velocities used, 7-151 of the impact force in the front end was
input force to the ladies in the particular row.

The pictures showing two or more peaks. observed at certain loads,
were due to load cell construction rather than real input
characteristics. The blurred peak picture makes the uncertainty
in determining peak force close to the maximum allowable tolerance
1'- 15: (25).

The air-cushions dimixdshed the input dynamic forces in front

with 50—651, and extended the duration; dependent upon number and

position of cushions. but almost independent of air-pressure in

the range 1.5 - A.5 psi.

The feedback on impact force is negligible.

(9)

(10)

(11)

(12)

26

In addition to dynamic forces. static pressure occurs.
Generally by increasing static pressure. severe vibrations are
superimposed on the main pulse.

These vibrations may be characterized as random, though 60-80
and 800-1000 cps. dominated quite often. Their duration is
20-50 maec. The magnitude went up to a maximum size order

that was the same as the peak force of the main pulse.

The vibrations were induced by internal friction in the load

of untreated wooden blocks - and normally supported by a

medium high level of static pressure. As soon as the friction
was brought under control by proper positioning and fastening,
in addition to smooth sliding surfaces, the superimposed
vibrations diminished. '

Coupling between load components and "feedback" of the characteristic
pulses through the load was supported, when two or more cushions
spaced about 50 in. were used (floating loud). "he“ only 090

cushion is used coupling only acours at the highest pressure (5 psi).

27

The theory of airkcuahiona

Figure 4 shows three sizes of pneumatic dunnage.

Many cushioning materials such as foam materials have their
effect because of air entrapped in then (pneumatic type cushioning).
However, their effect is not independent of the plastic and damping
properties. etc. of the material that encloses the air.

One sir-type cushioning material. classified for use in normal
light weight applications, is "Air Cap." 25) It consists of a trans-
parent, sir-tight, double aheet which encloses air-bubbles approximately
3/8 x 3/8 in., end fulfills the specification MIL~C~81013 for aircraft
use. etc.
the next smallest "pure" aircushion seemed to he PNEUPACK-Luftkisaen,26)
which is used for the packaging of electric and electronic parts in
Germany. It consists of high~frequency sealed PVC pillows vith an
air pressure of about 2.8 psi, when unloaded. It comes in 10 x 10 cn.
and 24 x 24 cm. pillows with a height of 4-6 cm., respectively 8-12 cm.
Maximal allowable static load is indicated to 1.6 psi (10 kg pr. 10 x 10,
cl.). by which load the deflection is approximately 2 cm. The material
itself contributes considerably to the.danping property of the cushion
(the enclosed air is principally an undamped spring).

Pneumatic dunnegg

Dividers between sections of the load, mainly the wooden dunnage

type and pneumatic dunnage, are used in transportation by rail, and

eventually by truck. They are not used inside the packages.

 

25) Sealed Air Corporation. 179 Goffle Road, Hawthorne, New Jersey.

26) Erich Gericke Plestic-Vererbeitung, Hedemannatraaae 11, 1 Berlin 61.
Gcmn’e

28

Most of the time these devices are used in the doorway part of
the car for the double purpose of spacing the load, filling voids
between load—sections. and providing a "floating load" type of protection
with complete recovery (26). It is indicated too, that air-cushions.
besides decreasing the acceleration response in longitudinal direction,
will damp vibrations in a vertical direction and lower the frequencies
transmitted (27). Standard sizes are 36 r 60 in.. 48 x 48 in. and
larger.27) Hanufacturers indicate placement of the cushions not only
vertically in different numbers along the car, but also horizontally at
the floor or between layers.
Host commercial types of pneumatic dunnage are ends from rubber.28)
One disposable type of bag is ends from paper-polyethylene:29) An inner
bag Iede of high-density polyethylene, approximately ‘0m11,thick and a

sturdy outer bag; six layers of 100 lb. kroft paper.

 

27) Interlake Steel Corporation, ACME STEEL Products, 135th Street 8

Perry Ave.. Chicago, Illinois.

28
) (l) ACRE Steal: Nylon reinforced, textured Neoprene.

(2) I. I. Goodrich Corporation. 3135 Euclid Ava.. Cleveland, Ohio
(distributed through Rubber Fabricators, Inc., Grantsville,
Heat Virginia): Rubber.

(3) RFD Comp. Ltd., Godalming, Surrey, England: Rubber.

(4) Firestone (5) Goodyear (6) U. 8. Rubber, Plastic
Product Division (Reference: Modern Haterials Handling,
June 1966, 1966 Directory. p. 286).

(7) TRELLEBORG's Gummifabriks Aktiebolag, Trelleborg, Schweden:
Butyl-rubber innerébag and Neoprene—polyamide outer texture.
9
2 ) Hanufactured by International Paper Company, 220 East 62 Street
new York, New‘York, Sale through lnterlake Steel Corporation.

Staticflpressurc

A test was conducted for the paper bag: (the company indicates
that the bag is able to withstand a pressure of 30 psi, but does not
say for how long a time). All tests at constant temperature and

humidity: 70°: and 502 an.

19/7-66 11 psi

21/7-66 0K

27/7-66 OK, inflated to 15 psi
l/8*66 exploded

TyPe of break:

It seemed remarkable that the bag could take the pressure of 15 psi
for four days before exploding. The break-starting point was the middle
of the front side in the long direction, from which it goes to the end
seams, where it breaks along the steel tubes (photograph, Figure 4).

Reasoning:

The strength of paper probably determined the moment of explosion
(non-stretchable paper was used). The polyethylene, stretchable to
about 2001, can be made slightly smaller than the paper, (if this is not
already the case) in order to prevent too much pressure on the paper.

The rubber manufacturers indicate that the bag should be able to
withstand 10 psi, under a load of about 8 psi. (Trelleborg) and in
free condition to not more than 6 psi (Trelleborg). These data seemed
more realistic than the data for paper bags, according to the test above.
It was remarkable, too, that the Trelleborg type cushion was reinforced
with a steel tube woven into the rubber at the middle in the long direction

of the front side, exactly where the explosion starts in the paper bag.

30

Rules of application

The American Association of Railroads (28), military authorities

(29) (30). as well as neutral literature (26), give rules and suggestions

for the application of inflated air cushions which are summarised in the

following:

(1) Used mostly for rigid types of loads, bricks. blocks. etc..

(2)

(3)

the pressure is 2-8 psi. For fragile items such as glass.
china. and fruits, use very low pressure and be especially
careful in loading the car (compressibility). For
semirigid goods like cans in fiberboard. determine the
pressure after "a rule of thumb"; weight of load behind
the cushion divided by the effective area between load

and cushion.30)
The main objective is to fill voids by using one bag for
each 2-12 in. of void with the center of the top of the

has slightly above the top of the load; and the bottom.of
the bog placed about one inch above the floor. An even
surfaca.nust be provided against the load (effective area).
The dunnago must be filled after being placed between loads.

For filling. it is, therefore, necessary with transportable

air-compression equipment, pressure gage, etc.

 

30)

The rule is used by AAR (37). It may be criticized because the
c-fector normally is 2-3 3, so that the pressure and compression
vill be large. On the other hand, the slow response in moderate
pressure air cushions accounts for a low amplification, so that
the approach udght be acceptable in most cases - also. it is very
practical.

31

leculatio§s_f r air spring; (31)

Air springs have low natural frequency combined with zero statical
permanent deflection. when combined with masses, the statements are
modified to some extent.

The behavior can be determined by looking upon the dunnage as a

cylinder having area A and length dimension 1.

d” J’

P? - Po V0 (gas law for adiabatic change)
where I, V, Po, Vo indicates pressure and volume after and before

deflection, and¢¥ is the ratio of specific heat CP I cv . 31)

— A’+'1 *
J'V ( ) . dV I dx , where x is deflection.

dr/dx-J'Povo
d? I dx - - A ; V - V0 - A - s
Stiffness k is the force per unit displacement:

-(J’+

k . A - up / dx - also A2 I v0- (1 - A): I v0) 1). which inplios cubic

elasticity.

Simplified, when the deflection is small:
2
i-J’roi Iva-lulu?-
Hatursl angular frequency and period is:

w-Vy7m- VJAg/Vo

 

rolz-‘fi‘velk-W‘Vvo/Iu

No damping will occur for an air spring since air is acting as an ideal
gas. By using a big volume V0 (by means of a connecting tank with a valve
release for a certain value of pressure), it is possible to obtain a soft
spring with an almost constant k-factor. A throttle valve in the air-line

may provide for an additional damping to the system.

 

31).
Mar different gases: Hz a 1.111 A z 1.67 He : 1.67 CO2 : 1.31
Atmospheric air: 1.6

ggperimcntglrwork and discussion

According to the rules of application. a trial cushion may not
extend beyond the top of the load.32) Therefore. cushions 18 x 30 in.
were used, which gave an effective area against the load of about
Ab- 200 in.2 (area of 2 rowm wooden blocks is 2 r 5.5 r 30 . 330 in.2;
but the shape of the pillow’will prevent it from contact at the whole
area). IRELEBORG was the only coupany responding to the request for
cushions in that special size.

The simple formula for to I 2 hasdu>consequence, that duration is
independent of pressure. The formula does not hold. The area becomes
greater during the impact. though with a lesser increase at higher
pressure (where deflection is less). Therefore by higher pressure,
average A is relatively small; and duration should be relatively high.
The experiments showed the apposite (Tables 9 and 10).

how we consider the more complicated expression for k which accounts
for the increase in k, when travel is not negligible; in fact, travel in
this case is about 4-2 in. at 2-5 psi.

With a thickness (void) of 8.5 in., the pillow is approximately a
cylinder. which volume is Vo - 3300 in.3

(77 0 diameter - 2 . width gives W‘o d - 2.18;

d-12:A1-/7I£ ~d2-110390-30 -110).
the factor: >
' -(?+1) -2.4
(l-A‘x/Vo) -(1-2-2oo/3300) -l.35forx-Zin.

" e‘
and (l - 6 ‘ 200 I 3300) 2 - 1.75 for r C Q in.

shows an expectedly lower k at higher pressure, where deflection is small.

 

32)
What happens, if it does, may be seen from Figure A.

33

The main factor <r Po A2 I Vo will contribute to a higher k, (an
increase of 70: from 3 to 5 psi), whereas the decrement from the other
factor is only about 252.

Total calculation at 5 psi as follows:

to - 1.4 . s - 2002 I 3300 - as lbs/in.

II: II .35 ' 1.35
max

115 lbs/in.

k

.v 100 lbs/in.

To / 2 . WT” - VSBB‘TSE'E’ ' 10!) - 330 use:
"a” is used as the total mass behind the cushion in position 2, and
comparison in duration might be to the load cell in position 1; in the
back end of the car. the duration will be zero according to the same
sort of calculation. Nou'the coupling effect between 2‘+ 12 cushions
will even out the difference, and we noticed the m of To I 2 front
and To I 2 hack to be approximately 260 msec (Table 9). only 22! below

calculated value. The "qualified" average between front and back

 

duration is \Ik1302+-1301) I 2 - 130 msec; which is only 552 of the
calculated value.
f Conclusivaly, our model is far too simple. We cannot neglect:

(l) the reaction from the backstop as a stiffening factor that sometimes
overlaps the main pulse transmitted.

(2) the friction forces in the load, which probably would give longer
pulses (see Tables 9 and 10, cardboard trials).

(3) the multiple degree of freedonvaystem that are involved, including
not only harmonic forces, but also friction forces that are

extremely difficult to treat analytically in a multi~degree of

freedom system.

34

(4) the influence of distributed mass, and last but not least:

(5) our model includes, to some extent, the odd shape of the cushion,
although we had to assume constant A. But. the elastic forces
from the rubber material, when working together with the air-
cushioa, is not considered - thus we assume an additional
stiffening factor. not from the point of impact since the load is
incompressible. but from the places wherethe cushion is not
loaded and the air therefore is able to expand. To completely
investigate that effect would require measurements of deflection
and forces in different directions of the inflated rubber air"
cushion in the load, and the effect depends upon the compressi~
bility of the load. Before such a study of restraint-influence
on duration - and mainly on forces ~ is undertaken, it has to be
remeabered that in actual use the fraction of the cushion area
free from the load to the cushion area under the load in l333_
than in our simulation. If we want to investigate these effects

at the 1/4 scale railroad, we therefore ought to adjust to the

area free
area—loaded '

 

right proportion of Beforehand. it is unknown if
a restraint will occur, or if the rubber material will correspond
to the normally involved dynamic impact forces and durations so

that no additional factor of importance will occur.

Tables aggwresult§_

Tables 7. 8, 9 and 10 are attempts to describe the observed pulses
under the indicated conditions. The vibrations could have been described
by use of the spectral density concept and probability distributions.
especially if the reproduction was refined by using the oscillograph ~
records instead of just photographs. However, the descriptions -
preferably held together. with the corresponding pictures in Appendix VI -
should support the findings that havs already been indicated.

Generally the dynamic input forces are diminished with at least 5212,33)
when air cushions are used in the front and of the car. We noticed that
forces do not necessarily decrease along the car. At positions for from
tho cushions with wooden blocks at both sides (for example. 2 + 12,
Position 5), the input force nay even be about 602 of the input force
without cushions. which is about twice as such as in the front position
(2 4' 12, Position 1).

Usually. there is no clearly directed influence on the peak force
of either pressure or the number of cushions as long as the volume of the
air cushion is maintained. It is believed that with other types of loads,
and especially when internal friction is brought do: control. we might
be able to see such a difference. .

the self-excited vibrations due to the floating load of non-
surface-crested wooden blocks that slide against each other producing

friction forces with randomly varying friction coefficients will also

1000 lbs. (Table 8) used in the denominator, probablyrclflz higher,
due to load hold-down blocks adding to the weight.

36

deterline the obtainsble peak forces: however. the phenomena ere not
easily described by differential equations. Even slight nonlinearity
in the terms of the equation of movement will influence the solution
considerably”) r

The experimental evidence for self-excited vibrations can be seen
trots the last four trials in Table 10 (with cardboard between rows of
wooden blocks) compared to trials without cardboard; otherwise conditions
are the seas as in Table 9.35)

Also, when there are no wooden blocks to move. as in 2 + 12.

Position 1, there are practically no vibrations. The vibrations from
othc parts of the car are being damped due to the cushion in Position 2,
except for a high static pressure which gives high support for
oscillations (2 + 12, Position 1. 5 psi) .'

Couplig is said to occur when there is a clear connection between
the values of peak force in the front and in the back of the car; that is.
when the value of corrupondins load cells are approachim each other
relatively and with significance, when goirg from one pressure to the next
highest.

Peak forces, values in parenthesis, are static pressures exerted by
the air cushions. Vibrations are being maintained partly due to high

levels of static pressure, becmse oscillations then will take place

around the static level represented by the cushion itself which has a

3‘) ,
Jacobsen In Ayre: ”Erginserins Vibrations," 1958, pp. 274-77.

5 .
3 ) It is a usual thing to place cardboard in the bottom of railway

cars before loading (32).

37

negligible mass. The static pressure has been measured once at 4.5 psi:
pillows in position 1 + 7 + 13. load cell in Position 4: 530 lbs.:
load cell in Position 10: 390 lbs. Average 410 lbs. corresponding to
an effective area of £10 I 6.5 - Min each of the two rows.
Durations, values in parenthesis, are rebound durations in
sequence indicated without letter. If a letter 1 is indicated,
severe vibrations (up to magnitude of approximately the static pressure)
occur; if a letter 1 is indicated. vibrations occur both before and
after the nain pulse; if letter _i_, the vibrations clearly super-impose
the main pulse; it the letter ldoes not occur in parenthesis. the
pulse cannot be diatimuished from vibrations; 11' letter (w) or (v).

inside another parenthesis, only medium vibrations occur.

38

 

 

 

 

 

 

 

 

Table 7. Input to Lading, Front End of Car
TONI Peak Force 1' Duration F [1’ Picture
Position Load (Paired Value) t . m'm ‘ " Reference
Lbs. Lbs. X) xx) 0

1 795 180 100 0.03 —-

2 - 330 90 0.11 78

3 - 620 95 0.1!: 79

b - 920 95 0.17 80

5 - 1250 (+90 st.) 100 0.20 81
---—qy---4$--------- ----- ——-----o -----

1 1020 350 140 0.12 82

2 - 1.30 130 0.12 83

3 - 620 (560) 120 0.12 (0.11) 8!.

6 - 820 (700) 115 0.13 (0.11) 85
- - - .. 4» .. .. .. .1 .................... .1 .....

1 1200 70 160 0.02 86

2 - 330 140 0.03 87

3 - 600 120 0.10 88

4 - 620 110 0.09 89
r)

or induced from pneumatic dunnage.

xx)

Otherwise fair-ed .

(+st.) iniicatea a static load left over from the preceding impact

“whats in parenthesis indicate the lower peak, when two peaks of
about 20 nsec in between occur.

39

Table 8. Forces Along and Upside-Down the Car

 

 

Position ‘ 960 Lb" _ 190 Lbs.
1050 1.12..
Total Load ”5 “m“ 85 use
1030 . 9 . .
(Picture. 1"” i 0 Lbs (300 Lbs 1: mac)
fin. 7" 75’ 76)- 110 ”“3 (reaction from bulk
head)

 

 

 

 

Comments to Tables 1 and 8

The percentagelof transmitted load to the ladies is most
consistent in the 1020 load area. where it is eeaaured in the Eggglg,
£233 In the lgggg.row, it varies from 8 - 202. higher by higher
severety of the impact; this is consistent with some indications in
part A of higher amplification factors by higher load factors: The
lower row is note close to the floor and the impact force. The 12!
corresponds fairly close to the values in.the bag-damage-sinulation
tests (7).

1n the upper row, the transmission factor is a little lower than
in the middle row, and from Position 1 of the incline the response is
very small.

The decrease of the nagndtude of the main pulse along the car
fron.;ggg_1bs. to 200 lbs. and the time delay of the main pulse. about
100 n 8. is a characteristic of the particular lading.

Because of the relatively small forces in the back end of the car
and the additional blurred pulse, that the rear bulk-head introduces.
it seems advisable always to use the front end of the car for damage

simulations.

(00

Table 9. Pulse-Characteristic for inmatic Dunne-go - Void 8 1/2 Inch

Haintained. Position 5 ~1050 Lbs. Total Load.

 

 

0 hi Load Prea- Peak Force Coup~ Pi
P u: 1°” Cell sure (Paired Value) Duration ling Velocity R Icture
as t on Poa- psi Lbs. (t ) msec é Check 9 erence
itio 0 xx)
2+12 l S 27!) (+3650) 130 (+120) + 1229 53. SA
(1 100v)

" 5 5 490 (+s650) 120 (+200?) GD 1174': 51, 52

" s s 180 (+uso) 120 (+2009) (:3 1174 51. 52

" l3 5 140 (+1650) 130 (+110) + 1229 53, 54

(i 50v)

u— - - and u- - 4} - e- c. ——————— H- ...... — s— - h - e- c e- —————
2+12 1 - - - - - -

" 5 I. 590 (+3360) 140 («0 (ED 1227 55

" 8 l: 200 (+s360) 30 (+14) GD 1227 SS

" 13 - - - - - —
as s- eu d o a. d)- :— o - ------- 1b ------ us- an — ..........
2+12 1 3 330 (+9270) 160 (+70) + 1228 56

N S - _ _ __ _ -

I! 8 _ - _ - _ ~

" 13 3 100 («270) 150 (+100v) + 1223 56

- - as u s. — dh .- e- - ------- he uuuuuu use I- - F u s- e- d .....
2+12 l 2 320 (+s 180) 150 + 1225 57. 58

i! s _ _ __ - _ _

fl 8 _ __ __ _ __ _

" 13 2 7o (+3180) 150 (+5w) + 1225 57. 53

.- es- .- T .- - 4P .- c- e.- aaaaaaa b ----- .- ... .- u. e- - e- e- and, .....
2+12 1 l 620 (4190) 2?.0 (3 1160 59

" 13 l 50 (4190) ~(+30v) GD 1160 59

 

 

 

 

 

 

 

 

61

Table 9. Contimed.

 

 

 

 

 

 

 

 

 

, Load Pras- Peak Force Coup-J
Cushion Cell sure (Paired Value) Duration pita Velocity Picture
Position Pos- psi Lbs. (t ) usec + Check Reference
ition ° 8 xx)
2+12 1 0 370 220 C3 1224 60
xxx)
” 13 0 .30 - Q 1224 60
e- - - .- 1b - so an e- a- - d} -------- b ————— d - O h- - .- O c- - u- e- u. e-
7 1 S 420 (41650) 120 (+60) + 1238 61
" ' 13 s 170 (M50) sway.) + 1238 51
- - use - «IL - - u a» e- - 4 ........ D a- - - a - all as- as L- ----- ee - — e-
7 1 I. 490 (+- 360) 120 (+60) - ' 1231 62
" 13 4 150 (+- 360) 9o"(1v.) - 1231 62
----4b------+D-OQODGC‘FQQ-thuulr--------.
7 1 3 440 (+s270) 160 (+30) - 1226 63
‘ (iv.)

" 13 3 150 (+s270) 70 v - 1226 63
----1+------qus---‘--1hu-ue--~--(L ..... ----
7 1 2 680 (+s180) 180 (+50) - 1228 66

(iv .)

" 13 2 80 (41180) 70 v - 1228 64
s- .— C- — 4b a- c- - - .- - J ------- 1p- ------ s- — 1h «- e- as. - ... r— .- - .-

7 l 1 590 (+390) 150 - 1232 65

" l3 1 80 (+890) 50 v - 1232 65
Qua—d}--—--- —------(b—----d--qb-----bee—-
l+7+l3 lo 5 670 (M50) 120 (+90+60) + 1150 72

" 10 5 170 (41450) 130 (+20 v) + 1150 72
Position
0 r.“ 1 2 3 6 5 6 7 8 9 10 11 12 13

 

 

Key for Position flashers

d The first reaction from the backstOp not indicated (only a few

vibrations and magnitude about the same as for the transmitted slain pulse).

leadirc at counter in 10"
value, the lower velocity.

Void fill-out is maintained and valve closed.

xx) see. by constant distance, that is: the higher

xxx)

62

“table 10. Pulse Characteristics for Pneumatic Dunnaxe ~ Variation of
Void, Check 2.. Rubber Material, Impact Chock. Paper Bag, and Cardboard

Material in Load .

 

 

 

 

 

 

Load
611st Cell Prea- Void. ) Peak Velo- Pic-
Position oaition sure in. 3: Force Duration city ture
7 1 4 71/2 070 (41350) 160 (+200) 1220 66
, (1.12.)
7 13 6 7 1/2 120 (41360) 50 (i.v.) 1220 66
7 1 lo 11 530 (+1360) 200 (+80) 1236 67
7 13 6 11 100 (H360) 23 (i .v.) 1236 67
unocqybcnocoundu---1p-----~-—-uuuqens-s”-.-
7 l 0 0 1320 120 1138 68
(rubber)
7 13 0 O ‘ 90 30 1158 68
”(alone)
a---d------%%-m--{bfifi--u- ...... ‘-'J--
2+12 1 3 8 1/2 300 (+a270) 160 1181 69
2+12 0 l 3 8 1/2 5200 130 1181 69
(impact)
Dwnodubudnnudwu-fllh ...... —~~-—-q--Jru~
7 s «2 WIt) 580 170 1231 70
(paper has
7 8 ~2 ”/0 ~ 0 ~ 0 1231 70
large)
----d--"---*----Tb- ..... clan--e-iocedLQ-e
2+12 5 5 8 1’2 500 (+3050) 110 (+80) 1236 71
(150 (V) l '
) 312 8 5 O 1/2 170 (41450) 90 (+ (10)) 1234 71
xx
1+ 7+13 4 0 1/2 a 1/2 370 (+3410) 100(+100+70 1236 13
1+ H13 10 0 1/2 31/2 150 («010) 110 0100.. 1236 73

 

 

 

 

 

1’ 7 1/2 in. is about the maxim compression possible with a rope tie-

down at the particular imact; neither steel-straps nor plastic-
straps could hold the load down.

1:) Cardboard, coated type, is placed between rows no. 1 and 2 in addition
to the C-flute corrugated in the bottom of the car.

PART C

REPMKS ON SIMULATION AND REPRODUCTION

OF DAFSACE

4'3

No

As a matter of fact, very little infatuation was available about
the correlation between the input characteristics and the damage on
particular goods caused by the input. The bag simulation, mentioned
before, is one of the few attempts. (7)

In order to simulate damage in the laboratory, it is necessary
to know the damage potential of the input compared to practice. The
attempt in the present study has been devoted to
A. Producing well defined impact conditions and pulses that in duration

and travel are comparable to practice.

B. {ensuring the relationships between the impact and the input to a

certain type of load.

Simulation of forces

The shortcomings of the laboratory-type simulation night he
first of all:

(1) the lack of force that in sire-order would be similar

to practice.

(2) the lack of reproducibility of vibrations and friction

which are also involved in practical operations.

Now concentrating on the ahortcmning (l) and thereby restricting
ourselves to discuss matters of importance for goods that are 32$
vibration~aenait ive (which we do not know too each about) . we have
produced input forces at about m. total in the front end of the
car. It is 2000 I 330 . 9.3.1:. dymic force (in fact we have no
intonation regarding how the input force is distributed over the. cross-
direction of the car, neither in practice nor in the laboratory; at

present, we lust assure even distribution of forces horizontally across

65

the car). In practice. input forces to the lsding (though a lading of
bags. probably giving less peak force than voodsn blocks) have been
Inasured to 1809 lbs. total force with a cross section area of 6x5' -
1300 in.2. is 1300 I 0300 - .92 2.1 dynamic force (for bags).

In this respect, too, our sinulstion seena fairly good. if the
causes for damage really are unidirectional forces centrally located and

evenly distributed over each container. How is this in practice?

Simulatigngf dsnagg_
vs decided to furthsr investigate the canned goods; namely. carbon-
ated juices in cans. According to (33). this type of goods is one of the
nest susceptible to damage; about 11! of revenues are paid back in clsins.
whereas the overall average is only 12. Causes for dasage are indicated
as follows:
Poor arrange-ant of the load 25!
Improper handling of the car in transit 232
Shift in loading due to a loose lead 151
Unfortunately. these firures suggest bad loading as the Iain source for
damage. In other vords, they refer to eaterial handling sathods. not to
packages. However. the figures isply that internal shift in lading (in
addition to forces acting on an integral type of load) are causing the
damage. Internal shift is difficult to sinulste in a laboratory device
vhere vs do not have the height. nunber of layers and packages necessary
to produce a rsndou.shift in different directions.
In a ray, gggpg_of packages in the car seen to be the cause of much
of the dosage. This brings us to the question. would drop-testing be

sufficient for investigation of damage? In fact. the forces acting in

66

vertical impacts (switching) have the suns sire-order as forces acting
in vertical impacts (drops) .36) The answer night he denying, because
the durations are really different (about 100 nsec versus 4 10 msec).
resultim from the long travel of draft gears. Therefore, elements of
the laditg, that would not be involved by a drop test because

Isplification factors (due to low value of eleggnt natural fregucncy)
Hput natural Fequency

will be <41, night be involved in impact tests with the approximately
10 tines as low "input natural frequency."

Also the horizontal shift in ladirg that produces dosage by a
second impact against other lading or the ends of the car ought to be
studied in a proper device with room for distribution of load. devices
for neasurim influence of friction, etc .. although it night he studied
analytically to sons extent (34). A drop-testiIg-device will not suffice
for investigations of that kind.

(1) agect simulation involves the use of the test itcn (package)
in the front end of the car with a rigid and unsovsble mass
behind it. The mass behind may consist of the same type of
items (packages) or a dummy load.

(2) Indirect simulation takes into account the smaller forces
due to Isaller nssses behind in the simulation than in
practice, by adjusting the test items (package). If it is
found, for ample, in a load of has that the input force

is only one half in the simulation in the front of the car

 

36
’ For ample, reference (24) (pp. 26~28) uses a Neoprene rubber spring

sinilar to ours, and a mass of 227 lbs. impacting with a velocity
of 50 inlsec. gives a peak of 38 g's, it is 8-9000 lbs. (comared
to Table 6: 3-5000 lbs.).

47

as it is in practice (in the position of a full scale car
that is under investigation). than the knowledge of package
strength as a function of the sire of the package. the
thickness of bag Interial, etc.. will make it possible to
adjust the package to the forces occurring in simulation
(for example. to make a bag with twice the dimensions of the
package in practice).

Maybe the only simple example describing that is a
wooden or plywood box which can be treated according to the
been concept (35); say a wooden board of length L. thickness I,

width D:
8 (Bending Stress) -

(Bandit; Moment) H 0 ‘i‘ I 2 ’ I Gin-ant of Inertia) -
1. ' L ’ T I 2 ' 12 F t L 3

a 01-3 m1 s

 

 

In terms of minu- deflection:

.2 2.
r 5 III. !.L 00

‘ I

 

 

max'“
33s r°r 0134-384

where I is the nodules of elasticity.

In order to perforn a simulation we choose to build a nodal

2 ' L of the original container. with the other dimensions
unchanged, the stress for the test container will be the same
by a simulation peak force of 1., as it is for the original
container in transportation with a peak force of 2 ' fl. etc.
The static concept could be codified by the influence of the
amplification factors. To check this the natural frequency

for the been

48

 

I ‘ 5
fn c Vhlg/le.

C . 3.56 for fixed beam may be calculated (25)
and compared to the input force frequency.

Friction forces that have been brought under control and
fulfill the simple equations (for dry friction: ’fr'- fL' V,
VhflfC/A is a constant friction coefficient, and w is the weight
component rectangular to the sliding surfaces), may be simulated
separately. Say the weight in vertical direction above the car
floor is only lid in the simulation than it is in practice
(possible height l/A of full scale, same type of commodity up
and down in the carload), then by choosing four times as high
friction coefficients as in practice between the load and floor
(choose a fairly rough or “sticky” floor material), the friction
force will he the same. It is noticed, that it would be
impossible to combine the latter mnthod with the former

dimension-adjusting method in one test.

The case of resilient lading_

(a) If the lading or parts of it have space for moving in the horizontal

direction (free slack), the input velocity to the lading relative

to the car is V - V1 +'ka (Law of inertia, friction 0). In practice,

since the coupling between cars will stop the striking car immediately

37)

after impact, the input velocity is only Vt or it might be Va

 

37)

The statement is slightly modified by the fact, that if only a few
of free slack is available, that is if the lading will already,
during the first half part of the impact move sufficiently to hit
other lading or bulk head (V I 2 x t I 3.5 in. for V - 70 in/sec.
and t - 0.1 sec.) then the input velocity is less the V even at
friction sero, but in that case the damage potential is ow too.

(b)

(c)

49

for slack in the other end of the car or the unit and is then
directed in the opposite direction. It was noticed then, that
although impact velocities are the same in the simulation as in
practice, then with the same relationship Va I V1,

(namelyrow37 I 61 - 602, Appendix V compared to 532 measured in
practice (11)) the impact simulation is more severe than impact

in practice for slack in the front end of the car or the unit.
whereas slack in the back end has no effect. we may be able to
adjust for this by changing to a lower impact velocity at the same

38) But the best

time considering forces involved, when possible.
solution would be to build the test track to include a coupling
device similar to practice.

If the load is to some extent floating or resilient, the input

velocity may be something between VR and V1‘+ VR, eventually 0,

y and we would have a very bad simulation, when friction has to be

taken into consideration (Part B).

we reached the conclusion that our work is successful only in
thecase of a rigid load, otherwise coupling, etc. will have to be
included on the test track in order to perfore a direct or

indirect simulation procedure.

Different types of friction forces occur.

Experimental work and calculations
A fiberboard box 27 x 5.5 x 19 in. (L r H'x D), 8~flute, 200 lbs.
Aullen), double faced, REC, containing 40 cans, tinecoated, 612 x 210,
each containing 12 oz. HAVIS Club Sodaeboverages. Soldered, seamed ends,
3/8 in. space above liquid. Total weight: 35 lhs. Taped over the flaps.
The package was exposed to impacts, V1 - 7O in/sec.
H D 975 lbs. (Fm O 6000 lbs. for wooden block load,
Fa - 1700 lbs. in front position)ard higher (table ll).
No damage except for small indentations under the lid occurred (cans
turned around occasionally. so that new indentations could he observed),
although dynamic forces as high as 1800 I 27.18 - §_p§i_were encountered.

Because we were unable to reproduce damage in the box, cans were
taken out and exposed to impacts in three directions (table 11). Some
authors indicate strength of cans (static) to about 1000 psf - nggi.(6)
and here tun cars with a top area of 2 r 5.5 in.2 were exposed to a force
of about 850 lbs.; 850 I ll - 21m33i_(if a pulse shape similar to the
earlier observed pulses are assumed) without a break.

It is quite complicated to consider the strength of the can analyti-
cally. It involves both elastic, plastic and viscous properties for a
rather complex item. However, the maximum force acting on the end of the
filled can may be calculated approximately by assuming that the can is
acting as a liquid spring, without damping (adiabatic), where the force
rises very fast to a constant value. The law of conservation of energy

applied internally in the load then gives:

FZ3S I l/2 M V12 , where

F is the constant reaction force from the can,

438 is the corresponding maximum deflection of the can,

( - displacement of the row of wooden blocks relative
to the car),

H is the mass of the row of wooden blocks, and

V1 is the total decrease in the velocity of the wooden
blocks during the impact, relative to the car.

If the time for total deflection of the can is not too much
smaller than the rise time for the input pulse (it can hardly be
believed, that it is longer), V1 is the cane as the measured impact
velocity, and friction forces will hardly be able to brake the movement
at a distance of 3/3 in. From the trials (1 can), it seems as if the

minimum breaking point is about V1 - 7O in/sec., and than we get
I ' 3/16 ' 1/2 ' 205 I 336 ' 702; F C 7000 lbl.; P . 700 I 6 ' 117 281.

The friction force involved and assumed constant means that P - 'can+ ’fr;
with cardboard underneath, the friction coefficient is less than 1;
Pfr<1200 lbs.; only slight difference in the calculation.vill therefore
occur.

Impact from the side gives l/B in. deflection at 70 ln/sec.. that
means an area of 9 in.2 for the two cans in the trial gives the (constant)
force F' l/8 - 1/2 205 I 386 ' 702; P - 16,400 lbs.: P - llg psi.

The present example is only meant as a rough calculated case
demonstrating that the present test track can be used to produce damage
and data relating to it, although it requires a lot of force even to
break cans on the test track. The observation supports that other causes

for can-breakage during railroad transportation are more important than

unidirectional forces.

mu 1 1 .

52

Insets on Lung Including Cons.

 

 

Pootttoo of ' W of
Lou Ionic-o boon Run-to
Lbs. ‘Cuobloo
975 7 (box) 1 -
13 (box) 2 -
1 (box) 3 mtmn intimation: odor
tho 11:! Mo con to mum
Ina—cob uuuuuuuuu u. ----- -.?Ehs.°=':-------c—
1175 1 (box) 2 Mining Mutations undo:
tho 11¢! Ihoro m to touch“:
-----P-—------- ------- .?£h:r-c:nzounuocncu
1050 4' (box) 5 ..
S 112 poi. 1.13 (cushion)
cotton 3) 1 (box) 5 -
1.13 (en-Mon)
12:0 3+4. (box) "’ s -
5 1/2 pot 1.13 («.111»)
1020 1. 1m and 4 (10 cm) Bo d‘qo up to 70 ill/us.
1. from ond lo ( 2 cono) No daugo up to 70 til/boo.
1, from and A ( 1 con) At 70 1nlooc. omhod 3/16 in.
ad looking o 113:1»
1020 1, fro. old: 3 ( 2 cm) A: 70 inlooc. 1/8 to

 

 

eta-proud in flu: bloc.
(2) o 11ttlo noro
(3) no furthor contusion.

 

 

Tabls ll. Continued.

53

 

 

Position of Number of
Load Box/Can Impacts Remarks
Lbs. Cushion
780 l (40 in/sec) 1 (1 can) alight impression,
1/16 in.
(Ono loysr of 1 (52 in/sec) 1 (1 can) 1/8 in. impression
wooden (measured across).
blocks
rooting on 1 (63 in/ssc)‘ 1 (1 can) 7/3 in. impression, leak.
csrdbosrd)
1 (70 in/ssc) l (1 can.) Total break and leak
diagonally
xxx)

 

 

 

 

X)

8tst1c prosours exerted for 1 hour before dynmic test.
2 gsi static pressurs on box.

Effective area:

xx)

96/270 gives

Box plsccd horizontally in the middlo row.

Special jig secured the can in.£5° position (Figure 8).

54

 

Figure 8

Cans after exposure to impact. Shown from left to right: impact from
the side, diagonally (3 stages) and from the end.

Jig used for diagonal testing of cans shown in place in the front end of
the car.

6. SUMMARY AND CONCLUSION. SWESTIOflS

with the objective of predicting damage potential caused by
humpirg and switching operations performed in practice on railway
freight cars, a laboratory scale railway test track with car and
lading was investigated as the model system.

Being interested in an output (readings concerned with damage)
from the model sflmilar to, or with a simple calculatory relationship
to the real system during Operation. we tried to compare the components
of the input to the two systems.

Impact coalitions. input—forces to the lading and self—excited
movements in the lading all constitute factors, that must be assumed
to be important. with a lack of comparable information on the detailed
behavior of the lading as a functiontof the switching Operations alone,
our attempt was to discuss and make suggestions in order to adjust the
model from an environnental analysis-standpoint, involving parameters

such as

impact velocity

magnitude, shape and duration of impact forces

impact quantum (equivalent mass of striked cars)

rebound velocity

movement during and after impact

magnitude and duration of input forces.
Having reached the input to the lading quite successfully, we
proposed dirwt and indirect simlation procedures using items or model

items from which it should be possible to describe damage. Correlation

to practice is practically unknown, and parameters such as
- superimposed vibrations on impact

- self-excited randou‘vibrations in the
load caused by internal sliding, etc.

seemed impossible to simulate in the laboratory. But we still suggest
that proposed procedures - after preper adaptation of the test track -
such as:

— securing strictly horizontal impact and
movement of all outer components

- rebuilding backstop to perform as 3 or 4
striked cars in actual operations,
preferably with no rebound travel
- reinforce construction and hauling devices
to operate at least masses of 1200 lbs.
securely
will make a valid prediction of damage possible for a wide range of
commodities being transported by the railroad.

Although it might be outside the direct scope of the present work,
it is suggested that trials including the analysis of damage potential
be made on full scale test tracks also, because this is believed to be
the only way to steadily adjust a laboratory scale device to account for
special phenomena and new developments in railroad transportation (7).

Air—cushions included to divide lading into sections were
investigated and showed up to have a decreasing effect on the magnitude
of input forces to the lading. Further investigations are suggested
including other types of loads than wooden blocks, preferably with a
controlled type of internal friction, using unitized loads and smoother

sliding surfaces. Also the effect of unitizirg, bracing and fastening

means (36) (37) such as steel accessories, non~skid plates, anchor plates

57

and retarder plates in connection with steel strapping, and vooden type
dividers can be investigated and compared to air-cushions.

An attempt to reproduce the damage of canned goods in fiberboard
boxes were not successful. but damage was obtained by exposing one or

two cans to the maximum obtainable input-forces.

(1)

"(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

7- BIBLIOGRAPHY

The Chesapeake and Ohio Railway Company, ”1M Scale Impact Test
Plant,” Research geport lb, 60, November 23, 1964.

American Society for Testing and Materials, "Incline Impact Test
for Shipping Containers, Tentative Method of,” 5??“ D 392::
63T, 1963.

Jacobean, Pails, Unpublished data from course work, PKG 623,
HSU Packaging SchoolL LAB Project 1&3le Spring 1966.

I-bnterey Research Laboratory, Inc. "t-iodular Impact Shock Testing
Systems“ (commercial information.)

at B. Hiner, Inc. "Roller Coaster for Testing Freight Cars,"
Illinois Central Magazim, February 1964, pp. 6-7
(special print.)

Paterson, Hillian 8., "Cushioning Requirements for Adequate Lading
Protection," £13m: Paper No,‘59-A-3l_2, Nov.,Dec. 1959.

Patterson, 0., Jr., and P. R. Lantos, "Rail Car Impact Simulator,"
2, I, du Pont de Nemaurs S Coupon}, 1963. (AS‘I‘M SIP-326).

Rector, R. 3., ”Some Shock Spectra Comparisons between the LII-Ix 600
Series Railroad Cars and a Railroad Switching Shock Test
Facility, "SandiawCorporation, Shock, Vibration and Associated
Environments, Bulletin R0. 30, Part III, February 1962,

PP. 1384‘s

Association of American Railroads, ”Approved Draft Gears for
Alternate Standard 36 in. Draft Pocket for Freight Service."
Qacification 21-901 C~63L 1963.

"Approved Draft Gears Having a Minimum Capacity of 18,000 Foot
Pounds at a Reaction of 500,000 Pounds for Freight Service.”
SpecificationJi-fiml D~62, 1962.

”Approved Draft Gears Having a Minimum Capacity of 36,000 foot
Pounds at a Reaction of 500,000 Pounds for Freight Service,"

Sggcification 14—201 2-62, 1962.

"Approved Draft Gears to be Rated by Impact Testing for Freight
Service," Specification H-901 0-6!“ 1966.

"Special Cushioning Devices for Freight Cara,” Spegijication 24-921-
§_§, 1965

Baillie, lhllace 3., "Impact as Related to Freight Car and Lading
Damage," ASHE: Papa; No. Mull—2433J Hov.,Dec. 1959.

R?»

S9

(11) Roggsveen, R. C., "Analog-Computer Simulations of End Impact of
Railmys Cars,” ASP-ill: Paper No. 65-RR-3, April 1965.

(12) Newcomer, G. R., "Study of Vibration Frequencies under Impact
Conditions," ASHE: Pager Ho. 594-239, Novnbec. 1959.

(13) walsh, J. P. and R. E. Blake, "The equivalent static accelera-
tions of shock motions," Proc. Soc. Fag), Stress Analysis,
Vol. 6, Me 2, PP. 150-58 (1949).

(16) Kuoppanaki, K., R. A. Rouchon, "Aerospace Shock Test Specified
and Monitored by the Response Spectrum,” .52.st and
Vibration Bulletin Ho. 3'5, Part 6, April 1966, pp. 163-42.

(15) Ostergrean, S. 3., "Shock Testirs to Shock aaectra Specifications,"
Shock and Vibration Bulletin No. 31, Part 6, April 1966,
PP. 155‘96e

(l6) Cartel, M., R. Holland, "Definition of Shock Design and Test
Criteria using Smock and Fourier Spectra of Transient
Environments," S. and V. Bulletin No. 35, Part 6, April 1966,
PP. 2‘9‘26‘e

(17) Painter, 6. 11., and B. J. Parry, "A continuous frequency Constant
0 Shock Spectrum Analyzer,“ A. and V. Bulletin No. 35,Part 4,
February 1966, pp. 129-36.

(18) Department of the Army, 'Transportability Criteria Shock and
Vibration,” TB 55-100, April 1966.

(19) National Aeronautics and Space Administration (MSA), "Transport-
ation, Shock and Vibration, Design Criteria Manual,“ MR 1262,
September 1965, VOIe 1. Po 6‘s

(20) Vigness, Irwin, ”Specification of Acceleration Pulses for Shock
Tests," 3. and VL Bulletin, No. 35, Part 6, April 1966,
pp. 173-83.

(21) Rooks, R. 0., "Shock Test Methods versus Shock Test Specifications,"

(22) Vigness, 1., and E. W. Clements, "Sawtooth and Half-Sine Shock
Machine for Medium Weight Equipment," US Naval Research
Laboratorz ml. Report 5%}J June 3, 1963.

(23) Cross, U. 11., and Max. McWhirter, “Shock Tester for Shipping
Containers,” AS‘I‘! Special Technical Pulletin F . 176, 1955.
(Symposium on Impact Testing), pp. 161-63.

(th) h’ooks, Richard 0., "Shock Sprims and Pulse Shaping on Impact
Shock Machines," S. and V. Bulletin, Ko. 35, Part 6, April
1966, pp. 23-so.

(25)

(26)

(27)

(28)

(29)

(30)

(31)

(32)

(33)

(34)

(35)

(36)

(37)

60

SAZDIA Corporation, "Standard‘Enviromnental Test Methods,"

liensch, Ing. K., “Transportschutzkissen, vorteil‘naft und
rationell," Die ncue Vcrpaclmng, Fcbr. 1965, pp. 143-44.

Anon., Railwnl Age, May 31, 1965, p. 120.

Association of American Railroads, "Rules Regulating the Safe
Loading of Freight in Closed Cars," 53.32 Pamphlet Mo. 16,
December 1963, p. 35.

Quartermaster Corps, U.S. Army, "Pneumatic Dunnage," =101m?-
fillitary-Industrj Symposiula Fort Lee, Vimi‘nia, October 1957.

 

Joint Military Packaging Training Center, "Packirg and
Carloading," Special Text. BIS—F2 (J1), January 1966, p. 8.31.

Thomson, William T., "Vibration Theory and. Applications,"
Rentice 11311, Inc., 1965, pp. 67-68.

 

Peterson, William 11., "Dynamic Behavior of Lading in Freight Cars,"
Agni Conference on P'Iethods for Evaluating Lading, Storage
and BracinyLPractices, Uccobet 3, 1962, p. 18.

 

 

Transportation and Packing Survey, Final Report, Oct. 14, 1952
(1415 K Street NAIL, Washington 5, D.C.), pp. 27, 63, 83.

Sela, Dr. Ing., "Transporbeanspruhungen und Verpackungspruei‘ung,"
018 “8:28 Vergackurg, hOe 6’ 1958’ PP. ‘92“5‘34.

have, Kenneth, "Package Design Ergineering,’ John Wiley 5 Sons,

Inca, 1959’ pp. 48’ 52. 319

Sherrick, Janes H. "Flcncible Tie-Downs for Railroad-lading,"
ASHE: Paper Izo. Gian-237, 1961, p. 5.

Anon., "Outline of Loading :dethods and Marginal Bracing Conditions,"
AAR lio. 15?3&,_internnl (Courtesy of Mr. Burt Hilliams,
Association of American Railroads).

1.

APPENJIX I
hpact Velocity Versus Position on Incline,

Theoretical and Heasured: Uncertainties Involved

Theoretical impact velocity:

 

v“ - VZg'ZNn hn/U
dtived from the law of conservation of energy:

-1/2w/gv z-hw+hw‘+hl=l,+hw+hw

2
”2’” 1c 11 22 33 44 55

it

ill - Height of load; 11- corretponding height - height at the
floor level (of center of gravity).
1.12 I Weight of rear bulk head, 135 11111., hz as above.

113 - Weight of front bulk head, 135 lbs., 113 as above.
1.". II Weight of car floor with steel-angles, 150 lbs.

Us 0 Height of wheels and suspensions, 80 lbs.

The weights are figured out from the information given in (1).
The heights of center of gravity were measured for each of the

componmts.

Snags Cslculat ion

Position 2, 645 11). load:

 

v - 27.8 V(5.0 - «as + 12.0 - 135 - 2.0 - 135 + 5.0 - 150+ 11.5 - amlmzo

(Only backwheels undergo a decrease in height.)

61

Table 1s

62

Impact Velocities

 

 

 

Position Theoretical Velocity by Load: Meeeured‘Velocity by Load:
in/sec in/sec
575 795 1020 129 575 795 1020 1200
lbs. lbs. lbs. lbs. lbs. lbs. lbs. lbs.
1 62,5 60,5 39,5 32,9
2 65.2 63,8 62.3 63,1 52,5 52,5 51,5 43,2
3 62.0 63.0 61.0 56,0
6 83,2 82.5 ~ 83,3 71.5 72,5 70.0 63.5

 

 

 

 

Comments to Table l:

The truly obtained velocity is less than the

 

 

DrliFriction

 

no friction occurred.

Positions 2 end A (Table l).

The available velocity has

 

 

 

available. except when

been calculated for

According to the equations for dry friction:
5/3 ' ecc. Ih'sinx «r4 Hcos it
Where 3 is the angle of the car with the horizontal, we get:

VF I (3 sins -g}icos s) t+VoF

}* in the friction coefficient that we want to determine. When
integratixg we esme that cos x is constant, in fact it varies with the
time; for Position 2 from 0.955 to 1.0, for Position 6 from 9.96 to 1.0.
If there is no friction, we get:

V - g(sin:n)t 4'Vo where V is the friction«free velocity.

V0 - vol? - 0 (initial velocity), and then

V -VF'PS (cos x) t

t was measured to ca. 3.0 sec. (by a stapmtct:) for position 2

(and 4), which gives
V- (64.0 - 52.0) / 335 - 0.995 - 3 - 0.0101.
The correspordim friction caefficieut P”, between bearings
and axles is determined by
H-r “114/2 coax
whee r is the radius of journals, R is the radius of the wheels,
and + is the angle of friction between bearitgs and axles:
sin + - 0.0104 . 0.995 ' a / 0.3 - 0.104
M:- t3 J, - 0.105
which is quite usual for a railway truck. Therefore. this point in
simulation is quite well justified.

2. Velocity as measured as follows: Just before the car hits
the backstop, the time interval t between the front wheel hits the
first and the second switch was measured at a Hewlett Packard electric
counter. The distance d between the switches was determined carefully

by rolling the car very slowly over then, and the followim formula

used:

V Id/t (in/sec )Id ' '60 ' 60 / t '12 ° 5978 (mph)

1

Because of the enall friction and the small distance (about 8-10")
just before impact, we get a high degree of accuracy.
As a result of approximately 70 measuraeents, we found a double

standard deviation (20’) of 1.24.82 for velocities less than 60 in/sec. ‘

64

and 2.5 - 31 for velocities 65-31) in/sec. with a sample site of 2-3.
for longer series from one nosition as in Position 4, though less than
1.52

As a rule. the velocity as checked each time in Parts A and B
of the experiments. This is considered to be a very important
parameter that goes into energy calculations as (V92, and even more
enplnsis could be placed on a higher dqree of accuracy.

With or without load on to and includim 1020 lbs" the impact
velocities are fairly constant with the same height and are consistent
with the conservation of encgrconcent discussed under calculation of
theoretical velocity. There is no influence of distributai mass (two
or higher lumped) before the impact.

The signif icant jump in impact velocity at 1200 lbs. for same
height is believed to be the result of the overloading of the suspension
and supporting systan of the car. thereby increasing friction. In other
words. we are close to the aaritnum capacity of the car, and normally
do not want to use as high a load for that car.

Another observation made by high load and high position at the
imline revealed that the back end of the car almost lifts itself from
the rails. The reasons for this jump that the suspension systes is
hardly able to prevent. is the rotational moment, that deveIOps just
after the impact because of the different level of the attack point of

the intact force and the center of gravity of the car.

APPENDIX 11 .

Selected photographic Dario: for

draft goat pals... Table. 2 and 3.

6S

66

 

 

 

 

A +~+-+ hH—++

 

 

 

«0+4§+¢46

 

 

 

 

 

 

 

 

4 § 9-0 + +—4-+fi+4

 

 

 

 

 

 

 

 

 

i
’ .
Y
¥

9‘94.

‘
§
r
0
’
§
*
+

1
Of.‘

00¢

9.. 9‘§,'|11:’+*.Y#‘.0 Q 0 o O

9
4
9
0
9

91.¢#§+4‘40$III++I++J

.
,
.

o?ov9.9.o&.0009o4‘9§|

46+49++§4:
I
O

¢oq§¢¢ooo.4

 

+4 +4 or+¢H +—

.
¥
1‘]
.
_
w

#:Ifii- .

H44+4o9000¢¢6‘

 

68

C

O

‘

‘41V09¢0§0009

 

 

 

    
     

 

IIIII III
IIIIIIIflII

III-W
IIIIw
.IIIII IIIII
IIIIIIIIII
IIIIIIII!

I, fill'I" .fl

 

 

 

69

.

,

1417+ 11.1 U ml
Li‘wirlrlrLI] . I

. . _
I 4 . a

.

_

- h
‘ +1:.; .+
.

1
O
.9
a.

.0 ‘
i
‘
§
I,
‘9wy
.9
'
o

' O

3 Y‘o‘OI'O‘Q
.
~9996 9+¢6¢

~OOQ9100
4¢+Oow644Y40+

49—449or9o44—4'

+4+¢¢Qoooo¢o¢
l

 

 

‘vdr'

__‘ -4»

 

 

 

 

D’Db

 

V Y

 

 

 

L A A4
1 T V Y

 

.__J.

 

 

 

AAAL
T‘VYf

>—--~——4rr———

L - ,--_JL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

JLJA

_‘ w T «. ll

. _ .

, _ .

_ _ .

. _

_ +

4 000+0#0+1xo+00 o «+fifio o fibfiqfi+#0%¢4flfi¥jt 4%0rv41; r0$
_ Ar

_ . I _

_ . 4 . _

n _ . A? ,

, _ r 41 _ I

J‘ 1 J l 1}- LAT» 1” l§+t

_ V J

‘ . _ _ 4 V _
. _ _ . Aw
. fl “ Lv

A . _

?£>}}> VI? PD Ll? [ptbr 'Ir‘ Ll
.

 

ALJ A
fYWY

._-_fi4>_-

 

 

b—_- 4r ___+~.- -

AAA

WYY—T

 

 

 

 

 

 

44+

 

 

 

 

 

 

r
.L
‘ fr
+}

 

+~
L_
i .
I
I
4++of++++
I
!
.,

 

 

 

 

 

 

 

 

 

..
.9
‘ Q
' 0 l‘l l .901" OI I‘IIIL‘" M “’1‘
0 . .9.
Q 1" -
‘-
_
.
‘ I Y fi {J
‘ ,
4
6
+ Ll‘vl+
'0... ....... Q9o‘7¢9‘99000§59.9lb 9v?6,0$.?9n§‘¢.‘|++¢ ATII
I A
-
4.
_
+
I -I ? 9 ' '. 7|. O ‘..|'|.0‘
.
.9
‘-
4
}>} >7 frirtL bl} D'rrr >£ELV|E>EP5I¢WX
4111‘! 11114 11 1 4J1 111 4 _
O
6 ,
.
9 . a
VI l.‘ § 0 o O 1" I'lol‘Lu‘lJ
.
v _
.
‘
O _ _
_ _
1 _ F
v I 0| 1 4 9 t .4‘ . ‘.u+ I‘ll»;f| T 41 4
_
O _ _
4 _
4 V

 

. , . .
‘u ,
H .. P I

‘u‘_ w WM

 

, .
iii
‘ _ .

III-III _
_

Emmi-flu ,w

 

72

 

IMIIIMBNII

I O
. .

an I...

     

   

INEMIIIII
IIIIIIIIII

 

73

 

NW.”

.. ‘
~
.

1“."

A . , V P M

4

H. .mh.

_ ._w‘. _
.‘I_.

.__s. _
IIIII _ _ m.
w “

 

finial-I

 

 

_HIIIIIIIII

75

 

APPENDIX 11!
Behavior of Draft Gears

with Reference to Tables t and 5

l. The 10 in. Polystyrene-scar is more damping than the 5 in.
felt gear. which might be seen from the rebound velocities measured .
For the sane thickness. the dampixg is almost the same for the two
materials. both having felt dampers, which also might be seen from the

area of the pulses after stabilizing.39)

(This measure is snore difficult
to obtain than rebound velocities.)

2. It takes 34 impacts to stabilize both felt gear and styrene
gear, and tho styrene has less mechanical strength (easy to tear. etc.).

3. None of tho gears have preferences compared to rubber gears;
therefore, they have not been used any further.

e. Honoring the rebound velocities 1. a handy way to follow the
stabilisation of a certain gear. for example, after 24 h and 68 in

compression the stability of polystyrene was measured as follow:

 

 

 

 

 

Impact No. l 2 3 k 48 h
Rob. Velocity 20.0 21.0 22.5 21.5 20 .0
(in/see .) After 24 h 68 h

 

 

 

 

w-

39) AV measured by the pulse area is a monotonous function of the
damping capability: when AV increases - for the same conditions
of mass and impact velocity - then the dsmpixg of the gear decreases
(more precisely: damping is the decrease in kinetic energy during
th. impact).

76

77

Solo regeneration - although not sufficient for practical application -
is occurring during rest-ties.

5. the felt damper at the backstop has ssee damping effect at
the noise vibrations as observed by the rubber gear, though a little
higher effect in the case of felt gears as by the styrene gear. A
rubber sandwich pad, normally used by drop-testing. showed up to
have a damping effect on vibrations too, but at the same time it drags
out the duration of the pulse again.

6. The polyethylene gear 8” has the better dbmension stability

of the 3 non-rubber draft gears tested:

 

 

Thickness Thickness I D.
As Nev After 5 Impacts crease
Felt 5" 2 7/8” ‘3:
folystyrene 10 '15) " 5 1/2 '13) as:
(501)
Polyethylene 8 '(3) S 3/4 " (1‘ 1/2) 281
. (502)

 

 

 

 

 

 

Although lhlt and polystyrene has about the same permanent deflection,
polystyrene has the lower decrease in the beginning, but continues to
decrease at a higher rate than felt after the S. impact.

When using the 3 in. polyethylene gear, (giving approximately
the right pulse-duration) the permanent deflection becomes about 22;.
as for the other non-rubber gears tested. It seems as though the
resr‘rirgs of the non-rubber gears are shielded in the beginning and

gradually will come into action by the longer gear; whereas, they

contribute to the energy absorption immediately by the shorter gears.

78

7. Because of the permanent set, the smaller mechanical strength
(mostly by polystyrene) and the not very outstanding damping capabilities,
respectively energy absorption capability by the travel of gear, by a
proper pulse duration (especially polyethylene), it was decided to use
only truly resilient typo gears like rubbc, and to go no further with
non-rubber gears.

8. The silicone rubber gear broke at the third impact along a
radial plane in the Ming-cylinders. The break surface was uneven.
According to the commercial information:

Duroneter A-llardness, ‘5, maximum 65

Tensile strength, psi ‘00, maximum 700

Elongation, 1 180, maximal 200
It seems improbable that any type of Silastic R'i'V (the solf-curim type)
can withstand the loads involved, if they are not strictly centered so

2-is

that the whole cross-section area of the gear - about 10 in.
involved. We ought to have a safety factor such as 3 in tensile
strength, which means a requirement of ca. 2000 psi. The urethane-
rubber, mentioned previously, fulfills this requirement. Even if we
have this safety factor, we do not know for sure whether the dynamic
type of load will brat it or not.

With 1750 lbs. peak force and to - 180 m s, the A 1/2 in. silicone

rubber has properties similar to the 8 in. polyethylene gear (Table 5).

79

9. To divide the total length of a draft gear into several
smaller rings - instead of using one cylinder—ring -may have the
effect that eventually skewed forces at the point of impact will be
directed in the proper direction perpendicular to the gear after

having; passed 1 or 2 rings.

Table 5. Degradation of Selected Materials Used in Draft Gear Tests

§.€Lt.:.‘?zsar ‘L..!1

 

 

 

 

 

 

 

 

 

Thickness Peak Force Duration Rebound
Impact After Impact (Paired (102 Limits) Velocity
number Imhes Height) Lbs 3) new in/sec .
l 3 l/ 2 - - l7 .5
2 3 1/5 2700 120 -
3 _ 3 3100 100 19.0
With 6 3 3401') 90 20.0
Felt S 2 7/3 3300 90 20.5
xx)
Damper: 6 2 3/4 3400 30 21.0
7 .. .. .. ..
8 ~ - - 22 .43
9 - - - 21 .0
10 2 5/3 - - 21 .0
With
Rubber - 2700 110 -
Pad:
2:)

The uncertainty in force is about 50 lbs. ('32), in duration about
1’.) meet: .

n) + 2
AV (area) of No. 6 measured to 4.5 _ 0.2 (cm ).

Table 5. Continued.

’31

Polystyrene-Gear 1,0 Inch Respectivelz 6 Inch

 

 

 

 

 

 

 

 

Thickness Peek Force Duration Rebound AV
Impact After Inpect (Paired (102 Limit) Velocity
Number Inches Height) Lbs. use: in/sec. (enz)
1“‘) a 1000 160 10.0
2 7 - - 12.7
3 6 1/0 - - 14.2
1 5 3/11 1500 - 15.8
1° !“‘h‘ s 5 1/2 1550 150 16.8
6 ‘ 5 1/4 1750 140 -
7 5 1300 140 15.7
--- ..... ------.-----------.----.---
a 3 1/4 2300 130 -
6 Inch’ 9 3 2600 110 17.5
- - - - - - - - 4 - - - - - - t - - - - - - - - - - - . - - -.- . - - -
7.1: 10 3 . 2650 110 20.0 0.3
D'””"‘ (11) 2 7/8 2700 110 22.0 0.3
1 1,2 In, 1 1/1 £650 60 20.0 4.1
(After 118 h

in Compression )

m)

Cherscteristic pulse with flet top lasting more than 80 01sec.

Tabl. 5 e

Polyethylene-Foam Gear

Continued.

’-

‘vd
r.)

8 Inch Respectivelx 3 Inch

 

 

Thickness Peak Force L Duration Rebound
Lepect After Impact (Peired 10! Limit) Velocity
number Inches Height) Lbs. usec in/sec.
1 6 1/0 1150 200 25.5
2 6 1400 200 22.5
3 6 1750 200 23.5
8 Inch:
6 6 1750 200 23.5
5 53" "" "' 2‘s,
--------P--------- uuuuu pun-cup-----
6 1 1/2 3900 120 26.5
3 Inch:
7 1 1/2 6100 90 23.5

 

 

 

 

 

83

Figure 7

6 fieoprene Gear + Pelt Dagger

 

 

 

 

F A [63 Position 3
005 0 Load
5100 .
3/100
1700 ‘
t
SEEC!
20x 60: 100x
10-3 10~3 10~3
V 'fi ‘n/sec
6O ..
40 ‘.
20 +_
L I L \ t
I’ fi '— I
100: $811
10—3

 

 

80

Figure 7 Continued

 

‘D
{a

20: , 60: 100:
10-3 10~3 10-3

”1%.. Psi

625 Characteristic for the Gear +'Load

at the Given Pulse

340

255

170

35.

 

 

SEC

d

 

. :
1 ’ 2
x)

.5,
3

in

Calculated for the Rubber tart of the Gear Having a Cross Section

Area of 10 in.’

APPENDIX IV

Characteristics for the Test Track System

After consenting on the travel-characteristic of the draft gear
(Figure 7) the characteristic diagrams for the whole system
(Pigurea 5 and 6. Table 6) will be discussed.

II£!£$,°f the draft gear

The combined travel 2.6 in. of the 6 ring rubber gear under the

stated conditions is composed of:
travel of felt damper ,\, 3/6 in.
backstopaovenent N 1/6 in.
due to rubber-spring. etc. ,«c 1.6 in.

With 6 rubber rings each 3/6 in. v 6.5 in. the available travel is
about 0.80 ' 6.5 in. -»3.5 in. So ve still havs some available travel
of the draft gear. If we want considerably more energy absorbed without
changing the duration of the pulse. the rubber or spring type gear must
be conbined with or replaced by other types of gears, as discussed before.
or the cross-section erasinust be increased.

The characteristic for the draft gear (Figure 7) has uncertainties
because of the several links involved in deriving it. Although. it
adsht be better than a characteristic taken by gtggig_load application as
IIIDIutOd in the earlier report (1).

Some correspondence is obtained between the static load character-
istic and the present characteristic. For example, in the earlier report
a force of 2800 lbs. corresponds to a travel of 2.0 in. of a 7 ring gear.
while here a force of 2800 lbs. corresponds to a travel of 1.6 in. of
6 ring, thus a smaller travel by the dynamic load is indicated.

85

Duration

The formula T/2 -fivm" (also valid for a rather high dampimg)
does not hold. eventually because of the two-lumped mass or rather the
distribute! mans (discussed below). When comaaring mag. (575 lbs .)
at 2900 lb. force with 625 load (1200 lbs.) at 2000 lb. force, the k-

factor should be the same, and duration should therefore deviate with

 

a factor of \/1200 I 575 :71.5. They only differ with a factor
of 1.2. and the observation therefore supports the concept of
distributed mass in the simulation. Several authors account for at
least a tin-lumped macs (info)

If the mass is assumed to be lumped,

k - (1.5 I 1.2)2 - k - 1.55 - R
1200. 2900 1 575, 2900 -—-~ 575, 2900

The apparent k-factor becomes higher, as the impacting mass is
imreasirg. although the force is the ewe. This is not consistmt
with the concept of k as a staticially determined factor (the spring
itself can be considered as nessless), and we interpret the result in
true of distributed mass 100 / 1.56 - 651' of the actual mass relative

to 0 load .

 

60)
When considerable amount of cubic elasticity, multiply with the factor:

 

V2 I B (1+3) (4+ VFW, where

BI blight/kZ-vazm'r/kz

 

wha'e r is the parameter appearing in the approach equation for
the characteristic curve:

I e- k ' x + l’ ' :3 Clindlinmp. 11—13)

Only a small amount of cubicity appears by moderate loads.

87

On the other hand, when cmpering 220 load (795 lbs.) st 1.550 lbs.
force with 625 loed (1200 lbs.) st 4250 lbs.. there is no such difference.
There is no difference in the relative distribution of use in these two

eeses, (es fer es effect upon duration is concerned) es the fsctor

 

V1200 I 795 '- 1.23 expleins the different observed durstione vell
vithin the uncertainty of the neseurenent (f, 5 e S) .
The rise time for the pulses is ebout A0! of the total pulse length

for ell pulses.

Mutat—steticdoed—curve (Figure 6)

tor ems inpect velocities. the loed fector ("G"-fector) is plotted
egeinst totsl use I (1'I . e . 6‘ II I . G . g). It is interesting to
capers to the eel]. known curves for cushioning neteriele, where the peak
eccelcetion is plotted easinst stetie loed of the dropping heed of the
testim nechine for different drop heights (equivalent to different inpact

velocities).
Expectedly the curves would go down (Gm II V1 Vk/n except for damping)

es long es k reeeins conetsnt or rises less then the ones. Not only
stiffness end duping, but influence of distributed nese (end eventuelly
uplificstioe fectors) complicete explenetion of the curves.

Aseunitu the dupim to be epproxinetely the same at 1020 lbs. es
st 575 lbs. (shout the sane percentege decrease in 6 relative to 60'
which is justified by competing hysteresis ereas under the load-

displscenent curves, Figure 7). we get:

2
c I a - 57s I 1020 - k I k ;
( 1020. 40 575, 40) ( ) ( 1020, 40 $75. 40 )
2
k - 2.8 3.0.5 ' 1020 575 -k -l. '1:
1020, 40 ( , ) ( I ) S75. loo --- 575. 40,

k I k - 1760 lbs/in. (stress-strain curve)
575, ’00 1600 lbl.

k - (k )- 1.5 . 1760 - 26m lbs/in.
1020, 1.0 2900 m.

88

Accordim to the static curve:

63 .
k2900 II 00 lbs/1n

Therefore, we may seems that the static Iva-concept is not valid

II ‘1)

in this case, or we may calculate an ”equivalent mass different

fro. the actual uses by working back:

(2.3 I 3.05)2 - (57s I w.q) - (1300 I 1760); ugq - 1840 lbs.

Amplification Am II 18110 I 1020 I- 1.8

”I.

Now the static characteristic is not worked out for higher loads;
it is suggested to determine the stiffness-characteristic more accurately
for further calculations and comparisons.

Also, another eethod is tried, as follows:

By surely crossirg the curves in Figure 5 and assuming I: constant

by constant peak force:
Pn'vi Vice; 7. zit-V12 'n-Conatant

(still seeming damping as a certain percentage of the total work being

done). we get to (for 1020 lbs. load):

w;q (2900 lbs.) - 57s - (532 I 40.52) - 900 lbs.; A. - 0.96

u.q (1100 lbs.) - s75 . (71.52 I 64.52) - 1400 lbs; A" - 1.43

 

41 ‘
) The ease participating in the particular impact divided by static
lease nay be thought of as the "enmlification factor" in this case
mimltn. p. 71).

89

It is reasonable to believe. the Afirvl by lower loads. The

value of H;q is consistent with the finding k I 2640 lbs/in.

2900
The earlier characteristic then seems to be wrong, but only for higher
values of force such as for the 2900 lb. load. The foregoing calculation
of A. II 1.8 then must be wrom; too.

It is therefore suggested to calculate equivalent load or the
amplification factor according to the crossing method. instead of giving
up the constant k-ooncept.

At load factors less than about 3. the amplification effect is
negligible. By higher load factors such as 4.5 (6600 I 1020), the
amplification is about 1.5. etc.

The influence of distributed lass cannot be distinguished from
the influence of amplification by these measurements.

Characteristic Curves (Figure 3)

The shape of the peakrforce curves are fairly consistent with the
fact that the characteristic of the draft gear is somewhat cubic. in
other words, shows increasing stiffness (increasing k-factor) with

increasing load (Figure 7):
P Ila-V 'w-V \Ik/m
n n n

rm - vn Vi;

90

Approximately right for small damping”) small cubicity of

the characteristic ‘0) ‘1)

and mplification factors close to 1.
The duration of pulse is between a low of 85 and a high of 125. At

the velocity, we want to use in the further work: 0.0 mph” 70 in/sec..
the duration is between 90 and 10! use which corresponds fairly close

to the duration found in the field (75-100 uec).

 

For a netter of interest. when knowing damping, energy loss 9’ - 11777
or Iqaritnic decrement A '- 277‘; for the spring, and by using

-fi‘fi‘ _. ~p77‘
f a,“ 'X/ CV“ VII/I /0

for snail duping (see Hindlin p. 50), we may calculate stiffness
fro- the neasured values of I. and V“, or reverse (seeming also

that there is no influence from amplification factors).

APPENDIX V
Sample Derivation of Velocity, Displacement and

Stress-Deflection-Curves for the 6 Ring Rubber Gear with Felt Damper

For the test gear No. 10: 6 ring Neoprene gear + felt damper on
backstop, Position 3, 645 1b. load (1020 lbs. total), a picture was
taken of the pulse (No. 14), that is redrawn in Figure 7. From that
we derived the velocity - pulse by graphical integration using the
formula:

t t
vt-v1+S(F/n) dt-V1+386lwgrdt
O O

U - 1020 lbs

v120 (the final velocity) - 48.31013“.

Avuo - 61.0 + 48.0 I 109.0 iglsec.

which is compared to the measured velocity:
61.0 + 37.0 I 98.0 in/sec.
The uncertainty involved in the graphical method of integration is hardly

more than 21.

Also we derived the displacement-tine curve by using
I
dt-d°+SVt dt; do-O.
dII (the maximum deflection) occurring at t - 60 1a 3 is 2.6 in.,
which—Ts compared to the measured combined travel of the draft gear:

2 3/8 in.
91

02

The travel is determined by successive movement of a light beam
with increments of 1!: in., until the bean: would not be switched off
by the impact. Then the distance from the backstop at floor level was
subtracted from the distance at the point where the draft gear was just
touchim the backstop at V1 I 0.

The rebound velocity was measured in the same manner as the impact
velocity, chem ing the switches around and coupling the counter into
the system just after the impact began. The switches were hit by the
proper front wheel Just after the impact was over.

Discussion
(1) Q_i_r_ectiona1 Forces and Velocitx

Dissipation of energy in the rubber gear itself and in the felt

deeper is relatively small. Absorption of energy in the backstop

is due to the movement of the backstop, which is not totally rigid.

According to the law of nouentua:

Vt om-Mvvzd'anR

\

“(vi-wig) -.

where
V1 . a1 is the impact quantum (IQ).
VB. ° n is the rebound mentum "for tea car.
V2 is the velocity of backstop just after the impact.
11 is the equivalent mass of the backstop, which is

expressing the. degree of rigidness.

(2)

93

Now the maximum displacement of the backstOp is about 1/4 in.
at the point of impact (measured by a pencil fastened to the back-
stop), and when a symmetric shape movement of the backstop is

assumed, the final velocity V2 will be determined by:
‘AVZ - (2000 I to (m S) ) - du? - (2000 I 120) 0.25 - 4.2 in/sec.

Substituting gives:

_§.' (51 ‘ (“ 43) ) I 4.2 x m a 26 m

 

The pencil test also revealed that the backstop moves at an angle
of about 20° with the horizontal, partly because the cylindrical
gear hits the tubular load cell excentrically (it is observable
by empty car and smell load at 220 lbs.), and probably the
impacting surfaces are not exactly parallel either (which might
be very difficult to obtain). Also the backstOp as a whole seems
to be too weak and the bolts in the floor are not able to hold it
down.

From this it is clear that the impact is acting partly in
another direction than the horizontal, and when we are measuring
the rebound velocity only in the horizontal direction, we come

to a result which is less than that derived from the pulse.

Other Sources of Errors Discussed:

Influence of Non—lumped Mass

As pointed out before, the load is not acting together with the
car as one mass. flow in this case the wooden blocks are rigidly

fastened and fill out the entire car length.

(3)

(4)

(5)

(6)

96

The Error in Heaaurirg Rebound Velocity
The several measurements of rebound velocity show that the error 25’
is less than 52 even with one single measurement. The measurement

is taken just after the impact.

The Static Calibration Eethod

For strsin gags accelerometers one may expect that under dynamic
conditions, the response would be somewhat lower so that the real
force is higher then measured. This error will force the result

in the opposite direction of our finding for¢3V, and therefore there

is no indication for that kind of error.

Influence of Distributed Maqg_

When the mess is not acting as one lumped mass, we may expect that
the mess participating in the impact is different inldifferent
stages and durations of the impact (the shock wave concept,elso
Part B) and less than the constant nsss we assume in calculation
of AV. Then the reel AV would be larger than calculated, whereas
we found 137 already too large compared to the measurements.

Therefore, the observation gives no indication for this error either.

Determination of Weight of Components

It is believed that the weight of the wooden blocks and the car
conponuts is still the same as that measured about two years ago
(before the author). The time and facilities did not allow for
controlling that information. Wooden blocks could have dried out

to sons extent.

The very fine correlation between measured and calculated
displacement does not necessarily indicate accuracy. In two
other cases (gear No. 15, felt gear, and gear No. 10, at Position 2

end 0 load) was found:

 

 

 

A V dm
_* calc. mean. calc. mean.
[
Felt gear 94 76 1.8 2.‘
Rubber gear
(0 load) 99 80 2.2 1.8
}

 

 

 

 

 

 

 

In both double integration and in the beam method for determining
travel there ia a considerable uncertainty.

Despite the deviationa, it is concluded that the comparisona
deacribed above prove the general validity of the methodology

and the aize order of the measurements.

APPENDIX VI.

Selected photographic eeriea for

lading pulaea. Tahlea 7, B. 9 and 10.

96

97

 

III-SI.
alga-will?

IIIIECIE—b I
V‘

 

magnesia-
Ill-men:-

Il' Ilil
J 4.

 

. a

U .r
Ye‘ivlevvre‘J eoeeeeOeee

98

b
v
+1

9 I ‘Iue vl‘l‘.‘ 14

'9‘96‘60§e605*‘,?+9‘1+f§|a

gooeeoooQO¢e++e

ev¢§e9§.6¢veeeeeeeoI‘eOO‘I.

Yl‘ 9.?0‘. +4

I

 

 

 

1 1: - .
flIIIrIIII ,
IIIS-Iul
illgn-al-

IIIIIIIIII

lull-all
Ill-IIEIII
Ill-Ie-III
IIIIIIIII

   

 

 

 

i 100

i . n . _..
. _
H . . s
. _ .

. i _

.
.

I
Isa“.

  

 

:EIIE

l I ‘1 ‘A
i . M ‘4 “!_il
Iii
" ’I ‘4 _
.

            

 

. ,zflIIfi
_ III

“Hun“;

IIIIfiiIII
IIIIEEIIII
IIIIIEBIEI
III-IIISSI
IIIIIIIEEI
IIIIII§III

a1 4144
”IIIIEIIIH
IIIIIIII

        

 

102

on .

Ill-IIIIII
linselilll
seagull-II
[genius-3:
Ill-IIIIII

    

 

  
 

  
  

IIIIIIIII
gal . 1
_. .

 

I. OIL

I
IIIIIIBIII
HIKINIIIII
IIIIIIIIII
IIIIIIIIII
IIIIIIIIII
IIIIIIIIII

         

       

IIIIIIII
IIII1IIIII
Minute-ea;
IIIIIIIIII
IIIIIIIIII
IIIIIIIIQI
IIIIIIIIII
IIIIIIIIII
ell-eel.-

    

       

          

 

 

ll’lIIII

ununmmnunu
. I

[Illusawll

       

‘.

 

  

 

IIVIIIIIU
III-milli-
III-engul-
Ill-lee»:-
Immune-Im-
wuwnmumwu
1 1flIII1IW1II“
m w ‘I w I III

.AIIIIII
WT.“ In

.- .1 «i r

 

     

   

! M r

_ M J .lll
_ - 1 1l+11 --

TF1 Ill-l
IIIIIII'

105

 

  

106