Local Buckling Strength and its Statistical Information of

Normal and Bridge High Performance Steel Plates

Journal:
Manuscript ID:
Theme:
Date Submitted by the Author:
Complete List of Authors:

Type of Structure:
Material and Equipment:
Other Aspects:

18th IABSE Congress on Innovative Infrastructures

Seoul-0132-2012.R1
Structures and Materials - extending the limits
28-Feb-2012
Dang, Viet Duc; Saitama University,
Okui, Yoshiaki; Saitama University, Civil and Environmental Engineering
Hagiwara, Koichi; Saitama University, Civil and Environmental Engineering
Nagai, Masatsugu; Nagaoka University of Technology, Civil and
Environmental Engineering
Bridges
Steel

Page 1 of 10

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Abstract for
18th IABSE Congress 2012, Seoul, Korea
Innovative Infrastructures Toward Human Urbanism
Date: 14/09/2011
Contact Author: Dang Viet, DUC, dangviet.duc@gmail.com

Local Buckling Strength and its Statistical Information of Normal and Bridge
High Performance Steel Plates
Dang Viet Duc, Yoshiaki Okui, Koichi Hagiwara, Masatsugu Nagai

The bridge high performance steels with advantages of high yield strength and good weldability have
been standardized as SBHS500 and SBHS700 in Japanese Industrial Standard (JIS) since 2008. These
SBHS steels exhibit different inelastic behavior from conventional steels, such as almost no yield
plateau and greater yield-to-tensile strength ratio. Hence, there is a need to examine the applicability of
SBHS to steel bridge structure and to develop design equations for SBHS.
In this paper, the compressive strengths of 4 edge simply supported normal and SBHS steel plates are
examined through parametric study employing elasto-plastic finite element analyses. In the parametric
study, 4 conventional steel grades SM400, SM490, SM490Y, SM570 and 2 SBHS steel grades
SBHS500 and SBHS700 were considered, and compressive strength of steel plates were evaluated at 10
slenderness parameter R values within the range from 0.4 to 1.4. The effect of different inelastic
behavior of different steel grades, the residual stress and the initial deflection, and the slenderness
parameter on the normalized compressive strength is discussed. It is found that the normalized
compressive strength of SBHS700 steel plates is about 6% greater than that of the SM400 steel plates
for R 0.7. The numerical analysis results with the conservative residual stress and initial deflection
levels show that the load-carrying capacity curve of Japan Specification for Highway Bridges (JSHB) is
unconservative for 0.45<R<0.75 and over conservative for R>0.8
In the second part of this paper, the mean value and standard deviation of the normalized compressive
strength is evaluated by using the response surface method and Monte Carlo simulation. To this end,
first, the response surface of the normalized compressive strength is identified based on 114 numerical
results for different residual stress and initial defection, where the response surface is assumed as a
simple algebraic function of the residual stress and initial deflection. Then, Monte Carlo simulation is
carried out to evaluate statistical distribution of the compressive strength. In this simulation, only the
residual stress and initial defection are considered as stochastic variables, and the probabilistic
distribution of residual stress and initial deflection are collected from previous experimental studies.
Comparing with the standard deviation obtained from experimental results, the standard deviation based
on this numerical simulation exhibits about half of that in a range of 0.6<R<1.2 and more reasonable
tendency.
Keywords: bridge high performance steels, compressive strength, residual stress, initial deflection, local
buckling

Page 2 of 10

Local Buckling Strength and its Statistical Information of Normal and Bridge
High Performance Steel Plates
Viet Duc DANG
PhD student, Dep. of Civil and
Environmental Eng., Saitama University,
Saitama Prefecture, Japan

Yoshiaki OKUI
Professor, Dep. of Civil and Environmental
Eng., Saitama University, Saitama
Prefecture, Japan

dangviet.duc@gmail.com

okui@mail.saitama-u.ac.jp

Koichi HAGIWARA
Master student, Dep. of Civil and
Environmental Eng., Saitama University,
Saitama Prefecture, Japan

Masatsugu NAGAI
Professor, Dep. of Civil and Environmental
Eng., Nagaoka University of Technology,
Nagaoka Prefecture, Japan

alexander.f.f.9@gmail.com

nagai@nagaokaut.ac.jp

1.

Introduction

Since 2008, the bridge high performance (SBHS) steels with advantages of high yield strength and
good weldability have been standardized as SBHS500 and SBHS700 in Japanese Industrial
Standard (JIS) [1]. However, SBHS steels exhibit different inelastic behaviour from conventional
steels, such as almost no yield plateau and greater yield-to-tensile strength ratio. Hence, it is
necessary to examine the applicability of the current compressive strength design equation of JSHB
[2] to steel plates with new steel grades.
The recent design specifications as Eurocode and AASHTO tend trend towards the partial safety
factor method. Regarding the predicting model on compressive strength of steel plates, in order to
determine the safety factors and the nominal compressive strength, statistical information, such as
the mean value and standard distribution of the compressive strength, is necessary. The current
study based on the Monte Carlo method and response surface to obtain probability distribution of
normalized compressive strength. The data of plate initial imperfections reported in [3] is
considered as the base of probabilistic characteristics of random inputs.

2.

Examination of the JSHB design equation

1

z
a

Simply supported edges

rc

0.8
y

W0

u/ y

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

0.6
0.4
0.2

JSHB

SM400

SM490

SM490Y

SM570

SBHS500

SBHS700
0

x y
z(x,y) = W0 sin sin
a b

Fig. 2: Idealized residual stress distribution

and sinusoidal initial deflection surface

0.3

0.6

0.9

1.2

1.5

Fig. 1: Normalized compressive strength of plate

with 6 steel grades in the case W0/b=1/400 and
rc/y=0.23

For the plate simulation, the steel material model applies assumptions of isotropic elasto-plastic
hardening, Von Mises yield surface and strain hardening hypothesis. The assumed uniaxial stressstrain relation of 6 steel grades is idealized from actual experiments. The four edges the of plate
model are assigned as simple supports. The residual stress and initial deflection are idealized as a
stress distribution and sinusoidal surface with surface function shown in Fig.1. The compression is
applied in the plate model by displacement control method. The ABAQUS S4R shell elements are

Page 3 of 10

applied for meshing the plate model with mesh size 30x30 elements.
The normalized compressive strength of steel plates (u/y) are compared between the mean value
of residual stress and the initial deflection from measurement data reported in [14] with rc/y=0.23,
W0/b=1/400. As shown in Fig.2 the compressive strengths of steel plates with six grades are quite
similar. The most significant difference occurs at R 0.7 and R 0.4, the normalized compressive
strength of SBHS700 steel plates (maximum value among six steel grades) is about 6% greater than
that of the SM400 steel plates (minimum value among 6 steel grades).For R > 0.4, the compressive
strength of SBHS steel plates with greater YR value is greater than that of the normal steel plate
with lower YR value. These results show the difference from conclusion in [4].

3.

Compressive strength scatterness

The histogram of compressive strengths of steel

plates is obtained by the application of a response
surface and a large number of random inputs of
0.8
residual stress and initial deflection. The
convergent mean and standard deviation values of
JSHB
0.6
compressive strength of steel plates are obtained
by processing 10000 random input couples of
M ean results
residual stress and initial deflection. In Fig.3, the
0.4
mean values of the normalized compressive
Fukumoto M curve
strength of steel plates with error bar equal to 2
0.2
Fukumoto M -2S curve
times of standard deviation values are plotted
along with the current JSHB design equation,
0
mean (M) curve and the mean minus 2 standard
0.0
0.2
0.4 0.6
0.8 1.0
1.2 1.4
deviation (M-2S) curve proposed in [3] .
R
As shown in Fig.3, the mean values of normalized
Fig. 3: Comparison between current study,
compressive strength of steel plates are similar to
JSHB [2], and Fukumoto el al [3] results
the mean curve reported in [3] which was
proposed from test results. Within the range 0.65R0.85, the mean values of the normalized
compressive strength of the current study are slightly greater than that of the mean curve reported in
[14]. The reason is that the current study doesnt consider steel plates with W0/b>1/150 and within
the mentioned range of R the influence of initial deflection on compressive strength is more
significant than other ranges. Also seen in Fig.3, the M-2S curve proposed in [14] is too
conservative if compared to corresponding results of the current study.
1

u/ y

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

4.

Conclusions

The mean values obtained in the current study are similar to the mean equation proposed in
Fukumoto et al (1984) [3] but slightly greater within the range 0.7<R<0.9.
The standard deviation values of compressive strength of the current study exhibits about half of
which was reported in [3] within a range of 0.6<R<1.2 and has a more reasonable tendency.
For R>0.4, for the same levels of R and initial imperfections, the compressive of steel plates is
similar, however; compressive strength of steel plates with SBHS grades with greater YR quantity
is slightly greater than that of normal grades, show the difference from conclusion in [4].

References
[1]
[2]
[3]
[4]

Japanese Industrial Standard, JIS G 3140, Higher yield strength steel plates for bridges,
2008
Japan Road Association, Specifications for Highway Bridges part II. Steel Bridges, 2002
FUKUMOTO. Y., and ITOH, Y., Basic compressive strength of steel plates from test data,
Proc. of JSCE No.334/I-1, 1984.
FUKUMOTO. Y., New constructional steels and structural stability, Engineering
structures, Vol.18, 1996, pp. 786-791

Page 4 of 10

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Local Buckling Strength and its Statistical Information of Normal and Bridge
High Performance Steel Plates
Viet Duc DANG
PhD student, Dep. of Civil and
Environmental Eng., Saitama University,
Saitama, Japan

Yoshiaki OKUI
Professor, Dep. of Civil and Environmental
Eng., Saitama University, Saitama, Japan
okui@mail.saitama-u.ac.jp

dangviet.duc@gmail.com

Koichi HAGIWARA
Master student, Dep. of Civil and
Environmental Eng., Saitama University,
Saitama, Japan

Masatsugu NAGAI
Professor, Dep. of Civil and Environmental
Eng., Nagaoka University of Technology,
Nagaoka, Japan

alexander.f.f.9@gmail.com

nagai@nagaokaut.ac.jp

Summary
The current Japanese design equation for load-carrying capacity of compressive steel plates is
examined for four conventional steel grades and new bridge high performance steel grades
SBHS500 and SBHS700. The analysis results of FEM plate models with initial imperfections show
that the current Japanese design specification is un-conservative within the range 0.5<R<0.75 and
over-conservative in the range R>0.8, where R is the slenderness parameter. Statistical distribution
of the normalized compressive strength is obtained by means of Monte Carlo simulation in
combination with the response surface which is obtained from sufficient number of FEM plate
model analyses. Both initial out-of-plane displacement and residual stress are considered as sources
of variability. The mean values of normalized compressive strength in this study are similar to those
obtained from experimental tests [14]. The standard deviation of the current study exhibits about
half of the experimental results [14] within the practical range 0.6<R<1.2.
Keywords: bridge high performance steels, compressive strength, residual stress, initial deflection,
local buckling.

1.

Introduction

Box columns and box plate girders consisting of unstiffened steel plates are widely used in bridge
structures. The local buckling strength of the steel plates frequently governs the load-carrying
capacity of these structural elements.
The current compressive strength design equation for unstiffened plates in Japanese Specifications
for Highway Bridge (JSHB) version 2002 [2] has been originally proposed in 1980 [3]. This
equation was based on experimental data for normal steel with yield strengths mainly less than 450
MPa. The bridge high performance steels, which poses high yield strength and good weldability,
have been standardized since 2008 as SBHS500 and SBHS700 in Japanese Industrial Standard (JIS)
[4]. However, SBHS steels exhibit different inelastic behaviour from conventional steels, such as
almost no yield plateau and greater yield-to-tensile strength ratio. Hence, it is necessary to examine
the applicability of the current compressive strength design equation of JSHB to steel plates with
new steel grades.
Regarding the compressive strength design equation of JSHB, Usami and Fukumoto [5], Usami [6],
and Kitada et al. [7] show that the design equation is un-conservative within the range 0.5<R<0.75
(intermediate range) and over-conservative in the range R>0.8 (slender range), in which R is the
slenderness parameter. However, studies [5], [6] consider only the normal (SM490Y) steel plates
and they employed the perfectly elstoplastic assumption for modeling the inelastic behavior of steel
material.
For theses reasons, the compressive strength design equation of JSHB need to be examined and

Page 5 of 10

developed for SBHS steel grades. However all the referred studies were based on the deterministic
method. The recent design specifications trend towards the partial safety factor method [9], in
which a safety factor separates into individual causes, such as the variability on material property
and the confidence of strength prediction method. The partial safety factor method with probabilitybased partial factors has been employed in Eurocode [10], [11] and AASHTO [12] as well. In order
to determine the safety factors and the nominal compressive strength, statistical information, such
as the mean value and standard distribution of the compressive strength, is necessary. Komatsu and
Nara [13] carried out a number of FE analyses of steel plates with considering initial deflection
based on collected measurement data from actual steel bridges to obtain the mean value and
standard deviation of compressive strength. However, this study considered only residual stress
rc/y=0.3 as a deterministic quantity, where rc is the compressive residual stress. Fukumoto and
Itoh [14] proposed the mean (M) and mean minus twice standard deviation (M-2S) curves of the
compressive strength on the basis of a database on single plate and box column compression test
data. The M and M-2S curves also show that the current JSHB design equation for steel plates
under compression is un-conservative within intermediate range and over-conservative in slender
range. Fukumoto and Itoh [14] also reported the statistical distribution of residual stress and initial
deflection. However, this study also considered the steel plates with initial deflection W0/b > 1/150,
where W0 and b are the maximum initial deflection and plate width, respectively.
This paper intends to examine the current JSHB design equation of steel plate compressive strength
for normal and SBHS steels, and to evaluate the mean and standard deviation of compressive
strength.

2.

Examination of JSHB design equation

2.1

Plate properties
Four normal steel grades SM400, SM490, SM490Y,
SM570 and two bridge high performance steel
SBHS500, SBHS700 are considered in the current study.
Fig. 1 shows yield strengths and slenderness parameters
R considered in the FE analyses, where R is defined by

800

Yield strength (MPa)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

600

400

R=
200
SM400

SM490

SM490Y

SM570

SBHS500

SBHS700

0
0.3

0.5

0.7

0.9

1.1

1.3

1.5

b y 12(1 2 )
.
t E
2k

(1)

where b, t, y, E, , and k=4.0 stand for the plate width,

thickness, yield strength, elastic modulus, poison ratio,
and buckling coefficient, respectively. The aspect ratio
of all steel plates is assigned to 1.

Fig. 1: Slenderness parameter and

yield strength in FE analyses
2.2

FEM model

Nonlinear FE analysis considering both material and geometric nonlinearity is conducted. PrandtlReuss equation is employed to model the steel plasticity. The idealized uniaxial stress-strain
relationships used in the numerical analyses are shown in Fig.1.
Regarding the boundary condition, all four edges of a plate model are assigned as simple supports.
Figure 3 shows the distributions of residual stress and initial deflection assumed in the FE analysis.
The probabilistic distributions of residual stress and initial deflection are based on measurement
data reported in [14].
The displacement control method is used to apply the compressive stress. ABAQUS S4R shell
elements are used for plate FE model with mesh size of 30x30 elements.

Page 6 of 10

800

z
a

Simply supported edges

1 1 (MPa)

600

rc
400

SBHS700

SBHS500

SM570

SM490Y

SM490

SM400

0.05

0.1

0.15

W0

0
0

200

x y
sin
a b

z(x,y) = W0 sin

0.2

1 1

Fig. 2: Idealized stress-strain

relations of steel grades considered
in current study

Fig. 3: Idealized residual stress

distribution and sinusoidal initial
deflection surface

2.3
Comparison of FE results with experimental results
The normalized compressive strengths obtained from the FE analyses as well as past experimental
results [1, 15] are plotted in Fig. 4 as a function of R. In these analyses, an initial deflection W0/b =
1/150 and a residual stress rc/y = 0.4 are considered as a conservative assumption. As shown in
Fig.4, the FEM results lay the lower bound of experimental results, which corresponds to the
conservative assumption.
1.2

1
1

0.8

u/ y

0.8

u / y

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

0.6
JSHB
Dorman Long
Chin
SM400
SM490Y
SBHS500

0.4
0.2

0.6
0.4

Rasmussen
Harrison
Moxham
SM490
SM570
SBHS700

0.2

JSHB

SM400

SM490

SM490Y

SM570

SBHS500

SBHS700

0
0.0

0.5

1.0

Fig. 4: Comparison of current study

results to previous test results

1.5

0.3

0.6

0.9

1.2

1.5

Fig. 5: Normalized compressive strength

of plate with 6 steel grades in the case
W0/b=1/400 and rc/y=0.23

2.4
Compressive strength of different steel grades
In this section, the normalized compressive strengths of different steel grades are compared in the
case of mean values of the residual stress and the initial deflection. The mean values of normalized
residual stress and initial deflection are obtained as rc/y=0.23 and W0/b=1/400, respectively from
the measurement data reported in [14]. In the evaluation of the mean value of the normalized initial
deflection, the measurement data for W0/b>150 are excluded owing to an allowable fabrication
upper limit in JSHB.
As shown in Fig.5, the compressive strengths of steel plates with 6 steel grades are quite similar in
the whole range of R. The largest difference occurs at R 0.7 and R 0.4, and the normalized
compressive strength of SBHS700 steel plates (maximum value among 6 steel grades) is about 6%

Page 7 of 10

greater than that of the SM400 steel plates (minimum value among 6 steel grades). For R > 0.4, the
compressive strength of SBHS steel plates with larger YR value is greater than that of normal steel
plate with lower YR value.

3.

Compressive strength scatterness

3.1

Response surface

numerical results
1.1
1

u / y

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

0.9
0.8

0.7
0

rc / y

0.5
0.005

W0/b

0.01

0.015

To obtain the probability distribution of the

compressive strength, Monte Carlo method is
employed. However, to obtain a convergent result
in the Monte Carlo simuration, it is essential to
implement a large number of deterministic
analyses, and accordingly it would take long time.
Hence, to overcome this problem, a response
surface of the normalized compressive strength,
which is an approximate algebraic function of the
initial deflection and residual stress, is instead of
the deterministic analyses. The current study
employs ten response surfaces corresponding to
the 10 considered R values.

Fig. 6: The response surface shape

presented along with FEM results for the
case R = 0.8
The response surface is expressed as a simple algebraic function,
u = p 00 + p10 r + p 01W0 + p 20 r + p11 r W0 + p 02W0 + p 21 r W0 + p12 r W0 + p 30 r + p 03W0
2

(2)

where u = u / y , W0 = W0 / b and r = rc / y . The constants pij in Eq.(2) were determined from a set
of 114 deterministic FE results for each steel grade and R value by using the least square method.
Fig.6 shows an obtained response surface along with FE numerical results for R=0.8. All constants
of ten response surfaces are presented in Table 1. The obtained response surfaces show good fit for
numerical results in the cases R0.7 with the coefficient of determination (R-square) > 95%. For
R<0.7, the R-square values become slightly lower due to the influence of the hardening of high
strength steels (SM570, SBHS500, SBHS700).
Table 1: Constant values of 10 Response surfaces
R value
0.40
0.50
0.60
0.70
0.80
0.92
1.04
1.16
1.28
1.40

p00
1.098
1.034
1.012
1.037
1.047
0.963
0.850
0.757
0.697
0.650

p01
-40.22
-15.15
-1.72
-34.54
-65.98
-40.44
-23.31
-12.66
-10.08
-7.44

p02
5442.0
1683.0
-1894.0
3169.0
8382.0
3081.0
1188.0
377.6
367.4
229.3

p03
-248100
-90660
100500
-135000
-389100
-112100
-34700
-11430
-12430
-8387

p10
0.007
-0.012
-0.087
-0.234
-0.584
-0.937
-0.788
-0.567
-0.435
-0.343

p11
-2.320
-8.400
-21.150
-32.100
-0.520
57.210
47.270
25.720
23.330
18.580

p12
-48.4
-42.4
507.9
1465.0
1309.0
-891.1
-769.2
46.8
-269.2
-208.5

p20
-0.007
0.069
0.284
0.568
1.006
1.368
1.169
0.856
0.617
0.462

p21
2.25
6.72
11.01
9.74
-15.13
-35.91
-29.12
-20.27
-15.11
-11.63

p30
0.000
-0.057
-0.198
-0.342
-0.492
-0.632
-0.556
-0.419
-0.288
-0.211

Page 8 of 10

3.2
Stochastic inputs of initial imperfections
In the Monte Carlo simulation, the probabilistic distribution of the initial deflection is assumed as
the Weibull distribution, and the residual stress as the Lognormal distribution as shown in Fig.7 and
8 [14]. In this figure, the generated random variables in the Monte Carlo simulation are also plotted
as the histogram chart. The generated normalized initial defections more than W0/b>1/150 are
excluded in the simulation due to an allowable fabrication upper limit in JSHB.
0.10

0.25

0.08
0.07
0.06

Relative Frequancy

Weibull dist.
=0.00280
=0
m=1.37
=0.00233
=0.00156

0.09

Relative Frequancy

0.05
0.04
0.03
0.02

Lognormal dist.
=0.23
=0.145

0.20
0.15
0.10
0.05

0.01
0
0

W0/b

0
0

7
x 10

-3

0.2

0.4

/
rc

Fig. 7: Generated random input of initial

deflection

0.6

0.8

1.0

Fig. 8: Generated random input of

residual stress

3.3
Results of Monte Carlo simulation
The probabilistic distribution of compressive strengths is obtained by application of the response
surface and a large number of random inputs of residual stress and initial deflection. The converged
mean and standard deviation are obtained by processing 10000 random input couples of residual
stress and initial deflection. The histogram of compressive strength with 10000 random input
couples in the case of R=0.8 is shown in Fig.9. Results of 10 convergent mean and standard
deviation of compressive strengths are presented in Table 2.
0.1

Relative Frequancy

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

0.08
0.06
0.04
0.02
0
0.5

0.6

0.7

0.8
0.9
/
u

1.0

1.1

1.2

Fig. 8: Probability distribution of compressive strength for R=0.8

Table 2: Mean and standard deviation of normalized compressive strength

R value

0.4

0.5

0.6

0.7

0.8

0.92

1.04

1.16

1.28

1.4

Page 9 of 10

M value
S value

1.039

1.006

0.982

0.938

0.862

0.766

0.701

0.653

0.617

0.586

0.0277

0.0157

0.0212

0.0413

0.0515

0.0399

0.0294

0.0216

0.0171

0.0145

In Fig.9, the mean values of normalized compressive strength with error bar equal to two times the
standard deviation are plotted along with the current JSHB design equation, mean (M) curve and
mean minus 2 standard deviation (M-2S) curve proposed in [14] .
0.20

0.20
1

Standard deviation

Current study
0.15

u/ y

0.8

u/ y

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

JSHB

0.6

M ean results

0.4

Fukumoto M curve

From Fukumoto Exp.

0.10

0.05

0.2

0.15

0.10

0.05

Fukumoto M -2S curve

0.00

0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.00
0.3

0.6

0.9

1.2

1.5

Fig. 9: Comparison between current

study, JSHB [2], and Fukumoto el al.
[14] results
As shown in Fig.9, the mean values of normalized compressive strength are similar to the mean
curve reported in [14], which was proposed from test results. Within 0.65R0.85, the mean values
of the current study are slightly greater than that reported in [14]. One of the possible reasons is that
the current study does not consider steel plates with W0/b>1/150 and the influence of initial
deflection on compressive strength is more significant within the mentioned range of R than other
ranges. Also seen in Fig.9, the M-2S curve proposed in [14] is too conservative if compared to
corresponding results of the current study.
A comparison of standard deviation of compressive strengths obtained in the current study and
previous study [14] is presented in Fig.10. Within the practical range 0.6<R<1.2, the standard
deviations of the current study exhibit about half of reported values in [14]. As seen in the figure,
the standard deviations obtained by the current study have a clearer tendency, and attained to the
maximum value at R0.8. Within the range 0.7<R<0.9, the ultimate state of compressive plates is
elastoplastic buckling, and accordingly the compressive strength is significantly influenced by both
residual stress and initial deflection. For R>0.9, the standard deviation decreases, and the elastic
buckling becomes dominant. The residual stress has almost no effect on the compressive strength.
For R < 0.7, the compressive strength tends to attain the yield strength and is mainly influenced by
initial deflection. In particular for R<0.5, the hardening behavior of high strength steel (SM570,
SBHS500 and SBHS700) starts to have a significant effect on the compressive strength.

4.

Conclusions

The mean values obtained in the current study are similar to those proposed in [14], but slightly
greater within the range 0.7<R<0.9. The standard deviation of compressive strength obtained in this
study is about half of values reported in [14] within a range of 0.6<R<1.2. The M-2S curve
proposed in [14] is too conservative if compared to corresponding values obtained in the current
study. The results of M-2S of the current study show that the current load-carrying-capacity JSHB
design equation of steel plates is un-conservative within the range 0.5 < R < 0.8 and overconservative for R > 0.85.

0.3

0.6

Fig. 10: Co
deviation v
study and r

Page 10 of 10

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Within the range of slenderness parameter 0.4R1.4, the normalized compressive strengths of
normal steel and SBHS steel are similar for the same levels of R and initial imperfections. However,
compressive strength of steel plates with SBHS grades with greater YR is slightly greater than that
of normal steel grades.

References
[1]

DWIGHT J. B., MOXHAM K. E., Welded steel plates in compression, The Structural
Engineer, Vol. 47, No. 2, 1969, pp. 49-66.

[2]
[3]

Japan Road Association, Specifications for Highway Bridges part II. Steel Bridges, 2002
Japan Road Association, Specifications for Highway Bridges part II. Steel Bridges, 1980

[4]

Japanese Industrial Standard, JIS G 3140, Higher yield strength steel plates for bridges,
2008

[5]

USAMI. T., and FUKUMOTO. Y., Deformation analysis of locally buckled steel
compression members, Journal of Constructional Steel Research, Vol.13, 1989, pp. 111135.

[6]

USAMI. T., Effective width of locally buckled plates in compression and bending, Journal
of Structural Engineering, ASCE, Vol. 119, No. 5, 1993, pp. 1358-1373.
KITADA. T., YAMAGUCHI. T., MATSUMURA. M., OKADA. J., ONO. K. and OCHI, N.,
New technology of steel bridge in Japan, Journal of Constructional Steel Research, Vol.58,
2002, pp. 21-70.
FUKUMOTO Y., New constructional steels and structural stability, Engineering
structures, Vol.18, 1996, pp. 786-791.

[7]

[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]

International Standard ISO 2394, General principles on reliability for structures, Third
edition, 1998
CEN, Eurocode3, Design of Steel Structures, Part 1-5, plated structural elements, 2004
CEN, Eurocode4, Design of Composite Steel and Concrete Structures, Part 2, General
rules and rules for bridges, 1994
American Association of State Highway and Transportation Officials, AASHTO LRFD
bridge design specification Fourth edition, 2007
KOMATSU S., and NARA. S., Statistical study on steel plate members, Journal of
Structural Engineering, ASCE, Vol. 109, No. 4, 1983.
FUKUMOTO Y., and ITOH, Y., Basic compressive strength of steel plates from test data,
Proc. of JSCE No.334/I-1, 1984.
RASMUSSEN K. J. R., and HANCOCK. J. G., Plate slenderness limits for high strength
steel sections, J. Construct. Steel Research. Vol.23, 1992, pp. 73-96