Professional Documents
Culture Documents
Equivalence Study Involving Rail Vehicle
Equivalence Study Involving Rail Vehicle
Abstract
In the context of rail vehicle collision tests, which incur high costs and consume substantial time and energy, the
equivalence between a single vehicle crashing into a rigid wall and two identical vehicles colliding with each other was
studied. Taking the car body as a rigid body, a three-dimensional multi-body dynamic model was built to simulate a single-
vehicle impact and a collision between two identical vehicles; the results showed that the condition of a single vehicle
crashing into a rigid wall at a speed of v=2 can be used to replace the condition of one vehicle moving at a speed of v and
crashing into an identical vehicle that is stationary. However, the actual collision is a strong nonlinear process, and it is
necessary to conduct the equivalent test of the condition of collision. Based on the similarity theory, the scaled equiva-
lent vehicle model is established. Through a series of scaled model tests, the following conclusion is drawn: if one vehicle
moving at a speed of vcrashes into another identical vehicle that is stationary, one can equivalently use a single vehicle
with a speed of v 0 ¼ 0:5248v þ 0:1608 (units: m/s) that crashes into a rigid wall. This study provides practical support for
the equivalence of vehicle collision test conditions and holds great value for engineering applications.
Keywords
Collision test, equivalence, multibody dynamics, similarity theory, scaled model
into two aspects: structural crashworthiness and the experimental research and thus expand the content of
interior occupant environment.9 Tyrell and Martinez experimental research to address the lack of research
provided an overview of the in-line, full-scale impact in this area; for instance, the force sensor which was
tests conducted by the FRA (which included single- fixed on rigid wall can be introduced in car-impact
vehicle impact into a fixed barrier, two-vehicle impact rigid-wall tests to obtain the impact force during the
into a fixed barrier, and cab car-led train collision collision.
with a standing locomotive-led train) and discussed The SAFETRAIN team conducted a coasting test
a strategy for preventing override between colliding in the laboratory of the British Automobile
equipment. Their results showed that under the Manufacturing Industry Research Association to
same impact conditions, the CEM equipment pre- verify the multibody dynamics and finite element cal-
served the integrity of the space for all occupants.10,11 culation; in this work, the C2 condition (two identical
Five interior occupant experiments were conducted as city trams colliding at a speed of 20 km/h) is simplified
part of the two-car impact test of CEM equipment. to a single vehicle with a test driver’s cab impacting a
The preliminary test results confirmed that the sec- rigid wall at a speed of 14 km/h.24 Chen proposed that
ondary impact environment in the two-car CEM test when two urban rail vehicles collide at a speed of
is indeed more severe than that of the previous tests.12 20 km/h, the energy absorption of the head car can
A prototype CEM coach car design has been devel- be calculated using the formula Ed ¼ 0:1875 Mv2
oped and successfully tested in two full-scale tests. (proposed by Lu25and Lu and Ronghua26), which is
The design showed significant improvements over equivalent to an urban rail vehicle impacting a rigid
the conventional equipment similarly tested.13,14 wall at a speed of 12.25 km/h. Similarly, when two
British researchers have conducted full-scale tests to urban rail vehicles collide at a speed of 30 km/h, the
validate concepts for energy absorption in crush zone energy absorption of the head car is equivalent to an
designs. The tests included a train-to-train impact that urban rail vehicle impacting a rigid wall at a speed of
demonstrated the energy absorption capabilities and 18 km/h.27 These conditions are different from each
crush distribution through the train. The results of a other; thus, there should be an exact method to sim-
car-to-car impact showed that the crush zone, which plify the test conditions. However, it is almost impos-
included pushback couplers and anticlimbers, pre- sible to conduct a full-scale train crash test because of
vented override.15 In 1994–1996, the French the high cost; therefore, as reported in this paper, a
National Railway Company (SNCF) sponsored mul- scaled equivalent model test was conducted to deter-
tiple full-scale tests of TGV equipment to qualify the mine how to simplify the test conditions.
crashworthiness features in the bi-level TGV and the The scaled equivalent model was established based
XTER DMU.16 Similarly, the UIC sponsored on the similarity theory. Scaled model tests have been
a research effort called SAFETRAIN, in which widely used in ship design, aerospace industry, bridge
a series of tests were planned to test the program’s structure engineering, traffic design, bullet explosion
prototype design in 2000.17–19 The Indian Railway shock analysis, bridge and tunnel deformation failure
conducted full-scale tests of redesigned GS and SLR analysis, and other engineering problems. Simplified
passenger railcars that included crush zones and con- small-scale model motor coaches were designed and
cluded that particular attention must be paid to tested under both static and dynamic loading, and
the collision loads that the crush zone support struc- attempts were made to relate the results to a full-
ture must sustain.20 In addition, a large number scale collision by Lowe et al.28 Kao and Chan29
of tests must be performed for the collision and Barley and Mills30 conducted a 1/10 and a 1/2
safety project TRAINCOL, SAFETRAM, and automobile scaled model test to study automobile
SAFEINTERIORS.21 The above-described experi- crashworthiness. Jiang et al.31 studied the scaled col-
mental studies involve multiple car-to-car impact lision law of a metal cone specimen. Wang et al.32
tests. In fact, the test of single vehicle impacting studied local damage effects in a concrete target
rigid wall is often carried out to evaluate the crash- under the impact of projectiles. Xu et al.33 showed
worthiness of energy absorber which was fixed on that scaled models represent real models if appropri-
the test vehicle.2,22 However, in order to evaluate ate similarity coefficients are applied. Rui et al. tested
the energy absorption capability of train head car, a one-eighth-scale model of train-to-train collisions
not only car-impact rigid-wall tests but also car-to- for a three-car set and simulated one-eighth-scale
car impact tests must be carried out,23 which incur train collisions for three-, five-, and eight-car sets;
high costs and consume substantial time and energy. the simulation results indicated that train sets do
In this paper, a series of studies are carried out to not significantly affect the energy absorption and
discuss how to simplify the car-to-car impact test con- energy dissipation patterns. The energy-absorbing
dition. A possible approach to simplify the test con- structure at the front of a train plays a major role in
ditions is to replace the conditions of car-to-car collisions.4
impact tests with the conditions of car-impact rigid- In this study, based on similarity theory, selecting
wall tests. This approach can not only save time and length, force, and time as the basic dimensions,
reduce test costs but can also improve the efficiency of according to the Buckingham theory, which is the
Yao et al. 75
1 1
Wa2 ¼ k2 þ k2 ð Þ ¼ k2 k2 2 ð15Þ
2 2
thus
1 1
Mv2 ¼ k2 k2 2 ð17Þ
8 2
where
Mv2
¼ þ ð18Þ Figure 4. Loading curve.
8k2 2
entered that collectively represent the initial state of and connection between bogie and car body are simu-
the actual rail vehicle of the initial position lated by springs and dampers. The dynamic friction
(0 m, 25 m), initial velocity (18 km/h for single car coefficient of the wheel–rail contact is set to 0.03. The
impacting a rigid wall condition, 36 km/h for car-to- coupler is considered a nonlinear spring that incorp-
car impact condition), and mass of each vehicle orates the loading and unloading characteristics of the
(60,000 kg). Connection between bogie components buffers and crushing tubes. The energy absorption
78 Proc IMechE Part F: J Rail and Rapid Transit 233(1)
Figure 8. Vertical displacement–time curve of wheels in condition 1. (a) Moving car and (b) stationary car.
mc: moving car; sc: stationary car.
Discussion
For condition 1, which specifies a moving car with a
speed of 36 km/h impacting an identical car that is
stationary, the calculated values of the total displace-
ment of the two vehicles and the speed change of the
two vehicles during the course of the collision are
shown in Figure 5. Figure 9. Vertical displacement–time curve of wheels in
For condition 2, which specifies a single moving car condition 2.
with a speed of 18 km/h crashing into a rigid wall, the mc: moving car.
calculated values of the displacement of the vehicle
and the speed change of the vehicle during the single car crashing into a rigid wall with the speed of
course of the collision are shown in Figure 6. v=2 can be used to replace the condition of one car
Based on Figures 5 and 6, we conclude the follow- moving at speed v crashing into an identical station-
ing: (1) the collision duration is approximately 150 ms ary car. This approach can help simplify experiments
in both conditions; (2) for the two-car collisions, the and improve efficiency.
speed of the stationary car is slightly greater than that The derailment stability of trains is considered in
of the moving car after collision, but each is close to many recent studies,34–36 and it has been studied in
5 m/s; and (3) for a single car crashing into a rigid this paper with the help of a three-dimensional multi-
wall, the car has a small rebound speed after the body dynamic model used in the ‘‘Design of multi-
collision. According to the simulation results, in the body model’’ section. The wheels are numbered such
collision process of two cars, the total deformation of as mc1-1 (moving car 1-1), sc1-1 (stationary car 1-1),
the energy absorber is 816.74 mm; thus, the deform- and the detailed number of wheels is shown in
ation of the energy absorber of each car is 408.37 mm; Figure 7.
in the process of a single car crashing into a rigid wall, The calculation results of vertical displacement of
the deformation of the energy absorber is 408.04 mm, wheels in conditions 1 and 2 are shown in Figures 8
i.e. the error is only 0.33 mm. We can conclude that a and 9. As for moving car in condition 1, the vertical
Yao et al. 79
During the collision, the maximum relative dis- Table 1. Scaling parameters.
placement of the two vehicles is
Physical parameter Dimension Scaling factor
V1 Length (L) L s
s1 ¼ ð28Þ
!n1 Area (S) L2 s2
Volume (V) L3 s3
Supposing the mass of each vehicle is the same, Density () FT2L4 1.0
qffiffiffiffiffi
!n1 ¼ 2k
M1 , and the deformation of the energy absor-
1
Mass (m) FT2L1 s3
ber of each vehicle is Load (F) F s2
Elastic modulus (E) FL2 1.0
sffiffiffiffiffiffiffiffiffiffiffiffiffi
Stress () FL2 1.0
s1 M1 V21
1 ¼ ¼ ð29Þ Strain (") – 1.0
2 8k1
Displacement () L s
For the condition of a single car crashing Velocity (v) LT1 1.0
into a rigid wall, it is assumed that the mass of the Time (t) T s
stationary car is infinite, i.e. M2 ! þ1, and M12 ! 0; Elastic and plastic energy (W) FL s3
k2 pffiffiffiffiffiffi qffiffiffiffiffi
k2 Kinetic energy (E) FL s3
thus, K2 ¼ M 1
, !n2 ¼ K2 ¼ M 1
, and the deform-
ation of the energy absorber of the moving vehicle is
1 1
sffiffiffiffiffiffiffiffiffiffiffiffiffi Supposing
that K 1 ¼ k1 M 1
þ M 2
and C1 ¼ c1
1 1
V3 M1 V23 M1 þ M2 , equation (32) can be simplified as follows
2 ¼ ¼ ð30Þ
!n2 k2
l€ þ C1 l_ þ K1 l ¼ 0 ð33Þ
The energy absorber in the condition of a moving The circular frequency of the vehicle longitudinal
car crashing into a stationary car can be considered as vibration in the undamped condition is
a series connection of energy absorbers of a single sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
car; for springs in series, we have k1 ¼ k22 . As a pffiffiffiffiffiffi k1 ðM1 þ M2 Þ
!n1 ¼ K1 ¼ ð34Þ
result, if the deformation of the energy absorber M1 M2
of each vehicle is the same, then 1 ¼ 2 ,
qffiffiffiffiffiffiffiffiffi2 qffiffiffiffiffiffiffiffiffi2 In the low-damping condition, the circular fre-
M1 V1 M1 V 3 V1
8k1 ¼ k2 , i.e. V3 ¼ 2 . quency of the vehicle longitudinal vibration is
qffiffiffiffiffiffiffiffiffiffiffiffiffi
We conclude that a single car with a speed of v=2 C1
!d1 ¼ 1 12 !n1 , and the damping ratio is 1 ¼ 2! .
crashing into a rigid wall can be used to replace the n1
condition of one moving car with a speed of v crash- The general solution of differential equation (33) is
ing into an identical car that is stationary.
If the collision scene of one moving car crashing lðtÞ ¼ e1 !n1 t ðC1 cos !d1 t þ C2 sin !d1 tÞ ð35Þ
into another stationary car is simplified as two lumped
mass blocks connected by a spring and a damper, The initial conditions of the collision are
as shown in Figure 12, supposing that the vehicles X1 ¼ X2 ¼ 0, X_ 1 ¼ V1 , and X_ 2 ¼ 0; thus, l ¼ 0,
are moving in the same direction when they collided, l_ ¼ V1 , and the special solution of differential equa-
the velocities of cars 1 and 2 are V1 and V2 , tion (33) is
respectively.
V1
Supposing that the longitudinal absolute displace- lðtÞ ¼ sin !d1 t ð36Þ
ments of the center of gravity of cars 1 and 2 are X1 !d1
and X2 , respectively, the motion differential equation During the collision, the maximum relative dis-
is placement of the two vehicles is
( V1
M1 X€ 1 þ k1 ðX1 X2 Þ þ c1 X_ 1 X_ 2 ¼ 0 s1 ¼ ð37Þ
ð31Þ !d1
M2 X€ 2 k1 ðX1 X2 Þ c1 X_ 1 X_ 2 ¼ 0
Supposing that the mass of each vehicle is the
qffiffiffiffiffi
same, !n1 ¼ 2k
M1 , and the deformation of the energy
1
For the condition of a single car crashing into a The V3 represents the equivalent velocity in car-
rigid wall, it is assumed that the mass of stationary car impact rigid-wall condition.
is infinite; thus M2 ! þ1, and M12 ! 0; as a result, The energy absorber in the condition of a moving
k2 pffiffiffiffiffiffi qffiffiffiffiffi
k2
car crashing into a stationary car can be considered as
K2 ¼ M 1
, ! n2 ¼ K2 ¼ M 1
, and the deformation of a series connection of energy absorbers of a single car;
the energy absorber of the moving vehicle is for springs in series, we have k1 ¼ k22 . As a result, if the
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi deformation of the energy absorber of each vehicle is
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V3 4M1 V23 M1 V21 4M1 V23
2 ¼ ¼ ð39Þ the same, then 1 ¼ 2 , and 8k M 2 ¼ 4k2 M1 C22
,
!d2 4k2 M1 C22 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 C1
8k M c2
i.e. V3 ¼ V21 8k 1M 14c22 .
1 1 1
l 0 1 l
Figure 13. Loading curves of a crushing tube and a ¼ _ ð40Þ
honeycomb.
_l K1 C1 l
Figure 14. Honeycomb used in the test. (a) 40 60 mm2, 3.3 MPa and (b) 40 60 mm2, 10 MPa.
Figure 15. Test arrangement. (a) Car-to-car impact and (b) car crashing into rigid wall.
Numerical integration methods, such as the fourth- the large prototype in engineering. Dimensionless ana-
order Runge–Kutta method, can be used to obtain the lysis is commonly used for testing models in practice in
solution of the above differential equation. situations where the equations related to the param-
Because the actual collision is a strong eters of a process are unknown. To ensure that the
nonlinear process, it is necessary to conduct the scale-test models are valid, according to the
equivalent test of the collision condition. However, Buckingham Theory, which is the basis of the major-
it is almost impossible to perform a full-scale ity of dimensional analysis, asserts that any complete
train crash test because the cost is too high; as a physical relationship can be expressed in terms of a set
result, as reported here, a scaled equivalent model of independent dimensionless products composed of
test was conducted to study how to simplify the test the relevant physical parameters,37 and we have
conditions.
fðq1 , q2 , , qn Þ ¼ 0 ð41Þ
Scaled equivalent model test Assuming that qn , qn1 , , qnkþ1 are k independ-
ent base physical quantities, there are k groups of real
Similarity theory
numbers an, i , an1, i , , ankþ1, i . Then, qi can be
If two systems are kinematically similar and have expressed as
similar mass distributions, then dynamic similarity
a a
exists.37 Therefore, the similar model, which is estab- qi ¼ i qann, i qn1
n1, i nkþ1, i
qnkþ1 , i ¼ 1, 2, , n k ð42Þ
lished based on similarity theory, can be used to study
Yao et al. 83
qi
i ¼ a an1, i ankþ1, i ð43Þ
qnn, i qn1 qnkþ1
Figure 17. Displacement–time and velocity–time curves of the test of a car impacting a rigid wall. (a) Test A1 (v = 2.628 m/s), (b) test
A2 (v = 2.759 m/s), (c) test A3 (v = 3.143 m/s), (d) test A4 (v = 4.554 m/s), and (e) test A5 (v = 5.881 m/s).
84 Proc IMechE Part F: J Rail and Rapid Transit 233(1)
parameters were obtained in a previous study.38 (40 60 mm2) and under a stress of 10 MPa, are used
A completely geometric similarity between a proto- to simulate the second-degree energy absorber of the
type and the scaled model is impossible; the less scaled equivalent model car, the initial length is
important parameters, such as the angular velocity, 62.5 mm; the honeycomb is shown in Figure 14.
can be ignored. The similarity coefficients of the phys- Comparisons of the corresponding parameters are
ical parameters are presented in Table 1. shown in Table 2. Since the mass and velocity of car
and force of energy absorbers have the greatest effect
on the collision energy absorption, the scaling factor
Test design for velocity is 1.0, so this study mainly listed two
Based on the similarity theory, a one-ninth-scale scaling parameters about the mass and force.
model of the train is established.39 The mass of the The test was conducted in the test bed for small-
scaled equivalent model is 135 kg, so the mass of the scale models at the Central South University. For
full-scale car is 98,415 kg, which is close to the mass of condition 1, the moving car was launched by an air
a locomotive. The plastic deformation platform force gun, collided with the stationary car. For condition 2,
of a honeycomb material is stable,40,41 and the loading the moving car was launched by an air gun, crashed
curves of a crushing tube (energy absorber of vehicle) into a rigid wall. The test arrangement is shown in
and a honeycomb are shown in Figure 13; thus, an Figure 15 and the test scenario in Figure 16. The
aluminum honeycomb can be used to simulate the entire collision process was recorded using a high-
energy absorber of vehicle. Honeycombs, each with speed camera; the displacement–time curves and vel-
a cross-sectional area of 2400 mm2 (40 60 mm) ocity–time curves of honeycombs can be obtained by
and under a stress of 3.3 MPa, are used to simulate subsequent processing of the high-speed camera data,
the first-degree energy absorber of the scaled equiva- and the acceleration sensor was placed in the center of
lent model car, the initial length is 50 mm, and honey- gravity of the car body to obtain the acceleration data
combs, each with a cross-sectional area of 2400 mm2 during the collision.
Yao et al. 85
Analysis of the test results shown in Figures 17 and 18, and the test data are
This study considered nine groups of collision condi- shown in Table 3.
tions, of which four groups of test conditions are for As shown in Figure 17, after the collision of a
two-vehicle collisions and the remaining five groups of single car with a rigid wall, the vehicle has a smaller
test conditions are for a single car crashing into a rigid rebound velocity. As shown in Figure 18, the speed of
wall. Processing the high-speed camera data obtained the stationary car is slightly higher than that of the
for each vehicle test conditions, the changing process moving car after collision, and the speeds of
of honeycomb deformation and vehicle speed are the moving and stationary cars are close to half of
Figure 18. Displacement–time and velocity–time curves of the car-to-car impact test. (a) Test B1 (v ¼ 5:330 m=s), (b) test B2
(v ¼ 5:407 m=s), (c) test B3 (v ¼ 8:490 m=s), and (d) test B4 (v ¼ 9:927 m=s).
86 Proc IMechE Part F: J Rail and Rapid Transit 233(1)
1
Mv2 ¼ Fs þ Mgs ð44Þ
2
the test speed. These results are consistent with the
prior simulation results.
The acceleration–time curves of the cars in the Because
impact test are shown in Figure 19; because the
length of the first-degree energy absorber was Mgs 0:03 135 9:8
¼ ¼ 1:65375 103 0
50 mm, the acceleration curves in tests A1, A2, A3, Fs 24000
B1, B2 have one platform which represents the com- ð45Þ
pression of first-degree energy absorber. The acceler-
ation curves can be divided into two main stages in we can ignore the friction of Mgs and have
tests A4, A5, B3, B4. The first platform is the compres-
sion of the first-degree energy absorber, and the 1
Mv2A ¼ FsA ð46Þ
second platform is the compression of the second- 2
degree energy absorber. The plastic deformation plat-
form force of a honeycomb material is stable, so 1
Mv2B ¼ FsB ð47Þ
the acceleration curves are also stable. The first 2
platform acceleration is about 6 g and the second plat-
form acceleration is about 18 g for scaled model, so If we divide equation (46) by equation (47), we
the first platform acceleration is about 0.7 g and the have
second platform acceleration is about 2 g for
2
prototype. vA sA
Because accurately controlling the collision speed ¼ ð48Þ
vB sB
before collision is challenging, the numerical
Yao et al. 87
Figure 19. Acceleration–time curves of the cars in the impact test. (a) Test A1, (b) test A2, (c) test A3, (d) test A4, (e) test A5, (f) test
B1, (g) test B2, (h) test B3, and (i) test B4.
Funding
The author(s) disclosed receipt of the following financial
support for the research, authorship, and/or publication
of this article: This work was supported by the National
Natural Science Foundation of China (grant number
51675537), the National Key Research and Development
Program of China (grant number 2016YFB1200404-04),
and the Fundamental Research Funds for the Central
Universities of Central South University (grant numbers
2017zzts196, 2017zzts585).
ORCID iD
Ping Xu http://orcid.org/0000-0001-9670-9871
References
Figure 20. Sample points and fitting curve. 1. Bae HU, Yun KM and Lim NH. Containment capacity
and estimation of crashworthiness of derailment con-
tainment walls against high-speed trains. Proc
IMechE, Part F: J Rail and Rapid Transit 2018; 232:
(units: m/s). The fitting error sum square is 0.2268, 680–696.
and the mean square deviation is 9.7761; thus, the 2. Xu P, Yang C, Peng Y, et al. Crash performance and
fitting degree is 0.9768. Figure 20 shows the fitting multi-objective optimization of a gradual energy-
curve and the sample points. absorbing structure for subway vehicles. Int J Mech
We conclude that we can use a single car with a Sci 2016; 107: 1–12.
speed of v0 ¼ 0:5248v þ 0:1608 crashing into a rigid 3. Wood DP, Simms CK and Walsh DG. Vehicle-pedes-
wall to replace a moving car with a speed of v crashing trian collisions: validated models for pedestrian impact
and projection. Proc IMechE, Part D: J Automobile
into an identical car that is stationary.
Engineering 2005; 219: 183–195.
4. Li R, Xu P, Peng Y, et al. Scaled tests and numerical
Conclusions simulations of rail vehicle collisions for various train
sets. Proc IMechE, Part F: J Rail and Rapid Transit
Presently, there are increasing numbers of research stu- 2016; 230: 1590–1600.
dies on the passive safety protection of rail vehicles, and 5. Peng Y, Hou L, Yang M, et al. Investigation of the
tests of the energy absorber and whole vehicle tests are train driver injuries and the optimization design of
ongoing. However, the vehicle collision test cost is too driver workspace during a collision. Proc IMechE,
high and the test is too time consuming. The collision Part F: J Rail and Rapid Transit 2017; 231: 902–915.
between vehicles is a complex nonlinear process, and the 6. Tyrell DC. US rail equipment crashworthiness stand-
energy absorption in the process of collision is achieved ards. Proc IMechE, Part F: J Rail and Rapid Transit
2002; 216: 123–130.
by both the energy absorber and the vehicle body. In
7. Technical Committee CEN/TC 256 Railway
view of this characteristic, in order to evaluate the
Applications. BS EN 15227-2008 railway applications:
energy absorption capability of a single car, we replaced crashworthiness requirements for railway vehicle bodies.
car-to-car impact conditions with a single car impacting London: British Standard Institution, 2008.
a rigid wall, and through simulations and experiments, 8. Hault-Dubrulle A, Robache F, Drazetic P, et al. Analysis
we found that a single vehicle with a speed of of train driver protection in rail collisions: part I.
v0 ¼ 0:5248v þ 0:1608 (units: m/s) crashing into a rigid Evaluation of injury outcome for train driver in desk
wall is equivalent to a moving car with a speed of v impact. Int J Crashworthiness 2013; 18: 183–193.
crashing into an identical car that is stationary. If we 9. Karina JM. Collision dynamics modeling of crash energy
equivalently use a single vehicle that crashes into a rigid management passenger rail equipment. Medford, MA:
wall to replace the test condition of one vehicle crashing Tufts University, 2008.
10. Tyrell D and Martinez E. A train-to-train impact test of
into an identical vehicle that is stationary, the possibility
crash energy management passenger rail equipment.
of derailment of trains can be evaluated to the greatest
Volpe National Transportation Systems Center,
extent. This research provides practical support for the Massachusetts, USA, 2006.
equivalence of the collision test condition of the rail 11. Tyrell D, Jacobsen K, Martinez E, et al. Train-to-train
vehicle and provides an effective method for studying impact test of crash energy management passenger rail
rail vehicle collisions in the future. equipment: structural results. In: ASME 2006 interna-
tional mechanical engineering congress and exposition,
Declaration of Conflicting Interests Illinois, USA, 2006, pp.35–44. American Society of
The author(s) declared no potential conflicts of interest with Mechanical Engineers.
respect to the research, authorship, and/or publication of 12. Severson KJ, Parent DP and Tyrell DC. Two-car
this article. impact test of crash-energy management passenger rail
Yao et al. 89
cars: analysis of occupant protection measurements. In: 26. Lu G and Ronghua XU. Energy absorption require-
ASME 2004 international mechanical engineering con- ment for crashworthy vehicles. Foreign Rolling Stock
gress and exposition, California, USA, 2004, pp.87–96. 2006; 43: 8–13.
American Society of Mechanical Engineers. 27. Chen S, Niu W and Wen T. Simulation of urban rail
13. Martinez E, Tyrell D and Perlman B. Development of vehicle crash and factors influencing anti-climbing abil-
crash energy management designs for existing passenger ity of its anti-climber. J Mech Strength 2015; 37:
rail vehicles. In: ASME 2004 international mechanical 924–929.
engineering congress and exposition, California, USA, 28. Lowe WT, Al-Hassani STS and Johnson W. Impact
2004, pp.107–115. American Society of Mechanical behaviour of small scale model motor coaches. Proc
Engineers. Instn Mech Engrs 1972; 186: 409–419.
14. Jacobsen K, Tyrell D and Perlman B. Impact tests of 29. Kao GC and Chan GC. A scale model study of crash
crash energy management passenger rail cars: analysis energy dissipating vehicle structures. Wyle Laboratories,
and structural measurements. In: ASME 2004 interna- Alabama, USA, 1968.
tional mechanical engineering congress and exposition, 30. Barley GW and Mills B. A study of impact behaviour
California, USA, 2004, pp.97–105. American Society through the use of geometrically similar models. Proc
of Mechanical Engineers. Instn Mech Engrs 1969; 184: 26–33.
15. Lewis JH. Validation of measures to improve vehicle 31. Jiang P, Tian CJ, Xie RZ, et al. Experimental investi-
safety in railway collisions. ASME Appl Mech Div gation into scaling laws for conical shells struck by pro-
Publ AMD 1995; 210: 17–34. jectiles. Int J Impact Eng 2006; 32: 1284–1298.
16. Cleon LM, Legait J and Villemin M. SNCF structural 32. Wang M, Liu Z, Qiu Y, et al. Study on the similarity
crashworthiness design strategy: design examples of laws for local damage effects in a concrete target under
duplex TGV and XTER diesel multiple unit. In: the impact of projectiles. Shock Vib 2015; 16: 1–16.
Tyrell DC and Rubin AH (eds) Symposium on rail vehi- 33. Xu LJ, Lu XZ, Smith ST, et al. Scaled model test for
cle crashworthiness, France, 1996. collision between over-height truck and bridge super-
17. Train crashworthiness for Europe. SAFETRAIN Final structure. Int J Impact Eng 2012; 49: 31–42.
Report for the European Community under the 34. Ling L, Dhanasekar M and Thambiratnam DP.
Industrial and Materials Technologies Programme, Assessment of road-rail crossing collision derailments
Belgium, 27 July 2000. on curved tracks. Aust J Struct Eng 2017; 18: 125–134.
18. SAFETRAIN B. EURAM Project n. BE-3092. 35. Sun YQ, Cole C, Dhanasekar M, et al. Modelling and
Dynamic tests. SAFETRAIN Technical Report, analysis of the crush zone of a typical Australian pas-
Belgium. senger train. Veh Syst Dyn 2012; 50: 1137–1155.
19. SAFETRAIN B. EURAM Project n. BE-3092. 36. Ling L, Guan Q, Dhanasekar M, et al. Dynamic simu-
Mathematical modelling, Belgium. lation of train–truck collision at level crossings. Veh
20. Railway tests crash-worthy coaches. The Tribune, Syst Dyn 2017; 55: 1–22.
India, 22 February 2006. 37. Baker WE, Westine PS and Dodge FT. Similarity meth-
21. Wei Q, Wang Y and Wen B. Safety research on collision ods in engineering dynamics: theory and practice of scale
of railway locomotives and rolling stock abroad. modeling. Spartan Books; [distributed by] Hayden Book
Foreign Rolling Stock 2012; 5: 014. Co., USA, 1973.
22. Yan J, Yao S, Xu P, et al. Theoretical prediction and 38. Lu G and Yu TX. Energy absorption of structures and
numerical studies of expanding circular tubes as energy materials. Amsterdam: Elsevier, 2003.
absorbers. Int J Mech Sci 2016; 105: 206–214. 39. Yao S, Yan K, Lu S, et al. Energy-absorption optimisa-
23. Reports for crash test of China standard EMU, http:// tion of locomotives and scaled equivalent model valid-
v.ifeng.com/oxygenvideo/special/160/index.shtml#_v_ ation. Int J Crashworthiness 2017; 22: 1–12.
www3 (2017, accessed 29 October 2017). 40. Wang Z, Liu J and Hui D. Mechanical behaviors of
24. Hecht M. The crashworthiness of tramcar and LRV. inclined cell honeycomb structure subjected to compres-
Foreign Rolling Stock 2005; 42: 39–41. sion. Compos Part B Eng 2017; 110: 307–314.
25. Lu G. Energy absorption requirement for crashworthy 41. Wang Z, Tian H, Lu Z, et al. High-speed axial impact
vehicles. Proc IMechE, Part F: J Rail and Rapid Transit of aluminum honeycomb-experiments and simulations.
2002; 216: 31–39. Compos Part B Eng 2014; 56: 1–8.