www.springer.com/journal/13296
Abstract
The probabilistic distributions of buckling strengths for compressive plates of normal and bridge high-performance steels
were obtained through numerical analyses in order to develop a nominal design strength and a corresponding safety factor. In
the numerical analyses, Monte Carlo simulation was used in combination with the response surface method to reduce the effort
associated with the finite element analyses. For each value of the slenderness parameter R, a response surface of the normalized
local bucking strength was determined based on the results of 114 finite element analyses using different residual stresses and
initial defections. The response surface is approximated as a simple algebraic function of the residual stress and the initial
deflection. Monte Carlo simulation is then carried out in order to evaluate the probabilistic distribution of the local bucking
strength. The mean values obtained in the present study approach those of a mean curve proposed based on experiments. The
standard deviation of the present study was approximately half that obtained based on experimental results in the range of 0.6
<R<1.2.
Keywords: bridge high-performance steels, local bucking strength, residual stress, initial deflection, local buckling
1. Introduction
Structural parts constructed from unstiffened plates,
such as box columns, box chord members in trusses, and
flanges of box girders, are widely used in steel structures.
In the design process of these structural parts, which are
under compression, load carrying capacity is frequently
governed by the local buckling strength (LBS) of the
constitutive unstiffened steel plates.
The current design equation of the Japanese Specifications
for Highway Bridges (JSHB, 2002) regarding the LBS of
compressive steel plates was originally proposed in the
1980 version of the specifications (Japan Road Association,
1980). Usami and Fukumoto (1989), Usami (1993), and
Kitada et al. (2002) demonstrated that the LBS design
equation in the JSHB is not conservative for 0.5<R<0.75
(intermediate range) and is overly conservative for R>
Note.-Discussion open until February 1, 2014. This manuscript for
this paper was submitted for review and possible publication on May
14, 2013; approved on August 22, 2013.
KSSC and Springer 2013
*Corresponding author
Tel: +81-48-858-3849; Fax: +81-48-858-9419
E-mail: email dangviet.duc@gmail.com
(1)
Dang Viet Duc et al. / International Journal of Steel Structures, 13(3), 000-000, 2013
Probabilistic Distributions of Plate Buckling Strength for Normal and Bridge High-performance Steels
Step 4
The LBS is evaluated by means of the response surface
for the generated values of initial deflection and residual
stress.
Step 5
The termination of the random simulation process is
decided based on the verification of the convergence of
the LBS probabilistic characteristics, such as the mean
value and standard deviation. The convergence verification
process is described in detail in Subsection 5.1.
4. Deterministic Analysis
Before conducting the probabilistic simulation, a series
of deterministic finite element analyses of four-edge simply
supported compressive plates is carried out for the purpose
of comparison with reported experimental data. Furthermore,
these numerical results are used to determine the response
surfaces.
Dang Viet Duc et al. / International Journal of Steel Structures, 13(3), 000-000, 2013
SHBS500
SM570
SM490Y
SM490
SM400
/y
/y
/y
/y
/y
/y
/y
/y
e/ey
s/sy
/y
/y
1
11
25
1
1.06
1.06
1
3
15
32
45
1
1
1.1
1.13
1.13
1
3
15
32
50
1
1
1.12
1.17
1.17
1
10
23
50
100
1
1
1.14
1.28
1.28
1
12
25
70
130
1
1
1.16
1.36
1.36
1
12
32
90
200
1
1
1.26
1.47
1.47
Probabilistic Distributions of Plate Buckling Strength for Normal and Bridge High-performance Steels
rc/y
W0/b
0.23
0.145
0.0025
0.0019
5. Response surface
The response surface method (Guan and Melchers,
2001; Gasper et al., 2012) is used to reduce the excessive
computational time for Monte Carlo simulation of the
LBS. A cubic response function is used to approximate
the dependency of the normalized LBS on the normalized
initial defection and residual stress:
Figure 8. FE analysis results (r= 0.4 and W0=1/150) for
SM400 and SBHS700 along with previously reported
experimental results.
cr = cr(x1, x2)
i j
pijx1x2
(2)
i, j = 0
i+j3
Dang Viet Duc et al. / International Journal of Steel Structures, 13(3), 000-000, 2013
Table 3. Response surface coefficients considering all steel grades for individual R values
R value
p00
p01
p02
p03
p10
p11
p12
p20
p21
p30
0.40
0.50
0.60
0.70
0.80
0.92
1.04
1.16
1.28
1.40
1.098
1.034
1.012
1.037
1.047
0.963
0.850
0.757
0.697
0.650
-40.22
-15.15
-1.72
-34.54
-65.98
-40.44
-23.31
-12.66
-10.08
-7.44
5442.0
1683.0
-1894.0
3169.0
8382.0
3081.0
1188.0
377.6
367.4
229.3
-248100
-90660
100500
-135000
-389100
-112100
-34700
-11430
-12430
-8387
0.007
-0.012
-0.087
-0.234
-0.584
-0.937
-0.788
-0.567
-0.435
-0.343
-2.320
-8.400
-21.150
-32.100
-0.520
57.210
47.270
25.720
23.330
18.580
-48.4
-42.4
507.9
1465.0
1309.0
-891.1
-769.2
46.8
-269.2
-208.5
-0.007
0.069
0.284
0.568
1.006
1.368
1.169
0.856
0.617
0.462
2.25
6.72
11.01
9.74
-15.13
-35.91
-29.12
-20.27
-15.11
-11.63
0.000
-0.057
-0.198
-0.342
-0.492
-0.632
-0.556
-0.419
-0.288
-0.211
0.4
0.5
0.6
0.7
0.8
0.92
1.04
1.16
1.28
1.4
R-squared
0.702
0.837
0.920
0.945
0.964
0.964
0.959
0.956
0.961
0.964
Probabilistic Distributions of Plate Buckling Strength for Normal and Bridge High-performance Steels
Table 5. Means and standard deviations obtained from Monte Carlo simulations for different R values
R value
0.4
0.5
0.6
0.7
0.8
0.92
1.04
1.16
1.28
1.4
value
value
1.039
0.0277
1.006
0.0157
0.982
0.0212
0.938
0.0413
0.862
0.0515
0.766
0.0399
0.701
0.0294
0.653
0.0216
0.617
0.0171
0.586
0.0145
Dang Viet Duc et al. / International Journal of Steel Structures, 13(3), 000-000, 2013
(3)
cr
cr, i cr, i
Var(cr) --------- Var(xi) --------------------- xi
2
2
i=1
(4)
i=1
(5)
Probabilistic Distributions of Plate Buckling Strength for Normal and Bridge High-performance Steels
(7)
0.4
1.039
1.05
1.05
1.07
0.5
1.006
1.03
1.03
1.04
0.6
0.982
1.04
1.04
1.05
0.7
0.938
1.08
1.09
1.11
0.8
0.862
1.11
1.13
1.16
0.92
0.766
1.09
1.11
1.14
1.04
0.701
1.07
1.09
1.11
1.16
0.653
1.06
1.07
1.08
1.28
0.617
1.05
1.06
1.07
1.4
0.586
1.04
1.05
1.06
10
Dang Viet Duc et al. / International Journal of Steel Structures, 13(3), 000-000, 2013
7. Conclusion
The mean results of the LBS obtained from approximate
random simulation are similar to the mean curve reported
References
Bijlaard, F., Sedlacek, G., Muller, C., and Trumpf, H. (2003).
Unified European design rules for steel and composite
structures. International Journal of Steel Structures, 3,
pp. 117-125.
Dassault (2008). Abaqus Theory Manual v.6.8-EF. Dassault
Systemes Simulia Corp., the place of publication.
AASHTO (2007). AASHTO LRFD bridge design specifications.
4th Ed., American Association of State Highway and
Transportation Officials.
Eurocode 3 (2004). Design of steel structures. Parts 1-5.
Plated structural elements. European Committee for
Standardization (CEN), Brussels, Belgium.
Eurocode 4 (1994). Design of composite steel and concrete
structures. Part 2. General rules and rules for bridges.
European Committee for Standardization (CEN), Brussels,
Belgium.
Dwight, J. B. and Moxham, K. E. (1969). Welded steel
plates in compression. The Structural Engineer, 47(2),
pp. 49-66.
Probabilistic Distributions of Plate Buckling Strength for Normal and Bridge High-performance Steels
11