You are on page 1of 13
ICCBT2008 Effect of Web and Flange Thickne ss of Profiled Web Girder using LUSAS M.

ICCBT2008

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

M. H. Abu Hassan*, Universiti Tenaga Nasional, MALAYSIA

H.

M. A. Al-Mattarneh, Universiti Tenaga Nasional, MALAYSIA

A.

Ibrahim, Universiti Teknologi MARA, MALAYSIA

H.

Abdul Hamid, Universiti Teknologi MARA, MALAYSIA

B.

S. Mohammed, Universiti Tenaga Nasional, MALAYSIA

ABSTRACT

The behaviour of steel girders with profiled web under shear load has been simulated using the finite element method. Three different buckling modes, namely local, zonal and global were, observed and discussed. The typical failure mode of a girder with profiled web is that it is initially in the local buckling mode occurring either at the top, middle or bottom of one flat rib upon reaching a peak load. Beyond the peak load, the ripple propagates into adjacent folds transforming the local to zonal or the ripple extends across a number of ribs forming a global buckling mode in a diagonal direction of the tension field action beyond the peak load (post-buckling load) and gradually buckled due to crippling of the web and subsequently buckled till the flanges yielded vertically into the web. In the process of buckling, the load displacement relationship of the girder switched to a sudden and steep descending branch. The buckling can reduce the post-buckling shear capacity up to 60% of the ultimate shear capacity.

Keywords: Profiled Web, buckling, plate Girder, Finite Element

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

1. INTRODUCTION

For many structures all of the beams may be selected from among the standard range of rolled sections. Certain industrial buildings have girders called gantry girders that carry rails for large-capacity overhead cranes. Normal (gantry) girders are made up of built-up sections, called plate girders. Nowadays it is a common practice to fabricate such sections simply by welding together three plates to form the top and bottom flanges, and the web. From time to time, a new generation of optimised steel girders is developed. In general, innovated girder systems would require less material and result in a lighter structure when compared to a conventional girder system having webs reinforced with vertical/horizontal stiffeners.

Reviewing the previous research the use of profiled webs is a possible way of achieving adequate out-of-plane stiffness out of webs without using stiffeners. The use of profiled web girder also leads to a structural system of high strength-to-weight ratio. This finding agreed with Khalid et al. [1], Chan et al. [2] and Khalid [3] who had reported that a profiled web weighed 10.6% less than the equivalent conventionally stiffened flat web. Recent research by Abdul Hamid et al. [4-6] on intermittent rectangular profiled web girders showed that, the ribs are able to act as stiffeners, anchoring the tension field zone. The web buckled in typical shear mode and developed large strain of inclined tension. It was also noted that if the depth and the width of the ribs are increased further, the tension field action would develop in the ribs instead, causing them to behave as sub-panels.

2. BUCKLING BEHAVIOUR OF PROFILED WEB GIRDER

In the early 1960s, the first attempt to establish a method to predict the ultimate load of girder of civil engineering proportions was made by Basler [7]. He assumed that flanges in practical plate girders do not possess sufficient flexural rigidity to resist the diagonal tension field. The diagonal tension field does not develop near the web-flange juncture and the web collapses after development of yield zone. In 1970s Rockey et al. modified these theories to achieve a better correlation between theory and tested results [8, 9]. They assumed that the flanges were able to anchor the diagonal tension field. They also established that the collapse mode of plate girder involved the development of plastic hinges in tension and compression flanges from after development of yield zone and finally web panel fails in sway mechanism. The different of this two model as shown in Figure 1.

However, developments of tension field and collapse behaviour of profiled web girder are similar to conventional flat web girder but the mode of buckling are different. According to Elgally et al. [10], buckling modes are categorized as either local or global but Lou and

Edlund [11] were categorised as local, zonal or global buckling mode. Figure 2 illustrates the three different buckling modes described by Lou and Edlund [11] occuring in a corrugated web. They are:

a. Local buckling: shear buckling occurs in the plane part of the folds and is restricted to this region only

b. Global buckling: shear buckling involves several folds and may give rise to yield lines crossing these folds

c. Zonal buckling: an intermediate type of shear buckling (between local buckling and global buckling), which involves several folds but only occurs over a part of the girder depth

516

ICCBT 2008 - C - (45) – pp515-528

τ

τ τ τ τ τ d 135° 45° τ a V u Figure 1(a). Unbuckled Behaviour of

τ τ τ d 135° 45° τ a
τ
τ
τ
d
135°
45°
τ
a

V

u

Figure 1(a). Unbuckled Behaviour of Shear Web Panel V u V u Yield Zone
Figure 1(a). Unbuckled
Behaviour of Shear Web Panel
V
u
V
u
Yield Zone
V u
V
u

Figure 1(c). Collapse Behaviour by Basler

M. H. Abu Hassan et. al.

τ−τ cr
τ−τ
cr
σ t τ−τ cr τ−τ cr σ t θ i τ−τ cr Figure 1(b). Post-buckled
σ
t
τ−τ
cr
τ−τ
cr
σ
t
θ
i
τ−τ
cr
Figure 1(b). Post-buckled
Behaviour of Shear Web Panel
Plastic hinge W X θ i V u V y σ u t Y Z
Plastic hinge
W
X
θ
i
V
u
V
y
σ
u
t
Y
Z
φ

Figure 1(d). Collapse Behaviour by Rockey et al (Cardiff Model)

Figure 1: Failure Mechanism of Shear Web Panel

Model) Figure 1: Failure Mechanism of Shear Web Panel (a): Local Buckling (b): Zonal Buckling (c):

(a): Local Buckling

1: Failure Mechanism of Shear Web Panel (a): Local Buckling (b): Zonal Buckling (c): Global Buckling

(b): Zonal Buckling

(c): Global Buckling

Figure 2. Buckling Modes of Corrugated Web by Lou and Edlund [8]

ICCBT 2008 - C - (45) – pp515-528

517

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

3. FINITE ELEMENT MODELLING OF PROFILED WEB GIRDER

Finite element models were developed for the specimens tested and nonlinear analysis was performed using LUSAS a finite element software to simulate the combined geometric and materials non-linear response of the girders. Figures 3 show the dimensions of the typical girder tested and the profiled steel sheet used. In this paper, all the dimensions are kept constant except web and flanges thickness and the others dimensions are tabulating in Table 1.

Applied

Load, P

522 65 522 50 125
522
65
522
50
125

(a): Experimental setup

d = 550

22.5

83.5 45 123.2°
83.5
45
123.2°

3.5

45

(b): Dimensions of profile steel sheets of PEVA 45

Figure 3. Dimensions of Test Girder and Profiled Steel Sheet in mm

Table 1. Dimensions and Properties of Profiled Web Girder

Model Name

Web Thickness

(mm)

Flange

Thickness

Flange

Width

Yield Stress

(mm)

Flange

Web

S550t0.8-Fe

0.8

9.0

*S550-Fe

1.0

9.0

S550t1.2-Fe

1.2

9.0

S550t2.0-Fe

2.0

9.0

S550T3-Fe

1.0

3.0

S550T6-Fe

1.0

6.0

125

302

405

S550-Fe

1.0

9.0

S550T12-Fe

1.0

12.0

S550t2.0-Fe

2.0

9.0

S550t2.0T20-Fe

2.0

20.0

Note: *Same in the experimental

All models were created in a 3D surface element where the entire plate components such as flanges, web and stiffeners were modelled with quadrilateral thin shell elements (QSL8) of different density and distribution. The material response was assumed to be elastic-perfectly plastic and non-linear geometry of the girder used Total Lagrangian approach. An isotropic stress potential with von Mises yield condition was adopted for the material attribute. The load

was applied on to the line feature along the width of the top flange to ensure that the load was transferred through the bearing stiffeners and to avoid the top flange from locally buckled into the web due to the concentrated patch load, thus simulating the experimental setting-up as shown in Figure 3.

518

ICCBT 2008 - C - (45) – pp515-528

a . E x p e r i m e n t a l S

a. Experimental Setup

M. H. Abu Hassan et. al.

Global Axis

Y
Y
X
X

Z

b. Finite Element Model

Figure 3. Simulating Experimental Setup using Finite Element

Since initial imperfection was not measured, a half-sine wave geometric imperfection was assumed, and applied to both web panels with the same magnitude and direction with a maximum initial imperfection of 0.1% of web depth. However, the arc-length control using Crisfiled arc-length procedure used in advanced non-linear incremental parameters solution did not refer to the current stiffness. The sign of the current stiffness parameter was good at coping with bifurcation points, but would always fail when a snap-back situation was encountered. According to Lou and Edlund [11], the snap-back phenomena generally occurred on numbers subject to shear loading. Lou and Edlund also used Crisfield arc-length procedure in ABAQUS, commercial finite element software, where it was able to efficiently handle snap-through situations. Notably in the presence of strain-softening, the arc-length method may converge on alternative and unstable equilibrium paths. To ensure the girder did not buckle prematurely due to unstable of geometry deformation and/or lateral torsional buckling, the flanges were pinned in global X-direction as shown in Figure 3.

4.

RESULTS AND DISCUSSIONS

4.1

Validation of Finite Element Analysis against Experimental Results

Figure 4 shows comparison of load deflection curves of analytical and experimental results for different test specimens. The elastic buckling (first slope) part of each curve shows that the entire finite element results were so stiff compared to the experimental results. The effect of the first slope of the load deflection curve could be due to the initial setting of experimental the setup. Loading, the specimens were not fully rested on the supports. Another reason effecting the first slope of the load deflection curve was because initial imperfection of the flanges (warping) due to welding, was not included in the finite element analysis. Initial imperfection was modelled only using half sine wave for web panels which was not exactly like the experimental specimens. However, comparison of their ultimate shear loads was

ICCBT 2008 - C - (45) – pp515-528

519

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

satisfactory. Table 2 shows the comparison of ultimate shear loads using finite element analysis against experimental results. Beyond the peak load for each load deflection curve there existed a snap back situation. Non-linear finite element study done by Lou and Edlund [11] also indicated this snap back situation as shown in Figure 5.

140 120 120 100 100 2 2.5 80 60 40 S550-Fe S550-1 (Exp) 20 S550-2
140
120
120
100
100
2
2.5
80
60
40
S550-Fe
S550-1 (Exp)
20
S550-2 (Exp)
S550-3 (Exp)
0
0
2
4
6
8
10
12
14
Shear, V (kN)

Deflection (mm)

Figure 4. Load Deflection Curves for S550 Series

Table 2. Comparison of Ultimate Shear Loads of Finite Element against Experimental Results

     

V

U

(exp)

Experiment

V u (exp)

V u (FE)

V

U

(FE)

S550-1

130.00

 

0.94

S550-2

119.10

122.29

 

1.03

S550-3

124.50

0.98

520

ICCBT 2008 - C - (45) – pp515-528

M. H. Abu Hassan et. al.

M. H. Abu Hassan et. al. Figure 5. Load-deflection Curves for Corrugated Web Girder Investigated by

Figure 5. Load-deflection Curves for Corrugated Web Girder Investigated by R. Lou and Edlund [11] under Shear with Different Corrugation Depths

4.2 Influence of Web Thickness

Figure 6 show plots of load deflection respond obtained using the non-linear analysis. Each curve shows that the load reduced suddenly after reaching the peak which was about 50% of the ultimate capacity. Table 3 shows the results for different web thickness. All of the models buckled in zonal buckling mode as shown in Figure 7. Models with web thickness 1.0 mm and 1.2 mm had the same failure mode as shown in Figure 7(b) and (c). No peak and dale occurred in the load deflection response as could be seen from Figure 6. When the models with web thickness 0.8 mm, 1.2 mm and 2.0 mm were compared to the model with web thickness 1.0 mm the ultimate shear capacity reduced by 24% or increased by 21% and 106% respectively.

Shear, V

(kN) 260 t = 0.8 mm 240 t = 1.0 mm 220 200 t =
(kN)
260
t
= 0.8 mm
240
t
= 1.0 mm
220
200
t
= 1.2 mm
180
t
= 2.0 mm
160
140
120
100
80
60
40
20
0
0
1
2
3
4
5
6
7
8
9
10

Deflection (mm)

Figure 6. Load Deflection Curves for Different Web Thickness

ICCBT 2008 - C - (45) – pp515-528

521

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

Table 3. Results of Non-Linear Analysis for Different Web Thickness

Comparison of Ultimate Shear

Capacity with

t =1.0 mm

Ultimate

Post-

 

Web

Shear

Buckling

Model

Thickness

Capacity,

Capacity,

(mm)

V

u

V

b

V

b

V

u

V

 

(kN)

(kN)

V

u (t

=

u

1.0 mm)

S550t0.8-Fe

0.8

92.41

50.15

0.54

0.76

S550-Fe

1.0

122.29

58.90

0.48

-

S550t1.2-Fe

1.2

148.13

68.88

0.46

1.21

S550t2.0-Fe

2.0

252.24

140.36

0.56

2.06

S550t2.0-Fe 2.0 252.24 140.36 0.56 2.06 (a): Web Thickness 0.8 mm (c): Web Thickness 1.2 mm

(a): Web Thickness 0.8 mm

252.24 140.36 0.56 2.06 (a): Web Thickness 0.8 mm (c): Web Thickness 1.2 mm ( b

(c): Web Thickness 1.2 mm

2.06 (a): Web Thickness 0.8 mm (c): Web Thickness 1.2 mm ( b ) : W

(b): Web Depth 550 mm

mm ( b ) : W e b D e p t h 5 5 0

(d): Web Thickness 2.0 mm

Figure 7. Buckling Modes Obtained at the End of the Analysis for Model with Different Web Thickness

522

ICCBT 2008 - C - (45) – pp515-528

4.3 Influence of Flange Thickness

M. H. Abu Hassan et. al.

All the models showed the flanges were buckle into the web as shown in Figure 8. For the flanges thickness 3.0 mm, the top flange was buckled sharply into the web and it looked like patch loading behaviour. This is because the flanges were very thin to anchor the tensile force from the web. That also showed that the tensile force was developing in the small region flat part of corrugation fold.

Figure 9 and Figure 10 show the load deflection curves with different flange thickness for constant web thickness 1.0 mm and 2.0 mm respectively. In both figures the flanges did not have great influence in term of strength. Compared to the thinnest flanges (T =3.0 mm), the ultimate shear strength increased only about 4% for web thickness 1.0 mm. For web thickness 2.0 mm, the ultimate shear strength only increased about 2%. The results in Table 4 show that when the slenderness of the flange element changed from slender to plastic, the influence on ultimate shear strength was insignificant.

However, the use flanges thickness of T = 3.0 mm could lead to a more abrupt reduction in the post-buckling shear capacity. Figure 8. (a) shows the flange was buckled into the web such as concentrated patch loading. Table 4. show the reduction of post-buckling shear capacity up to 86% of ultimate shear capacity for model S550T3-Fe (where the flanges thickness is 3.0 mm).

model S 550T3-Fe (where the flanges thickness is 3.0 mm). (a): S550T3-Fe (b): S550T6-Fe ( c

(a): S550T3-Fe

(b): S550T6-Fe

(c): S550-Fe

S550T3-Fe (b): S550T6-Fe ( c ) : S 5 5 0 - F e (e): S550T12-Fe

(e): S550T12-Fe

S550T6-Fe ( c ) : S 5 5 0 - F e (e): S550T12-Fe (f): S550t2.0-Fe

(f): S550t2.0-Fe

) : S 5 5 0 - F e (e): S550T12-Fe (f): S550t2.0-Fe ( g )

(g): S550t2.0T20-Fe

Figure 8: Buckling Modes Obtained at the End of the Analysis for Model with Different Flange Thickness

ICCBT 2008 - C - (45) – pp515-528

523

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

Shear, V

(kN) 140 120 100 80 T = 3 mm T = 6 mm 60 T
(kN)
140
120
100
80
T
= 3 mm
T =
6 mm
60
T
= 9 mm
T = 12 mm
40
20
0
0
1
2
3
4
5
6
7
8
9
10

Deflection (mm)

Figure 9: Load Deflection Curves for Different Flange Thickness, T with Web Thickness 1.0 mm

Shear, V

(kN) 260 240 220 200 180 160 140 120 100 80 60 T = 20
(kN)
260
240
220
200
180
160
140
120
100
80
60
T
= 20 mm
40
T
= 9 mm
20
0
0
1
2
3
4
5
6
7
8
9
10

Deflection (mm)

Figure 10: Load Deflection Curves for Different Flange Thickness, T with Web Thickness 2.0 mm

524

ICCBT 2008 - C - (45) – pp515-528

M. H. Abu Hassan et. al.

Table 4: Results of Non-Linear Analysis for Different Flange Thickness

Comparison of Ultimate Shear Capacity

for

Comparison of Ultimate Shear

Capacity for

t =2.0 mm

Ultimate

Post-

Web

Model

Thickne

Name

ss

Shear

Buckling

Capacity,

Capacity,

V

b

V

u

 

(mm)

V

u

V

b

t =1.0 mm

 

(kN)

(kN)

 

V

u

V

 

V

u (T

=

3.0 mm)

V

u

 

u (T

=

9.0 mm)

S550T3-Fe

1.0

117.79

17.00

0.14

-

-

S550T6-Fe

1.0

123.07

74.97

0.61

1.04

 

-

S550-Fe

1.0

122.29

61.00

0.50

1.04

-

S550T12-Fe

1.0

122.79

54.25

0.44

1.04

-

S550t2.0-Fe

2.0

252.24

140.36

0.56

-

-

S550t2.0T20-Fe

2.0

256.04

-

-

-

1.02

4.3 Buckling Behaviour of Profiled Web Girder

All of the tested specimens did not buckle in a symmetrical manner. Only one side panel buckled and pulled the flanges due to tension field action globally in a web panel or zonally in a few web sub-panels. The finite element study also found the same buckling phenomena. Figure 11 shows the unsymmetrical deformed mesh of the profiled web specimens using finite element analysis. However, local buckling mode only occurred after the load reached peak and transformed to zonal or global buckling mode. This kind of deformation behaviour was clearly observed with finite element analysis, where the web started to buckle in one flat part of the fold or in a few folds and then developed large deformation crossing fold lines over a part of panel width. Then, it subsequently buckled till the flanges yielded vertically into the web. The load deflection behaviour changed to what was referred to as a sudden and steep descending branch after reaching peak. That confirmed the load sudden and steep descending branch due to local buckling of flat part of corrugation fold.

ICCBT 2008 - C - (45) – pp515-528

525

Effect of Web and Flange Thickness of Profiled Web Girder using LUSAS

Web and Flange Thickness of Profiled Web Girder using LUSAS a. Specimen S550-1 after testing c.

a. Specimen S550-1 after testing

Web Girder using LUSAS a. Specimen S550-1 after testing c. Specimen S550-3 after testing b. Specimen

c. Specimen S550-3 after testing

S550-1 after testing c. Specimen S550-3 after testing b. Specimen S550-2 after testing a. FE model

b. Specimen S550-2 after testing

S550-3 after testing b. Specimen S550-2 after testing a. FE model S550-FE at load terminated Figure

a. FE model S550-FE at load terminated

Figure 11. Experimental and FE buckling behaviour of profiled web girder

5.

CONCLUSIONS

From the results obtained, the following conclusions could be made of present investigation:

Buckling modes of profiled web girder were categorized in three different buckling modes i.e. local, zonal or global. Local buckling mode occurs at the first stage of buckling generally after the load reaching the peak. Zonal or global buckling mode occurred at failure load terminated (final failure). From observation, the buckling phenomenon started locally in flat part of web sub-panel (local buckling) and propagated to another flat part of web sub-panel which then transformed to zonal or global buckling mode. Local flange buckling occurred depending on the web buckling modes. This behaviour occurred because the contribution of stress field in web was small and restricted only in these corrugation folds.

526

ICCBT 2008 - C - (45) – pp515-528

M. H. Abu Hassan et. al.

Three buckling modes had been found in this investigation but after initially buckled, no matter what kind of buckling modes it had to abrupt reduction of the post-buckling shear capacity. The buckling could reduce the post-buckling shear capacity in average about 30% to 50% of the ultimate shear capacity.

Increasing the flange thickness did not influence the ultimate shear capacity but the use of thinner flanges would reduce the post-buckling capacity.

REFERENCES

[1].

Khalid, Y. A., Chan, C. L., Sahari, B. B. and Hamouda, A. M. S., “Bending behaviour of corrugated web beam”. Journal of Materials Processing Technology; Vol. 150, 2004, pp. 242 – 254.

[2].

Chan, C. L., Khalid, Y. A., Sahari, B. B. and Hamouda, A. M. S., “Finite element analysis of corrugated web beams under bending”. Journal of Construction Steel Research; Vol. 58, 2002, pp 1391 – 1406.

[3].

Khalid, Y. A., “Bending strength of corrugated web beams”. ASEAN Journal on Science Technology for Development; Vol. 20, 2003, Issue 2, pp 177 – 186.

[4].

Abdul Hamid, H., Ibrahim, A. and Abu Hassan, M.H, “Intermediately stiffened webbed welded plate girder”, Proceeding 7 th International Conference on Steel and Space Structures, Singapore, October 2002, pp. 267 – 274.

[5].

Abdul Hamid, H., Ibrahim, A. and Abu Hassan, M.H, “Plate girder under shear load”, Proceeding of the 5 th Asia-Pacific Structural Engineering Conference, Johore Bahru, Malaysia, August 2003, pp. 451 – 466.

[6].

Abdul Hamid, H., Ibrahim, A. and Abu Hassan, M.H, “Buckling of singly and doubly- webbed corrugated web girders under shear loading”, Technical Post Graduate Symposium, Universiti Malaya, Kuala Lumpur, Malaysia, December 2003, pp. 627 –

629.

[7].

Basler, K., “Strength of Plate Girder in Shear”, Journal of Structural Division, ASCE,

[8].

Vol. 87 No. 7, pp 151 – 180, 1961. Rockey, K. C., Evan, H. R., and Porter, D. M., “A design method for predicting the

[9].

collapse behaviour of plate girders”, Proc. Inst. Civil Engrs., Part 2, pp.85 – 112, 1978 Rockey, K. C., Valtinat, G., and Tang, K. H., “The design of transverse stiffeners on

[10].

webs loaded in shear – an ultimate load approach”, Proc. Inst. Civil Engrs., Part 2, pp.1069 – 1099, 1981 Elgaaly, M., Himilton, R. W. and Seshadri, A., “Shear strength of beam with corrugated webs”, Journal of Structural Engineering; Vol. 122, No. 4, 1996, pp. 390 – 398.

[11].

Lou, R. and Edlund, B., “Shear capacity of plate girders with trapezoidal corrugated webs”, Thin-Walled Structures; Vol. 26, No. 1, 1996, pp 19 – 44.

ICCBT 2008 - C - (45) – pp515-528

527