Analysis of Flange Transverse Bending of Corrugated Web I-Girders Under In-Plane Loads
Analysis of Flange Transverse Bending of Corrugated Web I-Girders Under In-Plane Loads
Analysis of Flange Transverse Bending of Corrugated Web I-Girders Under In-Plane Loads
Abstract: This paper presents theoretical, experimental, and finite-element analysis results for the linear elastic behavior of corrugated
web steel I-girders under in-plane loads. A typical corrugated web steel I-girder consists of two steel flanges welded to a corrugated steel
Downloaded from ascelibrary.org by New York University on 05/18/15. Copyright ASCE. For personal use only; all rights reserved.
web. Previous research has shown that a corrugated web I-girder under primary moment and shear cannot be analyzed using conventional
beam theory alone, and a flange transverse bending analysis is required. A theoretical method, the fictitious load method, is presented
herein as an analytical tool for quantifying flange transverse bending in corrugated web I-girders. To validate this method, four-point
bending experimental results for a large-scale corrugated web I-girder are presented. The measured flange transverse displacements and
flange stresses were in good agreement with the theoretical results especially in regions of constant shear. To gain additional insight, finite-
element analysis results for the test girder are presented, and compared to both the experimental and theoretical results.
DOI: 10.1061/共ASCE兲0733-9445共2007兲133:3共347兲
CE Database subject headings: Experimentation; Finite element method; Flanges; Plate girders; Shear; Transverse loads; Webs.
stress resultant, Vt, associated with flange transverse bending is As de / dz = tan , where ⫽slope of the web profile as a func-
given by 共Abbas et al. 2006兲 tion of z, and dVY / dz = −pY , where pY ⫽in-plane load applied to
the girder, Eq. 共7兲 becomes
2VY e
Vt = − + A1 共2兲
h 2
pt = 关VY tan − epY 兴 共8兲
where A1⫽constant of integration that depends on the flange h
boundary conditions. It was shown by Abbas et al. 共2006兲 that the In the absence of distributed loads on the girder 共i.e., in con-
flange transverse moment, M t, is related to Vt by stant shear regions where pY = 0兲, Eq. 共8兲 reduces to
dM t 2
Vt = 共3兲 pt = 关VY tan 兴 共9兲
dz h
Experimental Results
Eq. 共9兲 indicates that the fictitious transverse load in a constant
shear region is directly related to . Hence, for corrugation pro- Test Girder, Setup, and Instrumentation
files with piecewise linear folds 共e.g., trapezoidal, triangular, rect-
To validate the flange transverse bending behavior predicted by
angular, etc.兲 pt is either zero for longitudinal folds 共as tan is
the fictitious load method, a laboratory experiment was conducted
zero兲 or uniform over the inclined folds 共as tan is constant兲.
on a simply supported corrugated web I-girder with a trapezoidal
This is illustrated by Case I in Fig. 4共a兲 which shows fictitious
web profile. The specimen belonged to a set of six that were
load patterns for a trapezoidal web profile in a constant shear
tested to investigate the fatigue strength of corrugated web
region. Note that as the corrugation slope approaches 90° 共i.e.,
I-girders 共Sause et al. 2006兲. The nominal geometry of the test
a rectangular web profile兲, as demonstrated by Case I in Fig. 4共b兲,
girder is shown in Fig. 5. The test girder is 7.4 m 共24.3 ft兲 in
the tangent of approaches infinity and a fictitious concentrated
length with an effective span of 7 m 共23.0 ft兲 that includes seven
load, Pt, is used, which can be expressed mathematically in terms
corrugation wavelengths, each 1 m 共3.28 ft兲 long. The web is
of the fictitious distributed load, pt, as follows:
1.2 m 共3.94 ft兲 deep and has a measured thickness of 6.25 mm
2VY hr 共0.25 in. 兲. The top and bottom flanges have a measured width of
Pt = lim ptdz = ± 共10兲 227 mm 共8.90 in. 兲 and a measured thickness of 20.4 mm
dz→0 h
共0.8 in. 兲. A pair of stiffeners was provided at both the reaction
where hr⫽corrugation depth 关Fig. 1共a兲兴. The positive sign in and load points to minimize local effects due to concentrated
Eq. 共10兲 is used when the sign of web eccentricity changes from loads. A full-scale trapezoidal corrugated web profile, shown in
negative to positive, and the negative sign is used when the sign Fig. 5共b兲, with a bend radius of 120 mm 共4.72 in. 兲 between the
of web eccentricity changes from positive to negative 共Fig. 4兲. In longitudinal and the inclined folds, was used. The geometry of the
a similar fashion, the pattern of fictitious loads in the presence of web profile was developed based on shear strength considerations
uniform loads, pY , can be determined from Eq. 共8兲, as shown by and fatigue and fracture characteristics, especially in the bend
Case II in Fig. 4. Although the patterns are slightly more compli- region 共Abbas 2003兲.
cated than for Case I, it is seen that the fictitious loads are either The test specimen was fabricated by High Steel Structures,
共1兲 concentrated, 共2兲 uniform, or 共3兲 linearly varying, depending Inc., a bridge fabricator in Pennsylvania, using ASTM A709 steel
on the corrugation profile. Finally, in the special case when VY is material 共ASTM 2001兲. The web and flanges were grade HPS
discontinuous, a fictitious concentrated load is required at the 485W, whereas the stiffeners were grade 345W. The nominal
point of discontinuity. It can be seen from the second term in yield stress for the HPS 485W steel material is 485 MPa 共70 ksi兲.
Eq. 共7兲 that this fictitious load is directly related to the product of To corrugate the web, a custom-made die was used that required
e and the change in VY at the point of discontinuity. a four-stoke process with manual incrementing 共Sause 2003兲. The
Once the fictitious loads are determined, the flange transverse web-to-flange fillet welds were made using a semiautomatic gas
bending problem can be solved using conventional structural metal arc welding 共GMAW兲 process, due to its low heat input and
analysis. Although manual calculation is possible, it is often more potential for automation. A detailed description of fabrication can
convenient to use commercial structural analysis software, espe- be found in Sause 共2003兲.
cially as the number of corrugations, n, increases. The flange The girder was loaded statically in four-point bending as
transverse bending problem can also be solved numerically shown in Fig. 5. The central region between the two loads was
共Abbas 2003兲. thus subjected to constant bending moment 共zero shear兲, whereas
It is worth noting that the method of analysis described herein the end regions were subjected to constant shear 共linearly vary-
for corrugated web I-girders under moment and shear is concep- ing moment兲. This loading arrangement, combined with the geo-
fold with the inclined fold were placed on the top surface of the
bottom 共tension兲 flange only and were intended for the subsequent
fatigue test. The strain gauge arrangement for the top flange was
similar to that of the bottom flange except that strain gauges were
not placed along the centerline 共Abbas 2003兲.
Six LVDTs were used to measure vertical displacements of the
bottom flange at three locations along the span, and seven LVDTs
were used to measure transverse displacements at five locations
along the bottom flange and two locations along the top flange.
These locations are indicated in Fig. 5. The six vertical LVDTs
were placed in pairs approximately 125 mm 共5 in. 兲 apart 共trans-
versely兲 at each of the three locations. The transverse LVDTs
were fixtured at least 1.2 m 共4 ft兲 away from the bottom flange to
reduce the effect of the vertical displacements on the horizontal
Fig. 5. Test specimen geometry, loading conditions, and displacement data acquired from these LVDTs.
instrumentation The static load was applied slowly under manual control. Six
static load cycles were conducted to verify the measurements
from the instrumentation and to ensure repeatability. Each jack
applied a maximum load of approximately 489 kN 共110 kips兲.
metric characteristics of the test girder, provided a good opportu- Results from the final cycle, load cycle 6, are presented in the
nity to observe the behavior of corrugated web I-girders under following section.
bending moment and shear. The girder was mounted on two stiff-
ened steel pedestals. The end supports allowed for the free exten-
Experimental Results and Comparison with Theory
sion of the bottom flange at one end, thus simulating the intended
simply supported boundary conditions. The loads were applied Fig. 6 shows a plot of the applied load, P, versus the midspan
using two 489 kN 共110 kips兲 jacks. The jacks were supported by vertical deflection, ⌬max. Fig. 6 shows that the behavior is linear
overhead braced steel frames, and the load points were laterally elastic. Using beam theory, the initial stiffness, k0 共in the expres-
braced using hollow steel tubes that were attached to the columns sion P = k0⌬max兲, including both flexural and shear deformations,
of the overhead frames. is given by
Instrumentation was selected to measure the flange transverse
bending behavior. A total of 85 uniaxial bondable electrical resis- k bk s
k0 = 共11兲
tance strain gauges and 13 linear variable displacement transduc- kb + ks
ers 共LVDTs兲 were used. Two load cells 共one at each jack location兲 where kb 共the bending stiffness兲 and ks 共the shear stiffness兲 are
were also utilized. The test data were acquired and processed given by
using a data acquisition system with a 16-bit analog/digital board
and a personal computer. The strain gauges were located on either 24EIX
kb =
the bottom or the top flange and were oriented in the longitudinal a共3L2 − 4a2兲
direction. Seven characteristic sections along the span 共four
within the shear regions and three within the constant moment
region兲 were instrumented with strain gauges that were, whenever
possible, provided in pairs to enable the flange middle surface
ks =
a
冉
GAw  + cos ␣
+1
冊 共12兲
strains to be estimated. The middle surface strains were used for where E and G⫽elastic and shear moduli, respectively; a and
direct comparison with theoretical results, which are based on the L⫽shear span and overall span, respectively; Aw⫽web area; and
assumption of thin-walled cross-sectional components. All strain IX⫽moment of inertia of the two flanges about the X axis. The
gauges were either placed along the flange centerline or at a dis- factor in parentheses involving ␣ 共the corrugation angle兲 and 
tance of approximately 13 mm 共0.5 in. 兲 from the flange edge. 共the ratio of the longitudinal fold width to the inclined fold width兲
Fig. 5 shows the strain gauge arrangement for the bottom flange. in the expression for ks takes into account the reduced stiffness in
Note that strain gauges near the intersection of the longitudinal shear for a corrugated web as compared to a flat web. This factor
stresses of the test girder were observed in regions where the strength of steel beams with corrugated webs.” J. Struct. Eng.,
behavior is dominated by primary bending. To provide additional 123共6兲, 772–782.
insight into this behavior, an FE analysis of the test girder was Grubb, M. A. 共1984兲. “Horizontally curved I-girder bridge analysis:
conducted and the results were presented. The FE analysis results V-load method.” Transportation Research Record. 982, Transporta-
tion Research Board, Washington, D.C., 26–36.
suggest that the differences between the experimental and theo-
Hibbitt, Karlsson, and Sorenson, Inc. 共HKS兲. 共2002兲. ABAQUS version
retical flange stresses can be attributed to the contribution of the
6.3, Pawtucket, R.I.
web to flexural resistance. Further research to quantify this effect
Johnson, R. P., and Cafolla, J. 共1997兲. “Local flange buckling in plate
is needed. girders with corrugated webs.” Proc. Inst. Civ. Eng., Struct. Build.,
122共2兲, 148–156.
Lindner, J. 共1990兲. “Lateral torsional buckling of beams with trapezoi-
Acknowledgments dally corrugated webs.” Proc., Int. Colloquium of Stability of Steel
Structures, Budapest, Hungary, 79–86.
This research was conducted at the Advanced Technology for Lindner, J. 共1992兲. “Zur bemessung von trapezstegträgern.” Stahlbau,
Large Structural Systems 共ATLSS兲 Center at Lehigh University. 61共10兲, 311–318 共in German兲.
The writers are grateful for the financial support of the ATLSS Lindner, J., and Aschinger, R. 共1990兲. “Zur torsionstegsteifigkeit von
Center, the Federal Highway Administration 共FHWA兲, the Penn- trapezsteg trägern.” Stahlbau, 59共4兲, 113–120 共in German兲.
sylvania Department of Transportation 共PennDOT兲, and the Penn- Machimdamrong, C., Watanabe, E., and Utsunomiya, T. 共2004兲. “Analy-
sylvania Infrastructure Technology Alliance 共funded by a grant sis of corrugated steel web girders by an efficient beam bending
from the Pennsylvania Department of Community and Economic theory.” Struct. Eng./Earthquake Eng., 21共2兲, 131–142.
Poellot, W. N. 共1987兲. “Computer-aided design of horizontally curved
Development兲. The opinions, findings, and conclusions expressed
girders by the V-load method.” Eng. J., 24共1兲, 42–50.
in this paper are of the writers, and do not necessarily reflect the
Protte, W. 共1993兲. “Zur gurtbeulung eines trägers mit profiliertem steg-
opinions of those acknowledged here. blech.” Stahlbau, 62共11兲, 327–332 共in German兲.
Sause, R. 共2003兲. “Corrugated web girder fabrication: Work area 2, Penn-
sylvania innovative high performance steel bridge demonstration
References project.” ATLSS Rep. No. 03-19, Lehigh Univ., Bethlehem, Pa.
Sause, R., Abbas, H. H., Driver, R. G., Anami, K., and Fisher, J. W.
Abbas, H. H. 共2003兲. “Analysis and design of corrugated web I-girders 共2006兲. “Fatigue life of girders with trapezoidal corrugated webs.”
for bridges using high performance steel.” Ph.D. dissertation, Lehigh J. Struct. Eng., 132共7兲, 1070–1078.
Univ., Bethlehem, Pa. Vlasov, V. Z. 共1961兲. Thin-walled elastic beams, 2nd Ed., Israel Program
Abbas, H. H., Sause, R., and Driver, R. G. 共2006兲. “Behavior of corru- for Scientific Translation, Jerusalem.