Abstract - IRCOBI · ut on the s itudinal sea the shape ench‐end pa h‐end part id that a str r...

Abstract In railway collisions, passengers are thrown from their seats and translate long distances because

they are not restrained by a seatbelt. This creates the possibility of serious injury caused by collision with

interior components at high velocity. It is important, therefore that passengers’ behaviours are estimated. This

study focuses on a passenger seated on longitudinal seating away from a bench‐end partition. They translated

and then pitched over with its head leading any other physical part. Sled tests for imitating a situation where a

passenger collides with the bench‐end partition were conducted. Then, numerical simulations for reproducing

the sled tests were performed with MADYMO. The dummy’s behaviours of the numerical simulations, which are

collision position and collision velocity, were almost equivalent to those of the sled tests. On the other hand,

the head accelerations of the simulations were different from those of the sled tests greatly. Thus the joints of

the dummy model’s neck were improved. The head accelerations of the simulations after the improvement

almost corresponded with those of the sled tests.

Keywords behaviour analysis, longitudinal seating, multi‐body dynamics, seated passenger, railway collision.

I. INTRODUCTION

It is important that severe injuries are prevented when a railway collision happened. Since the railway

passengers do not have a seat belt, they are thrown from their seat by the acceleration of collision and collide

with interior components. The collision between the vehicle and external objects is called “primary impact” [1]

and the collision between passengers and interior components is called “secondary impact” [2]. This study

focuses on the secondary impact.

There has been a great discussion about the secondary impact on a railway vehicle. Severson (2000)

conducted a full‐scale collision test and numerical simulations to evaluate occupant protection strategies [3].

Wang (2011) analysed secondary collision injuries to occupants and influence factors of railway vehicle interior

impact injuries using the software MADYMO [4]. Li (2007) carried out numerical simulations using the software

PAM/CRASH [5]. Hecht (2005) conducted experiments and numerical simulations to improve the

crashworthiness of tramcars and light rail vehicle [6]. Some studies have shown the head injury value as a

function of secondary impact velocity. Tyrell (1998) analysed secondary impact against a seat back with

MADYMO [7]. Simons (1999) researched the survivability for a general high‐speed train system when a

passenger strikes a forward seat back [8]. Xie (2013) investigated the secondary impact of a passenger faced

with a table in a compartment [9]. In bus crashes, McCray (2001) simulated frontal rigid barrier test [10]. Elias

(2001) evaluated potential of safety restraints on large school buses and indicated that seat spacing could affect

dummy kinematics and responses [11]. Elias (2003) discussed relative performance of compartmentalization,

lap belt restraints and lap/shoulder belt restraints, as well as the effects of seat back height and seat spacing on

the performance of these safety restraint strategies [12]. Mitsuishi (2001) focused on passenger protection in

frontal collisions and showed the limitations of lap belts in preventing head impacts and the potential of

reducing passenger injuries by optimizing seat spacing [13]. In this way, although there has been extensive

study of train and bus crashes, they focus on transverse seating layout. In the United Kingdom, the standard for

safety assessment of transverse seating passengers exists [14], in which methods for sled tests using a dummy

has been prescribed. On the other hand, there is no standard for longitudinal seating passengers. The

longitudinal seating is typical as a commuter vehicle in Japan and can been seen in a European subway. Safety

assessment in such a layout is also important.

D. Suzuki (tel: +81‐42‐573‐7348; e‐mail: [email protected]), K. Nakai, S. Enami, T. Okino and J. Takano are researchers at the Railway Technical Research Institute in Japan. R. Palacin is a senior researcher at Railway Systems Group of NewRail at Newcastle University, UK.

Proposal of Simulation Method for Behaviour Analysis of Passengers on Longitudinal Seating in Railway Collision

Daisuke Suzuki, Kazuma Nakai, Shota Enami, Tomohiro Okino, Junichi Takano, Roberto Palacin

IRC-15-44 IRCOBI Conference 2015

- 327 -

ar

ac

fo

id

Th

Th

pe

M

Slusevinwse

Fi

Fi

pl

ut

be

co

ve

The seco

re thrown f

cceleration a

or considerin

dentify the v

herefore the

he error rate

Sled test

erformed.

Methods of th

ed tests wersed in the svent of headnterior consiswas used for eating (figure

g. 1. ES‐2 du

g. 3. Longitu

Existing i

late (Height

tilised in act

ench‐end pa

ollision posit

ehicle partiti

ondary impac

from their s

and initial po

ng safety m

vehicle acce

e purpose in

e of the seco

ts using a d

he Sled Test

re conductedled tests bed‐on collisionsts of a floorcommuter ve 3).

ummy.

udinal seating

nterior equip

: 1500 mm,

tual railway

artition influ

tion. Differen

on because

ct velocity is

seat by the

osition of the

easures. If a

leration for

this study is

ndary impac

ummy were

d to reproduecause the mn. Figure 1 sr, a longitudvehicles in J

g.

pments, such

width: 100

vehicles. A

uences the s

nce of collisio

the existing

s important

acceleratio

e passengers

a simulation

suppressing

to propose t

ct velocity by

I

e conducted

uce the secomovement oshows the ESinal seating apan. A feat

h as divider o

00 mm, mat

reason of

everity of th

on position a

vehicle parti

for safety as

on. The seco

. It is necess

n is able to

g the second

the simulatio

y the proposa

I. METHOD

and numer

ndary impacf a passengeS‐2 dummy. and a benchture of the

Fig. 2.

or partition,

erial: SS400

choosing th

he injury. T

affects the se

ition has har

ssessment in

ondary impa

ary to condu

predict pas

dary impact

on method t

al simulation

DS

rical simulat

ct of a passeer on longitFigure 2 shoh‐end partitilongitudinal

Layout on th

were not us

) was used

e steel plat

he severity

everity of the

rd position a

n railway col

act velocity

uct case stud

sengers’ be

velocity wit

o estimate t

n method wa

ions reprodu

nger. A side udinal seatinows the layoon. The longseating was

he sled.

ed for the be

as the benc

e is to avoi

of the injury

e injury in th

nd soft posit

llision becau

depends o

dies with the

haviours, it

thin an upp

the passenge

as 3%.

ucing the sl

impact dumng is mainlyout on the sgitudinal seas the shape

ench‐end pa

ch‐end part

id that a str

ry depends o

he case of us

tion (Figure 4

use passenge

on the vehic

ese paramete

is possible

er limit valu

ers’ behaviou

ed tests we

mmy (ES‐2) wy lateral in tsled. The moating was whof the buck

artition. A ste

ition, which

ructure of t

on stiffness

sing an existi

4).

ers

cle

ers

to

ue.

urs.

ere

was he ock hat ket

eel

is

he

of

ng


- 328 -

pa

pa

th

se

sh

re

ca

st

w

Fi

Fi

Fi

Compone

assenger ca

assenger’s h

he stiffness a

et at 5.0 m/s

hows the da

epresents fo

alculated us

tiffness in th

wall and the w

g. 4. Existing

g. 5. Appear

(a) Partitio

Stiffne

g. 6. Results

0

2

4

6

8

10

0

Force (kN)

Result

ent experim

n collide w

ead, collided

at two positi

sec which wa

ata measured

orce. The de

ing accelera

is study. The

window was

(a) Inte

g component

rance of the

on wall (Soft

ess: 350 N/m

of the comp

5 10 15

Stroke (mm)

measured

ments using

with, were c

d with them

ons, which w

as approxima

d and the st

eflection was

ation measu

e stiffness o

350 N/mm,

rior fitting.

t in the train

partition wa

position).

mm

ponent expe

20 25

Stiffness defined

a partition

conducted in

(Figure 5). S

were hard po

ately the sam

tiffness defin

s measured

ured by acce

f the soft po

1100 N/mm

vehicle.

ll experimen

(b) Partitio

Stiffne

riments.

0

2

4

6

8

10

0

Force (kN)

Result

wall and a

n order to

Since the sti

osition and s

me as second

ned. Horizon

with displa

elerometer

osition of th

and 2300 N

nt.

on wall (Hard

ess: 1100 N/m

5 10 15

Stroke (mm)

measured

window, w

measure th

ffness of the

soft position

dary impact v

ntal axis rep

acement gau

in the impa

e partition w

/mm, respec

(b) B

d position).

mm

20 25

Stiffness defined

which are in

heir stiffness

e partition w

n, were meas

velocity of th

resents repl

uge using m

actor. The s

wall, the har

ctively.

Back side of

Stif

0

2

4

6

8

10

0

Force (kN)

Result

nteriors a h

s. An impac

wall varies at

sured. Impac

he dummy’s

lacement an

magnescale. T

slope was d

rd position o

partition wa

(c)Windo

ffness: 2300

5 10 15

Stroke (mm)

measured

ead of sitti

ctor, imitati

t the position

ct velocity w

head. Figure

nd vertical ax

The force w

defined as t

of the partitio

ll.

ow.

N/mm

20 25

Stiffness defined

ng

ng

ns,

was

e 6

xis

was

he

on


- 329 -

sh

he

m

be

4.

al

va

Fi

Fi

thwthreth

secoofhebecasldiinm

The same

hows the ap

eld by a dev

mm from the

etween strok

.5 mm was 2

most covere

alid to imitat

(a) C

g. 7. Appear

g. 8. Compa

A corridohe RGS [14]. was difficult the sled was eference to 7he dummy in

Dummy’secondary imollision positf leaning of ead was meaefore the veauses the raed test withifferent fromnjuries arisingmethod is the

e experimen

pearance of

vice. The imp

e short side.

ke (mm) and

2250 N/mm,

ed a range of

te the stiffne

Condition of

rance of the

rison of the

or of the accAn accelerao reproducewithin the c7.0 G pulse nitially transls behaviour pact velocittion from thethe dummyasured usingelocity of dupid decelerah ES‐2 dummm ES‐2 in strug from the lae same as the

nts were con

f the steel p

pactor collid

Impact velo

d force (N). T

, respectivel

f the stiffnes

ess of existing

holding a pa

steel plate c

stiffness.

celeration totion impulsee the plateaucorridor (Fig(Figure 10). ated, then pand head y of the due floor is imp. The degreeg a high‐speemmy’s headation of dummy on the lucture. The hateral impace head injury

Partition w

Partition w

ducted by us

late compon

ed with the

ocity was se

The stiffness

y. Figure 8 s

ss of the inte

g interior co

artition

omponent e

o evaluate cre of 7.0 G wau part of the ure 9). An aFigure 11 shitched over winjury weremmy’s headportant to eve of leaning ed camera wd decreased mmy’s head. longitudinalhead performct, which is uy criterion (H

wall (Soft positi

SS400 (1.6m

wall (Hard positi

SS400 (3.2m

SS400 (4.5m

Wind

sing SS400 (T

nent experim

bench‐end

et at 5.0 m/s

value of 1.6

shows the c

eriors. It is fo

mponents.

experiment.

rashworthineas made in rcorridor in tacceleration hows the initwith its heade evaluated.d (SIVH) wervaluate the dof the dumhose frame suddenly waAs for head seating. Th

mance criterused in the fHIC) calculati

350

400

0 1

on)

mm)

on)

mm)

mm)

dow

St

Thickness: 1

ments. Four s

partition at

s. The stiffne

mm was 40

omparison o

ound that SS4

(b) I

ess of transveference to the performaimpulse of tial position d leading any. The collisire used to edummy’s behmmy influencrate was 100as defined ainjury indexe dummy uion (HPC) is field of automon method i

0

0

1100

1200

1000 2000

tiffness (N/mm

.6 mm, 3.2 m

sides of the

200 mm fro

ess value wa

0 N/mm, 3.2

of the stiffne

400 of 1.6mm

mpactor coll

verse seatingan upper limance of the t6.0 G, 5.0 Gof the dumy other physon position evaluate the haviour becaes the SIVH00 frame pers SIVH becaxes, there aresed in the Rused to indicmobile crashn the RGS.

2250

2300

0 3000

m)

Interior

SS400

mm and 4.5

bench‐end

om the long

as defined b

2 mm was 12

ess. The stiff

m, 3.2mm an

liding positio

g passenger mit value of ttest device. TG and 4.0 G my. It is a hical part. from the dummy’s buse it indica. The velocitr second. Veause the secoe no requireRGS is Hybrcate the likeh test. The H

mm). Figure

partition we

side and 10

by the relatio

200 N/mm a

fness of SS4

nd 4.5mm w

on

is provided the corridor.The velocitywere made ypothesis th

floor and tbehaviour. Ttes the degrty of dummylocity data juondary impaements for trid III which lihood of heHPC calculatio

e 7

ere

00

on

nd

00

was

in . It of in

hat

he he ee y’s ust act he is ad on


- 330 -

Fi

Fi

Fi

M

N

7.

(V

la

be

an

ex

g. 9. Compa

g. 10. Test c

g. 11. Initial

Methods of th

umerical sim

.4.2). MADY

Version 3.1)

yout of the

ench‐end pa

nd the floor

xperiments w

Although

0

20

40

60

80

0

Acceleration [m

/s2]

(a) Relation

rison betwee

onditions of

position of t

he Numerica

mulations un

YMO is used

was used as

interior com

artition mod

were mode

were substitu

h the longitu

0.1

Ti

of accelerat

en pulse 7G

acceleration

the dummy.

al Simulation

nder the sam

widely in t

s the passen

mponent mo

el. The long

elled with a

uted for con

dinal seating

0.2

ime [sec]

Acceleration

[m/s

2]

tion

and RGS cor

n pulse.

n

me condition

the field of

nger. Figure

odel which c

itudinal seat

plane. They

tact characte

g used in the

0.3

Upper lim

Lower lim

7.0G

0

20

40

60

80

0

[/

]

rridor.

ns of the sle

the automo

12 shows th

consists of a

ting was mo

y were rigid

eristics of th

e sled test w

3

mit

mit

Vl

i[

/]

0.1Time [sec]

(

ed tests wer

obile crash s

he ES‐2 facet

a floor mode

odelled with

bodies. The

e bench‐end

was bucket se

0

2

4

6

8

10

12

14

16

0

Velocity [m

/s]

0.2 0]

7.0G

6.0G

5.0G

4.0G

(b) Relation o

re performed

simulation.

t dummy mo

el, a longitud

an ellipsoid

results of t

d partition m

eat shown in

0.1

Time

Upper lim

Lower lim

7.0G

0.3

G

G

G

G

of velocity

d with MAD

ES‐2 facet d

odel. Figure

dinal seating

d. The bench

the steel pla

model.

n figure 3, it

0.2

e [sec]

mit

mit

DYMO (Versi

dummy mod

13 shows t

g model and

h‐end partiti

te compone

was modell

0.3

on

del

he

d a

on

ent

ed


- 331 -

w

du

Fi

nu

Re

Th

of

ac

ob

po

m

10

hi

co

w

he

th

Fi

with ellipsoid

ummy’s beh

g. 12. ES‐2 fa

Comparin

umerical sim

esults of the

he actual acc

f the sled a

cceleration a

bserved.

Figure 15

osition (0.00

measurement

020mm in th

igher the co

ondition of t

Figure 17

was determin

ead. Figure 1

he larger the

g. 14. Time h

because of

aviour, the i

acet dummy

ng the resu

mulations wa

e Sled Test

celeration of

cceleration

almost corres

5 shows the

00 sec), and

t procedure

he case. Tab

llision positio

he same acc

7 shows an e

ned as show

18 shows re

HPC. In the

history of the

difficulty of

nfluence wa

y model.

lts of the n

s inspected.

f the sled wa

(Acceleratio

sponded wit

dummy’s be

the right sid

of the colli

le 1 shows t

ons. The dum

eleration pu

example of t

n in figure 1

lation betwe

e case of the

e target acce

Acc

eler

atio

n [m

/s2]

modelling su

s simulated

umerical sim

as compared

n pulse: 5.0

th the target

ehaviour in t

de (b) was th

ision positio

he results of

mmy pitched

lse, the colli

the velocity

18, since sec

een SIVH and

almost same

eleration and

-20

0

20

40

60

80

0 0

[]

uch a shape.

by varying th

Fig. 13.

III. RESULTS

mulations w

with the tar

0G, Thicknes

t acceleration

the sled test

he time of se

on from the

f the collisio

d over large

sion position

change of t

condary impa

d HPC. In the

e SIVH, the t

d the sled ac

0.1 0.2

Time [sec

Tar

Sle

. Since the sh

he friction co

Layout of th

S

with those of

rget accelera

ss: 3.2mm) a

n. In the oth

t in the sam

econdary im

floor. The

n position. T

ly in the cas

ns were almo

he dummy’s

act causes a

e case of the

hicker the th

celeration.

0.3 0

c]

rget acceleratio

ed acceleration

hape of the

oefficient fro

he interior co

f the sled te

ation. Figure

and the targ

her condition

e case. The

mpact (0.214

collision pos

The higher th

e of the low

ost the same

s head over t

sudden dec

e same thick

hickness, the

.4

n

bucket seat

om 0.3 to 1.0

omponent m

ests, the pr

14 shows th

get accelera

ns, similar te

left side (a)

sec). Figure

sition from

he accelerati

w acceleratio

e.

time during

celeration of

kness, the hi

e larger the H

influences t

0.

model.

ecision of t

he time histo

tion. The sl

ndencies we

was the init

e 16 shows t

the floor w

ion pulses, t

n pulse. In t

the test. SIV

f the dummy

gher the SIV

HPC.

he

he

ory

ed

ere

tial

he

was

he

he

VH

y’s

VH,


- 332 -

Fi

Fi

Fi

Fi

g. 15. Behav

ig. 16. Measu

g. 17. The ve

g. 18. Relatio

(a)

viour of the d

urement of c

elocity chang

on between

Initial positi

dummy.

collision posi

ge of dummy

SIVH and HP

0

2

4

6

8

10

0

Volocity (m

/s)

5

10

15

20

HPC

on (0.000 se

ition from th

y’s head ove

PC.

50 100

0

00

00

00

00

0 2

4.5

3.2

1.6

ec) (b) T

TA

RE

he floor.

r time during

0 150 2

Time(ms)

4

SIVH (m

mm

mm

mm

Time of seco

ABLE 1

ESULTS OF TH

g the test.

4

5

6

7Acceleration

200 250

Seconda(214ms

Jusecond(215m

6 8

m/s)

ndary impac

HE COLLISION

1.6m

4.0G ‐

5.0G 1040

6.0G 1153

7.0G 1155

300

ary impacts, 6.4m/s)

ust afterdary impactms, 5.7m/s)

10

ct (0.214 sec)

N POSITION

m 3.2mm

960

0 1020

3 1140

5 ‐

Stiffness

)

4.5mm

990

1070

‐

‐

s


- 333 -

Results of the Numerical Simulation

The results of the dummy’s behaviours and the head accelerations in the simulations were compared with

those of the sled tests. First as for the dummy’s behaviours, error rates between the sled tests and numerical

simulations were measured in the case of varying the friction coefficient between the dummy model and the

longitudinal seating model. Figure 19 shows the error rates of the collision position and figure 20 shows those of

SIVH. In the collision position, the error rate was 19% when the friction coefficient was 0.3. The error rate was

decreased with increasing the friction coefficient. The error rate was minimum 3% when the friction coefficient

was 0.9. In the SIVH, the error rate was 21% when the friction coefficient was 0.3. The error rate was decreased

with increasing the friction coefficient. The error rate was minimum 3% when the friction coefficient was 0.9.

From the results, it was found that the dummy model’s behaviours of the simulations almost corresponded with

those of the sled tests in the condition that the friction coefficient was 0.9. Figure 21 shows relation between

SIVH and collision position in the sled tests and the simulations. Horizontal axis represents SIVH and vertical axis

represents collision position. In the sled tests, the higher the collision position, the higher the SIVH. The

simulations reproduced a similar tendency. Figure 22 shows examples of the dummy model’s behaviour in the

numerical simulation (Acceleration pulse: 5.0G, Stiffness: 1200N/mm, Friction: 0.9). The left side (a) was the

initial position (0.000 sec), and the right side (b) was the time of the secondary impact (0.215 sec). The

difference of the time of the secondary impact between the sled test and the simulation was only 0.001 sec. In

comparison with the results of the dummy’s behaviour in the sled test, the posture of the dummy model at

secondary impact was almost the same.

Fig. 19. Error rates of collision position. Fig. 20. Error rates of SIVH.

Fig. 21. Comparison of SIVH and collision position.

0

10

20

30

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Friction

Error rate (%)

0

10

20

30

40

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Friction

Error rate (%)

800

900

1000

1100

1200

1300

3 4 5 6 7 8 9 10 11

Collision position (mm)

SIVH (m/sec)

Sled test

Simulation


- 334 -

Fi

co

18

ac

se

Fi

A

in

th

co

on

th

ne

de

er

st

ex

im

im

si

de

ne

g. 22. Behav

Next, h

omparing tes

82G and HPC

cceleration w

even times th

g. 23. Head

Characte

lthough the

n the sled tes

he dummy m

onducted in

ne. Relations

he error rate

eck’s joints.

efault. Cont

rror rate wa

tandard devi

xample of t

mprovement

mprovement

gnificantly. B

efault dumm

eck’s joints.

Head

acceleration (G)

(a)

viour of the d

head acceler

st to simulat

C was 781. In

were observ

hat of the sle

G versus tim

ristics of nec

characterist

st. Therefore

model’s neck

order to im

s between s

es of the max

“D” means “

trolling the j

as 19% in th

ation of the

the head ac

, maximum

was seven

By the result

my model. T

0

100

200

300

400

500

600

700

200

) Initial posit

dummy mod

ration was i

tion (Acceler

n the simulat

ved which is

ed test. A co

me plot comp

ck’s joints we

ics of the joi

e the charact

. ES‐2 has fiv

mprove the c

tatus of the

ximum value

“default” an

oints one by

he status th

error rates,

cceleration

value was

times that

ts, it is diffic

The precision

210 220

Sled tes

Simulat

ion (0.000 se

del.

nvestigated.

ation pulse:

tion, maximu

not observe

mpletely diff

paring test to

IV

ere consider

nts simulate

eristics of th

ve joints at it

haracteristic

neck’s joint

e of the head

d “C” means

y one, error

at only the

the status th

in the statu

227G and H

of the sled

cult to evalua

n of the simu

230

st

tion

ec). (b) Tim

. Figure 23

4.0G, Thickn

um value wa

ed in the sle

ferent tende

o simulation.

V. DISCUSSIO

red as the rea

e those of the

he neck’s join

ts neck. Exa

cs of the five

ts and the he

d acceleratio

s “controlled

rate was 21

bottom join

hat all joints

us that all

HPC was 11

test, the e

ate the injur

ulation was

240 250

Time (ms)

me of second

shows exam

ness: 4.5mm

as 708G and

ed test. The

ency was pro

ON

ason of the l

e dummy’s n

nts were imp

minations to

e joints. The

ead accelera

on. The horiz

”. The error

% in the sta

nt was defa

were contro

joints were

146. Althoug

error rate af

ries of passe

improved by

260

dary impact

mples of the

). In the sled

HPC was 567

HPC of the s

ovided in the

ow precision

neck, head ac

proved. Figur

o control mo

e neck’s joint

ation were e

zontal axis re

rate was 75%

tus that all j

ult. Conside

olled was ad

e controlled

gh HPC in t

fter the imp

engers in the

y controlling

270 280

(0.215 sec).

e head G ve

d test, maxim

76. Vibratio

simulation w

e head accele

n of the head

cceleration d

re 24 shows

ovement of t

ts were con

examined. Fig

epresents th

% when all f

joints were c

ering the ave

opted. Figur

. In the sim

the simulatio

provement w

e railway col

g the charact

290

rsus time pl

mum value w

ns of the he

was more th

eration.

d acceleratio

did not vibra

appearance

the joints we

trolled one

gure 25 show

e status of t

ive joints we

controlled a

erage and t

re 26 shows

mulation aft

on before t

was decreas

lision with t

teristics of t

300

lot

was

ad

an

on.

ate

of

ere

by

ws

he

ere

nd

he

an

ter

he

ed

he

he


- 335 -

Fi

Fi

Fi

co

m

co

0.

28

Fi

g.24. Five jo

g. 25. Error r

g. 26. Head

For more

oefficient be

maximum and

ondition that

.90, the rang

8% which wa

gure 28 sho

Head

acceleration (G)

ints of the d

rates of max

G versus tim

e improvem

etween the d

d the minim

t reproduces

ges of the e

as minimum

ows relation

0

50

100

150

Error rate of maximun head

acceleration (%)

0

100

200

300

400

500

600

700

200

ummy mode

ximum head

me plot comp

ment of the

dummy mod

mum value o

s the results

rror rates w

m. When the

between SI

D

D

D

D

D

D

D

D

D

C

D

D

D

C

C

210 220

Sled te

Simula

el.

acceleration

paring test to

precision, a

del and the b

of the error

of the sled t

ere almost 4

friction coef

IVH and HPC

D

D

C

C

C

D

C

C

C

C

Statu

230

est

ation after im

n.

o simulation

a simulation

bench‐end p

rates of th

test thorough

40%, respec

fficient was

C in the con

C

C

C

C

C

C

C

C

C

D

C

C

C

D

D

us of five joint

240 250

Time (ms)

provement

after improv

analysis w

partition mo

e HPC were

hly (figure 27

tively. When

more than 0

ndition that

C

C

D

D

D

C

D

D

D

D

ts

①

②

③

④

⑤

260

vement.

as conducte

del from 0.8

e investigate

7). When th

n the friction

0.91, the ran

the friction

D: Defau

C: Contro

①

②

③

④

⑤

Average

Standard devia

270 280

ed by varyin

86 to 0.95.

ed in order

e friction wa

n was 0.91,

nges were m

coefficient

ult

ol

ation

290

ng the fricti

Ranges of t

to extract t

as from 0.85

the range w

more than 35

was 0.91. T

300

on

he

he

to

was

%.

he


- 336 -

results of the simulation showed the tendency same as that of the sled test

Fig. 27. Range of error rate of HPC.

Fig. 28. Relation between SIVH and HPC.

The simulation method in this study has a limitation on the precision of the HPC. Although the error rate of

the SIVH was only 3%, the error rate of the HPC was almost 20%. Two reasons are considered. One is the

characteristics of the neck’s joints of the dummy model. Since the ES‐2 model was developed for automobile

crash simulation, the default model was not able to evaluate the head injury in the railway collision accurately.

The other is that the bench‐end partition model, which was the steel plate, was the rigid body. Actually the

steel plates deflect due to the impact. It is necessary to model them with FE.

V. CONCLUSIONS

This study focused on a passenger seated on longitudinal seating at the time of the railway collision. The

purpose was to propose the simulation method to estimate the SIVH. In order to confirm the situations of the

secondary impact, the sled tests were conducted. Numerical simulations reproducing the sled tests were also

performed. The results of the SIVH in the numerical simulations almost coincided with those of the sled tests.

As for the head acceleration, the results of the simulation were different from those of the sled tests greatly. It

was found that the dummy model of the default is not usable for the evaluation of the head injury in the railway

collision. Controlling the characteristics of the neck’s joints, the precision of the head acceleration was

improved. In the future, conducting parametric studies with various accelerations, relation between the vehicle

acceleration and the secondary impact velocity is going to be clarified.

VI. REFERENCES

[1] Chatterjee S and Carney III JF. Passenger train crashworthiness – primary collisions. Transportation Research Record, 1996, 18, pp. 1–12.

0

20

40

60

80

0.86 0.87 0.88 0.89 0.90 0.91 0.92 0.93 0.94 0.95

Range

of error tate of HPC

(%)

Friction

0

500

1000

1500

2000

0.0 2.0 4.0 6.0 8.0 10.0

HPC

SIVH (m/s2)

4.5mm (Simulation) 3.2mm (Simulation) 1.6mm (Simulation)

4.5mm (Sled test) 3.2mm (Sled test) 1.6mm (Sled test)


- 337 -

[2] Chatterjee S and Carney III JF. Passenger train crashworthiness – secondary collisions. Transportation Research Record, 1996, 18, pp. 13–19.

[3] Severson JK, Tyrell DC and Perlman AB. Rail passenger equipment collision tests: analysis of structural measurements. International Mechanical Engineering Congress and Exposition, Orlando, FL, 2000.

[4] Wang W. Influence factors of railway vehicle interior impact injury. Applied Mechanics Materials, 2011, 79, pp. 227–231.

[5] Li L. Research on designing crashworthy structure for urban rail vehicle and occupant safety by simulation method. China Academy of Railway Sciences, 2007, Beijing.

[6] Hecht M. The crashworthiness of tramcar and LRV. Foreign Rolling Stock, 2005, 42(5), pp. 39–41. [7] Tyrell, D., Severson, K. and Marquis, B. Crashworthiness of passenger trains – safety of high‐speed ground

transportation systems. U.S. Department of Transportation Federal Railroad Administration Office of Research and Development, 1998, pp. 7–8.

[8] Simons, J. W. and Kirkpatrick, S. W. High‐speed passenger train crashworthiness and occupant survivability. IJ Crash, 1999, 4(2), pp. 121–32

[9] X., Suchao and T., Hongqi. Dynamic simulation of railway vehicle occupants under secondary impact. Vehicle System Dynamics, 2013, 51(12), pp. 1803–17.

[10] Linda McCray. Simulation of large school bus safety restraints. 17th International Technical Conference on the Enhanced Safety of Vehicles, 2001, Paper No.313.

[11] Jeffrey C. Elias, Lisa K. Sullivan and Linda B. McCray. Large school bus safety restraint evaluation. 17th International Technical Conference on the Enhanced Safety of Vehicles, 2001, Paper No.345.

[12] Jeffrey C. Elias, Lisa K. Sullivan and Linda B. McCray. Large school bus safety restraint evaluation – Phase II. 18th International Technical Conference on the Enhanced Safety of Vehicles, 2003, Paper No.313.

[13] Mitsuishi, H., Sukegawa, Y., Fujio, M. and Okano, S. Research on bus passenger safety in frontal impacts. 17th International Technical Conference on the Enhanced Safety of Vehicles, 2001, Amsterdam, The Netherlands.

[14] Railway safety and standards board. Railway Group Standards GM/RT2100 issue 5: Requirements for rail vehicle structures, 2012, Railway safety and standards board, London, UK, pp.27–30.


- 338 -

Abstract - IRCOBI · ut on the s itudinal sea the shape ench‐end pa h‐end part id that a str r...

Documents

Transcript of Abstract - IRCOBI · ut on the s itudinal sea the shape ench‐end pa h‐end part id that a str r...