Technical Report 1 - Version 3

Test Specimen Setup

Andr Biscaya da Graa*
University of Lisbon
November 30, 2016

R. Rovisco Pais 1, 1049-001 Lisboa, Portugal. E-mail: [email protected]

Contents
1

Introduction ..................................................................................... 3

Test Specimen ................................................................................. 3

2.1

Steel Grade ............................................................................................. 3

2.2

General Layout....................................................................................... 4

2.2.1

Flanges ............................................................................................ 4

2.2.2

Web ................................................................................................. 4

2.2.1

Longitudinal Stiffeners .................................................................... 5

2.2.2

Load Spreading Devices .................................................................. 5

Section Properties ............................................................................ 5

3.1

Gross Section .......................................................................................... 5

3.2

Effective Section ..................................................................................... 6

3.2.1

EN1993-1-5 Formulation ................................................................. 6

3.2.1

Numeric Formulation ...................................................................... 7

FEM Simulation ............................................................................... 8

4.1

Equivalent Geometric Imperfection ....................................................... 8

4.2

Choice of the compression load .............................................................. 9

4.3

Force-Displacement Results ................................................................... 9

Discussion...................................................................................... 11
5.1

M, V, N Interaction ............................................................................. 11

5.2

Previous research.................................................................................. 12

5.3

Codified Methods ................................................................................. 14

5.3.1

BS 5400-3:2000 .............................................................................. 14

5.3.2

EN1993-1-5 .................................................................................... 17

References ..................................................................................... 19

1 Introduction
The scope of this document is to design and preliminary assess the ultimate limit
state behaviour of steel plate girder beams subjected to both vertical loads and
horizontal compressions by means of experimental testing.
Consequently, two simply supported I-girders subjected to mid-span point loading and different levels of compression force, namely GL (Girder Low) and GH
(Girder High) are to be studied.
Girders will be provided with 2 web panels having different longitudinal stiffener
arrangements, thus each girder will be used to perform 2 loading tests, namely
GL_1S (GirderLow_1Stiffener), GL_2S (GirderLow_2Stiffener), GH_1S (GirderHigh_1Stiffener) and GH_2S (GirderHigh_2Stiffener), with low to moderate,
and high compressions, respectively.
Prior to the vertical load, compression will be introduced using prestressing bars
positioned in the middle of the girder depth and anchored on load spreading devices
(Figure 1).

2 Test Specimen

Figure 1 - Test Specimen Visualisation.

2.1 Steel Grade

S235 is chosen due to load capacity requirements.

2.2 General Layout

The general layout of the two test girders is given in Figure 2 with dimensions
per Table 1. Both girders are double-symmetric. Transverse stiffeners and endplates have the same width and thickness as the flanges.
2.2.1 Flanges
Designed to comply with the limits of class 2 and to be top and bottom braced
at supports and mid-span.
2.2.2 Web

Designed to have typical geometric slenderness of / > 180 and aspect ratio
of / = 2,5 to follow typical cable stayed bridge ratios.

(a)

(b)

Figure 2 - General layout: typical cross-section (a) and elevation (b).

Table 1 - Dimensions of the test sections.

section

web
flanges

[mm] [mm] [mm] [mm] [mm] [mm]

1S
2S

2250 1500
2250 1500

8
8

250
250

20
20

750
450

[mm]
80
80

stiffeners
1
2
[mm] [mm]
160
160

80
80

[mm]
5
5

2.2.1 Longitudinal Stiffeners

The cross-section dimensions of the longitudinal stiffeners are such that no torsional buckling can occur. In addition, the / ratios of the stiffener wall components should comply with the limits of class 3 sections in order avoid premature
buckling.
Thus, closed-section trapezoidal stiffeners were considered with relative bending
stiffness = 50 > 25 to fulfil strong stiffness requirements and to be higher than
the minimum stiffnesss which prevents pure compression and bending global buckling, although lower than the minimum stiffness which prevents pure shear global

buckling when 1 or 2 stiffeners are used 1

= 140 and 2
= 167, respectively.
Position in depth has been considered in the middle of the girder for 1 stiffener
section and with a distance to the flange of = 450 for 2 stiffners section.
2.2.2 Load Spreading Devices
In order to introduce the compressive force, a load spreader is needed. This requires 2xHEB500, strongly stiffened, either end of the girder to achieve 1000 mm
of spreading length.
Four Macalloy bars will then be anchored on the HEB500 flanges, two either side
of the web, forming a self-balanced system. The position of the prestressing bars
ties with the elastic neutral axis of the girders to avoid prestressing losses due to
girder bending.

3 Section Properties
3.1 Gross Section
Table 2 and 3 show the geometric and resistant properties for the gross sections.
Table 2 - Geometric properties of gross section.

section
Aw

Af

area
As

Af/A

[cm2] [cm2] [cm2] [cm2] [%]

Iw

If

inertia
Is

[cm4]

[cm4]

[cm4]

[cm4]

[%]

355

If/I

1S

120

100

13

233 43% 225000 562500

787855

71%

2S

120

100

26

246 41% 225000 562500 24010 811510

69%

Table 3 - Resistant properties of gross section.

section
Nw+s

axial
Nf.Rd

moment
Mf.Rd
Mpl.rRd

Mf/Mpl

shear
Vy.Rd

Npl.Rd

My

[kN]

[kN]

[kN]

[kNm]

[kNm]

[kNm]

[%]

[kN]

1S

3124

2350

5474

2469

1763

2820

63%

1628

2S

3428

2350

5778

2543

1763

3003

59%

1628

3.2 Effective Section

3.2.1 EN1993-1-5 Formulation
The effective section properties were calculated according to the EN1993-1-5 with
the dedicated softwares CTICM Z4 V1.0.1 and EBPlate V2.0.1.
Considering * the ratio between the most compressed fibre and the less compressed or tensile fibre stresses on the gross-section, then Table 4 and Figure 3
illustrates the effective properties for a given ratio and the interaction curve between Neff.Rd and Meff.N.Rd, respectively.
1

N/N

eff.rd

0.75

0.5

EN1993-1-5: 1S

0.25

EN1993-1-5: 2S
N/Neff.rd + M/Meff.rd = 1

0
0

0.25

0.5

0.75

M/M eff.rd
Figure 3 - Interaction curves using the EN1993-1-5 formulation.
It is worth noting that the interaction does not follow a linear variation due to
the changing of the effective area and elastic moduli with the stress field, and it is
more quadratic for section with 2 stiffeners.
*

=1 represents pure compression and =-1 pure bending.

Table 4 - Effective section properties.

section
1

0.75

0.5

0.25

-0.25

-0.5

-0.75

-1

Aeff [cm2]

177

180

183

188

196

210

213

216

219

Wel,sup [cm3]

9242

9268

9301

9335

9359

9360

9425

9500

9589

Wel,inf [cm3]

9242

9305

9408

9567

9871 10255 10247 10240 10236

eNz [cm]

0.0

0.1

0.4

0.9

1.9

3.3

3.0

2.7

2.4

N [kN]

4160

3701

3225

2761

2303

1851

1251

635

Meff.N.Rd [kNm]

261

519

772

1010

1254

1586

1919

2253

Class

205

212

219

223

228

231

241

242

243

Class

1S

Aeff [cm2]

2S

Wel,sup [cm3]

10120 10185 10272 10333 10443 10461 10478 10502 10518

Wel,inf [cm3]

10120 10193 10389 10421 10449 10464 10523 10565 10562

eNz [cm]

0.0

0.0

0.4

0.3

0.0

0.0

0.2

0.2

0.2

N [kN]

4825

4359

3852

3275

2679

2036

1416

711

288

584

886

1200

1511

1826

2147

2472

Meff.N.Rd [kNm]

3.2.1 Numeric Formulation

For completeness, FEM numeric interactions have been carried out (
Figure 4) to predict the same interaction curve.
From the figure, it can be depicted a similar variation but with more non-linear
reserve than the EN1993-1-5 curves.
1

N/Neff.rd

0.75

0.5

N/Neff + M/Meff = 1

0.25

FEM Test 1S
FEM Test 2S

0
0

0.25

0.5

0.75

M/Meff.rd
Figure 4 - Bending-Compression interaction curves using FEM formulation.

4 FEM Simulation
To predict the girders ultimate capacity a GMNIA analysis was carried out using
ABAQUS finite element software. The material properties considered an elasticplastic behaviour with yield stress of fy = 235 N/mm2, Youngs moduli of E = 210
kN/mm2 and a nominal yielding plateau slope of E/10000. S4R shell elements with
size of 25 mm were used and the analysis was undertaken using the RIKS method.

4.1 Equivalent Geometric Imperfection

Residual stresses and geometric imperfections were considered using the equivalent geometric imperfection concept as per the EN1993-1-5 Annex C. Thus, two
types of imperfections were considered: (i) global longitudinal stiffener imperfection
= min(400 ; 400) and (ii) local plate imperfection = min(200 ; 200).

(100% )
[mm]
3,7

3,7

2,3

1,5

5,1

4,8

(70% )
[mm]
=

0 [mm]

Figure 5 - Equivalent Geometric Imperfections.

In combining these imperfections, a leading imperfection was chosen and the

accompanying imperfections value reduced to 70%. Regarding the number of halfwaves considered on the local imperfection, a sensibility analysis was carried out

and the results shown that square half-waves gives the lowest ultimate resistance.
These imperfections are shown in Figure 5.

4.2 Choice of the compression load

In order to carry out the experimental tests it is necessary to define the levels of
compression for the GH and GL cases. Two possible candidates are: (i) GH - when
all the web is in compression; and (ii) GL - when the girder is subjected to some
level of compression but the web show tensile stresses. Accordingly, Table 5 shows
a set of compression forces deemed to satisfy these requirements.
Table 5 - Compression levels.

N/Neff.Rd

% Fpuk

[%]

[kN]

[-]

[%]

GL_1S

25

1251

-0.50

30%

GL_2S

25

1416

-0.50

34%

GH_1S

50

2303

55%

GH_2S

50

2679

64%

test

It is noted that according to the EN1993-1-5, the case where all the web is in
compression is considered as the turning point to change the M-V-N interaction
equation to clause 7.1(1) taking . = 0 which neglects the stress redistribution
between the web and the flange and therefore contradicts the reduced width method
principles.
The compression load is fully applied before the girder is vertically loaded and
kept constant during the test.

4.3 Force-Displacement Results

Following the numeric simulations, the force-displacements curves are shown in
Figure 6 (a) and (b), and the maximum load capacities summarised on Table 6.
A skim look at the results prompt the following comments:
1. Girder with 1S shows lower ultimate capacity and higher post-peak dropdown which is increased with the presence of the compression;
2. Both girders present a non-linear decrease of the load capacity with the
axial compression. In comparison, while the loss in capacity is 10% for the
GL case, for the GH situation this loss is around 30%. This drop-down is
even more accentuated in terms of ductility;

3. GH FEM simulations gave shear capacities below half the pure shear resistance where bending-shear interaction is not expected, as the preliminary results shows.

(a)

(b)
Figure 6 - Load-Displacements curves: (a) girder 2S and (b) girder 1S.

10

Table 6 - Ultimate load capacity.

test

FFEM

Loss

FFEM/
2Vy.Rd

MFEM

MFEM/

MFEM/

Meff.N.Rd EN1993-1-5

Meff.N.Rd FEM

[kN]

[kN]

[%]

[-]

[kNm]

[-]

[-]

2256

0.69

2160

0.96

0.90

GL_1S

0
1251

2030

10%

0.62

1943

1.13

0.97

GH_1S

1416

1517

33%

0.47

1452

1.22

1.04

G0_2S

2414

0.74

2354

0.95

0.86

GL_2S

0
2303

2188

9%

0.67

2133

1.10

0.93

GH_2S

2679

1653

32%

0.51

1612

1.18

0.98

G0_1S

5 Discussion
5.1 M, V, N Interaction
Based on the results obtained in the previous section, we can compare them with
the existing interaction formulations (Figure 7) and draft some remarks herein:
1. Interaction curve proposed by the EN1993-1-5 (red curves) seems unsuited
to describe the results obtained from the FEM analysis;
2. When the moment is divided by Meff.N.Rd.FEM, and despite out of scope of his
work, the interaction curve proposed by F. Sinur works well when compared with the results from the FEM tests from no compression to high
levels of compression. Since the capacities Meff.N.Rd.EN1993-1-5 are lower than
Meff.N.Rd.FEM (Table 6) then normalising the bending moment with the effec-

0.75

0.75
Ved/Vbw.rd

Ved/Vbw.rd

tive bending capacity given by the EN1993-1-5 will be on the safe side.

0.5
EN1993-1-5

0.5
EN1993-1-5

FEM Test 1S

0.25

FEM Test 2S

0.25

Meff.N0_1S FEM

Meff.N0_2S FEM

Meff.N25_1S FEM

Meff.N25_2S FEM

Meff.N50_1S FEM

Meff.N50_2S FEM

0
0

0.25

0.5

0.75

0.25

Med/Mpl.rd

0.5

0.75

Med/Mpl.rd

Figure 7 - Moment, shear, compression interaction.

11

N/Neff.rd
1.0

, ,

+ (1

, ,
2
) ( 1) = 1
, ,
,

, , =
(1

0.25
0.5

VN.rd/Vbw.rd

,
=
2

Nf.rd/Neff.rd

(1
)

1 +

(1 ,
)

> ,

Meff.N.rd/Meff.rd

1.0

1.0
M.f.N.rd/Meff.rd
V/Vbw.rd

0.5

M/Meff.rd

Mf.rd/Meff.rd

Figure 8 - M-V-N interaction to further investigate.

3. Due to laboratory constrains, regarding the capacity to provide higher

jacking forces and compression levels, the range of test specimens does not
cover the area where shear is higher than half the pure shear capacity with
full web in compression;
4. Figure 8 is deemed to be a reasonable candidate to describe the overall
behaviour. In fact, all the tested levels of compression are lower than
Nf.Rd/Neff.Rd where it is not expected any shear reduction due to the compression. The curves shown in Figure 7 are the horizontal planes plotted
in Figure 8.

5.2 Previous research

From 1971 until 2014, Hasan et al. (2015) showed an extensive state of the art
on steel and steel-concrete composite plate girder bridges. Whilst several works
refers to M-V interaction, there is no reference to the effect of axial compression.
Ziemian (2010) highlighted that the post-buckling strength of plate elements subjected to the combined action of shear and compression is limited. Since the analytical and experimental results on stiffened plates are relatively scarce, recourse is
usually made to data for unstiffened plates supplemented with whatever data are
available for the type of longitudinally stiffened construction considered to be most
important in practice, typically that shown in Figure 9. This example was reported

12

by Harris and Pifko (1969) and reference is made to the fact that this was done for
an integrally stiffened panel made of aluminium.

Figure 9 - Analytically predicted elastic and inelastic interaction curves for an integrally stiffened panel.

The predicted interaction curves are shown in this figure for both the elastic case,
which agrees very well with the parabolic relationship, and the inelastic buckling
case. Because of the limited nature of the data, no general relationship for the
inelastic buckling case can be derived. However, it should be noted that the circular
relationship lies above the analytical curve of the elastic buckling case.
Regarding directly the M-V-N interaction, Goczek and Supe (2014) derived a
formulation for class 1,2 and 3 sections. It regards to an extrapolation from the
EN1993-1-1 M-N interaction to the V-N interaction and combination of both (Figure 10).

Figure 10 - M-V-N interaction from Goczek and Supe (2014).

13

Despite developing a very interesting idea, it does not show any numeric or experimental results to justify it.
It is worth noting that the experimental works from Sinur and Beg (2013) were
the kick-off outline to the setup shown previously.

5.3 Codified Methods

5.3.1 BS 5400-3:2000
In the British codified method BS 5400-3 (2000) the stiffened web is checked on
a panel-by-panel basis. The design procedure consists of calculating the longitudinal
stresses using simple bending theory and gross areas. Shear forces are assumed to
be distributed uniformly through the cross section. Each panel is then checked for
yielding under combined compression, bending, and shear using the interaction formula,
+ 0.77

(
) + 3( ) 1

And for buckling using

( ) + ( ) + 3(
) 1



2

Where Kc, Kb and Kq (Figures 11,12,13) are coefficients for ultimate panel
strength and were derived from large-deflection elastoplastic computer analysis of
isolated plates, with results adjusted according to test data, and the interaction
formula was shown to give lower bound solutions by similar analyses.
In checking yielding, any proportion of the longitudinal stresses up to 60% maximum in a panel, can be assumed to be shed to the flanges while maintaining overall
equilibrium. In checking stability, up to 60% of these stresses can be shed from the
restrained inner panels, but none can be shed from outer panels, which are unrestrained.
This concept implicitly allows to stress redistribution between plates, the resultant stress distribution after such redistribution must be such that the whole of the
applied bending moment and axial force is transmitted and equilibrium is maintained.

14

Figure 11 - Buckling coefficient Kc.

The percentage reduction in stress in the web panels can vary from panel to panel
but is assumed to be uniform within any one panel. No redistribution is permitted
from panels containing holes larger than a specified size. Similarly, stresses that
cause yielding of the tension flange, but not buckling or yielding of the compression
flange, may be redistributed within certain restrictions as outlined in BS 5400: Part
3.

15

Figure 12 - Buckling coefficient Kq.

In summary, the interaction between moment, compression, and shear in girders

with longitudinal stiffeners is dealt with by relieving the web of some of the longitudinally destabilizing compressive stresses caused by bending and compression,
and distributing the load to the compression flange while maintaining overall equilibrium of the cross section.

16

Figure 13 - Buckling coefficient Kb.

This method could be understood as a reduced stress method allowing for limited
stress redistribution.
5.3.2 EN1993-1-5
The EN 1993-1-5 (2006) uses a different approach based on the effective width
method. While the web is not fully compressed, it considers the interaction between
bending, compression and shear as per a global action check given by
2
,,

+ (1
) (2
1) 1
,,
,,
,

On the other hand, when the axial force is so large that whole web is in compression then Mf,Rd should be taken as null and the interaction reads

+ (2 1) 1
,,
,
2

These curves are represented in Figure 7 and were found to be unsuitable to

describe the M-V interaction. Although, most recent research from Sinur (2011),
resulted in an improved M-V interaction equation defined by

17

+ (1
) (2
1) 1
,
,
,

In this equation , is used instead of , and the lower bound value for

the index is given by

=(

15
,
+ 0.2) + 1
,

Table 7 - Resistance decrease when = 1.

,
,

0.5
0.6
0.7
0.8
0.9
1.0

1.00
1.04
1.21
2.00
5.18
16.41

0.5

0.6

0.7

0.8

0.9

0.0%
0.0%
0.0%
0.0%
0.0%
0.0%

0.1%
0.5%
1.8%
3.3%
2.0%
0.0%

0.1%
0.6%
2.3%
5.2%
4.1%
0.0%

0.1%
0.6%
2.2%
5.5%
5.6%
0.0%

0.1%
0.4%
1.4%
3.8%
5.3%
0.0%

0.0%
0.0%
0.0%
0.0%
0.0%
0.0%

It is worth noting that this value has been calibrated without compression been
taken into account. Table 7 shows the resistance decrease if is taken conservatively as one. Results show that the maximum difference between resistances lies
below 6% and the average difference is 2.2% for VEd/Vbw.Rd between 0.6 and 0.9,
and Mf,Rd/Meff,Rd lower than 1.0.
Considering compression, a further development on this equation would be

,,

+ (1
) (2
1) = 1
,,
,,
,,

Where

,, =
(1

,, = ,

18

,,

(1
)

=
1 +

(1 ,
)

>

6 References
BS 5400-3. 2000. Steel, Concrete and Composite Bridges-Part 3: Code of Practice
for Design of Steel Bridges, no. December.
EN 1993-1-5. 2006. EN1993-1-5: Design of Steel Structures - Part 1-5: General Rules
- Plated Structural Elements. Eurocode 3. Vol. 5.
Goczek, Jerzy, and ukasz Supe. 2014. Resistance of Steel Cross-Sections
Subjected to Bending, Shear and Axial Forces. Engineering Structures 70.
Elsevier Ltd: 27177. doi:10.1016/j.engstruct.2014.02.016.
Harris, H. G., and A. B. Pifko. 1969. Elastic-Plastic Buckling of Stiffened
Rectangular Plates. In Proc. Symp. Appl. Finite Elem. Methods Div. Eng.,
edited by Vanderbilt University. Nashville.
Hasan, Q. A., W. H. Wan Badaruzzaman, Ahmed W. Al-Zand, and Azrul A.
Mutalib. 2015. The State of the Art of Steel and Steel-Composite Plate Girder
Bridges. Part I: Straight Plate Girders. Thin-Walled Structures. Elsevier, 1
32. doi:10.1016/j.tws.2015.01.014.
Sinur, Franc. 2011. Behaviour of Longitudinally Stiffened Plate Girders Subjected
to Bending-Shear Interaction. University of Ljubljana.
Sinur, Franc, and Darko Beg. 2013. Moment-Shear Interaction of Stiffened Plate
Girders - Numerical Study and Reliability Analysis. Journal of Constructional
Steel Research 88. Elsevier Ltd: 23143. doi:10.1016/j.jcsr.2013.05.016.
Ziemian, Ronald D. 2010. Guide To Stability Design Criteria for Metal Structures.
6th ed. John Wiley & Sons, Inc.

19