Vous êtes sur la page 1sur 8

Color profile: Disabled

Composite Default screen

160

NOTE / NOTE

Study of shear provisions for stiffened plate girders


Robert Loov and Narayana Parthasarathi

Abstract: Steel design in Canada generally follows the provisions of CSA-S16-01 “Limit states design of steel struc-
tures” (CSA 2001). The provisions in this standard governing the shear design of stiffened plate girders set limits to
the choice of web thickness, girder depth, and the spacing of intermediate stiffeners. This paper reviews the influence
of each of the equations that govern the shear design of stiffened plate girders. The study reveals that many of the
equations are unlikely to have any effect. Of the 13 equations that could restrict the design, only 3 are likely to have
any influence on a typical design. This reveals avenues for possible simplification of design procedures.

Key words: anchor panel, shear design, stiffened plate girder, tension field panel.
Résumé : La conception des charpentes en acier au Canada suit généralement les stipulations de la norme CSA-S16-01
« Règles de calcul aux états limites des charpentes en acier ». Les stipulations de cette norme gouvernant la conception
face au cisaillement des poutres à âmes pleines renforcées déterminent les limites du choix de l’épaisseur de l’âme, de
la profondeur de la poutre et de l’espacement des raidisseurs intermédiaires. Cet article examine l’influence de chaque
équation gouvernant la conception face au cisaillement des poutres à âmes pleines renforcées. L’étude démontre que
plusieurs équations n’auront probablement aucun effet. Des treize équations qui pourraient restreindre la conception,
uniquement trois influenceraient probablement la conception typique. Cela indique des chemins pour une simplification
possible des procédures de conception.
Mots clés : panneau d’ancrage, conception face au cisaillement (shear design), poutre à âme pleine renforcée, dalle
sous tension.
[Traduit par la Rédaction] Loov and Parthasarathi 167

General both flexure and shear and be the most economical with re-
spect to the total cost of material, fabrication, and construc-
The geometric parameters that determine the shear buck- tion. There are 13 equations in CSA-S16-01 (2001) that
ling strength of stiffened plate girder webs are the web slen- govern the shear design of plate girders. Shear design can be
derness, h/w (girder depth to web thickness ratio), and the a tedious and time-consuming exercise unless the designer
aspect ratio of the stiffened panel, a/h (stiffener spacing over has sufficient experience to know which of these equations
girder depth). will govern.
The stress limits specified by CSA-S16-01 (CSA 2001)
are elastic buckling, inelastic buckling, and yield. Tension
field action increases the shear stress that can be supported Parametric plots
by tension field panels in the elastic buckling zone and a A Mathcad® computer program was developed that takes
portion of the inelastic buckling zone. into consideration all the design equations. The program in-
The designer’s task is to find a combination of girder cludes commonly used web thicknesses and provisions for
depth, web thickness, and stiffener spacing that will support varying the stiffener spacing and the yield strength of flange
and web material.
Received 30 September 2002. Revision accepted Graphs for one of the case studies (case 1, discussed later
29 October 2003. Published on the NRC Research Press Web in this article) are presented in Figs. 1a and 1b for anchor
site at http://cjce.nrc.ca on 13 February 2004. and tension field panels, respectively. The lines defining the
minimum and maximum web depths are also shown. The
R. Loov. Department of Civil Engineering, The University of graphs show all the zones of influence, such as inelastic
Calgary, 2500 University Drive, NW, Calgary, AB T2N 1N4,
Canada.
buckling, elastic buckling, and “fabrication and handling”,
N. Parthasarathi.1 Bantrel, 1400 – 700 6th Avenue SW, as well as the zone corresponding to unstiffened girders. No
Calgary, AB T2P 0T8, Canada. tension field effects are considered for anchor panels and
within the narrow inelastic buckling zone adjacent to line 8
Written discussion of this note is welcomed and will be in Fig. 1b for tension field panels.
received by the Editor until 30 June 2004.
These graphs can be used for final design, as they show
1
Corresponding author (e-mail: partha@shaw.ca). the combinations of web thickness, web depth, and stiffener

Can. J. Civ. Eng. 31: 160–167 (2004) doi: 10.1139/L03-096 © 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:50 PM
Color profile: Disabled
Composite Default screen

Loov and Parthasarathi 161

Fig. 1. Typical graphs of h versus a (case 1): (a) anchor panel, and (b) tension field panel.

2
spacing that meet all the design conditions. Provided a suffi-  F 
cient web thickness is used along with the corresponding [3] a = 67500h  yf 
web depth, any stiffener spacing less than the limits defined  83000 
by eqs. [1] and [2] can be used.
where Fyf is the specified minimum yield stress for flange.
Equations for the zone boundaries Equation [4] is the upper boundary of the elastic buckling
The equations that demarcate the different zones on an h zone. This equation combines the elastic buckling require-
versus a graph have been derived from the limits in CSA- ment of Cl. 13.4.1.1(d) and the limit of Cl. 14.3.1 to prevent
S16-01 (CSA 2001). All clause references refer to this stan- vertical buckling of the compression flange.
dard and all symbols are defined in the list of symbols.
5. 34
Some of the equations are not explicit functions of h and a [4a] a=h for kv > 9.34
so require iterative solution. kv − 4
Equation [1] is simply the maximum stiffener spacing
specified in clause (Cl.) 14.5.2. This line forms a boundary and
between stiffened and unstiffened girders.
4
[1] a = 3h [4b] a=h for kv ≤ 9.34
kv − 5. 34
Equation [2] is the boundary between elastic buckling of
unstiffened webs mentioned in Cl. 13.4.1.1(d) and the zone where the shear buckling coefficient for stiffened girder kv
governed by fabrication and handling specified in Cl. 14.5.2. and the minimum web thickness to avoid vertical buckling
These code limits were combined to formulate an expression of compression flange wvb are given by
for a as a function of h.
Vf h
1/ 3 kv =
   
2
180000 φ w vb
3
1 V
[2] a = 67500   f
 
 h  180000 φ kvu  
  and

where the shear buckling coefficient for unstiffened girder Fyf h


w vb =
(kvu) is 5.34 and Vf is the shear force in a member or compo- 83000
nent under factored load.
Equation [3] is the upper boundary of the fabrication and Equation [5] is eq. [4] revised for the tension field panel.
handling zone governed by Cl. 14.5.2. The depth of the web The elastic buckling equation must include tension field ef-
is restricted by Cl. 14.3.1 to prevent vertical buckling of the fects from Cl. 13.4.1.1(d) and the limit to prevent vertical
compression flange. buckling of the compression flange from Cl. 14.3.1.

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:51 PM
Color profile: Disabled
Composite Default screen

162 Can. J. Civ. Eng. Vol. 31, 2004

1/ 2 2
   2
83000Vf φ  05 . h 
[5] h=   [9] kv = Fy3    
 φ Fyf [ Fcre + (0.5 Fy − 0.866 Fcre ) / 1 + (a/h) ]   f  866  
2 290V 0.
 
where Fy is the specified minimum yield for web and the Equation [10] is the boundary between inelastic buckling
elastic critical plate-buckling stress in shear Fcre is given by (Cl. 13.4.1.1(c)) and elastic buckling (Cl. 13.4.1.1(d)). Again
2
solve for a as in eq. [4] with kv defined by
 F 
Fcre = 180000kv  yf  [10] kv =
Vf h
 83000  180000 φw ie3
where where the web thickness on the dividing line between elastic
5.34 and inelastic stress zones wie is given by
kv = 4 + for a/h < 1
(a/h) 2 180000Vf
w ie =
2902 φ Fyh
and
Equation [11] is the boundary between inelastic buckling
4
kv = 5.34 + for a/h ≥ 1 from Cl. 13.4.1.1(c) and elastic buckling including the ten-
(a/h) 2 sion field effect from Cl. 13.4.1.1(d). It is given by
1/ 2
Equation [6] combines the equations for inelastic buckling  Vf (h/w) 
from Cl. 13.4.1.1(b) and vertical buckling of compression [11] h=  
 φ[ Fcre + (0.5 Fy − 0.866 Fcre ) / 1 + (a/h) ] 
flange from Cl. 14.3.1. The solution is the same as eq. [4] 2

except that kv is changed to


2 where Fcre and h/w using kv from eq. [5] are given by
1  Vf 
[6] kv =   180000 kv
Fy  290φw vb 
2 Fcre =
(h/w) 2
Equation [7] combines inelastic buckling including the and
tension field effect, Cl. 13.4.1.1(c), and vertical buckling of
compression flange, Cl. 14.3.1. 180000 kv
h/w =
1/ 2 290 Fy
 83000Vf 
[7] h=   Equation [12] is the boundary between elastic buckling
 φ Fyf [ Fcri + (0.5 Fy − 0.866 Fcri ) / 1 + (a/h) ] 
2
given in Cl. 13.4.1.1(d) and fabrication and handling from
Cl. 14.5.2 using kv from eq. [5]. It is given by
where the inelastic critical plate-buckling stress in shear Fcri
3/4
using kv from eq. [5] is given by Vf  67500 
[12] h=  
 F  180000φ kv  a/h 
Fcri = 290 Fykv  yf 
 83000  Finally, eq. [13] is the boundary between elastic buckling
including the tension field effect (Cl. 13.4.1.1(d)) and fabri-
Equation [8] must satisfy the limit imposed by yield cation and handling (Cl. 14.5.2). It is given by
stress, Cl. 13.4.1.1(a), and inelastic buckling, Cl. 13.4.1.1(b).
1/ 2
Solve for a as in eq. [4] with kv based on  Vf (h/w) 
2 [13] h=  
1  V   φ [ Fcre + (0.5 Fy − 0.866 Fcre ) / 1 + (a/h) ] 
2
[8] kv=  f 
2
Fy  290 φw y 
where Fcre and h/w using kv from eq. [5] are
where the minimum web thickness based on yield stress 180000kv
Fcre =
limit wy is given by (h/w) 2
Vf
wy = and
0.66 φ Fyh
h 67500
=
Equation [9] forms the boundary between the inelastic w (a/h)
buckling zone as specified in Cl. 13.4.1.1(b) and inelastic
buckling with tension field limited by Cl. 13.4.1.1(c). Solve Equations [1]–[4], [6], [8], and [10] can be used directly
for a as in eq. [4] with kv given by to determine h as a function of a. Equations [5], [7], and

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:51 PM
Color profile: Disabled
Composite Default screen

Loov and Parthasarathi 163

Table 1. Case studies: design information.


Reference w (mm) h (mm) Fy (MPa) Mf (MN·m) Panel Vf (MN) a (mm) a/h Design basis
Case 1
Kulak and Grondin 10.00 1400 350 6.20 Anchor 1.082 1600 1.14 CSA (2001)
(2002, p. 241)
Tension 0.970 3000 2.14
Case 2
Galambos et al. 6.35 2032 248 4.99 Anchor 0.973 635 0.31 AISC (1995)
(1980, p. 222)
Tension 0.935 1092 0.54
Case 3
McCormac 9.53 1575 248 7.28 Anchor 1.413 1067 0.68 AISC (1995)
(1992, p. 483)
Tension 1.368 1105 0.70

[11]–[13] require kv to be determined based on a chosen a/h Fig. 2. Case 1: load and shear force diagram. All dimensions are
ratio before the depth can be calculated. in millimetres.

Lines for specific web thicknesses


The lines representing the possible combinations of web
depth and stiffener spacing for each given web thickness
provide insight into the most feasible web thickness to be
considered. For a given depth, any panel length that is less
than the limit plotted for the chosen web thickness will be
satisfactory.

General comments on h versus a diagrams


The economical solution will generally be found in the
central region of these graphs (Fig.1). Unless there is a ma-
jor economic advantage to a shallow depth, shallow mem-
bers adjacent to eq. [8] are likely to be unduly expensive, as
they require large flanges to support the moment. The inelas- Discussion of case studies
tic buckling zone is small, and hence both the zone without
Practical cases from previously engineered structures from
tension field effect and the zone including the tension field
several sources and example problems from various struc-
effect are generally of little interest. Web depths near the top
tural steel design textbooks were investigated (Parthasarathi
of the graphs require thick webs and will not be economical.
2000; Parthasarathi and Loov 1999). Three representative
samples of these case studies from various textbooks are
Anchor panels presented in Table 1. Each case study was analyzed to exam-
The appropriate choice of depth, web thickness, and panel ine the influence of each equation. Table 1 lists the geomet-
width is almost certain to fall within the elastic buckling ric parameters and material strength assumed in the design,
zone because it covers most of the feasible region. The the factored shear force as well as the factored moment and
choice for the girder depth and web thickness of anchor pan- the source of the case study.
els must be coordinated with the adjacent tension field pan-
els, as the girder depth and web thickness are the same. Case 1 (Kulak and Grondin 2002)
The girder details, loading, and factored shear force dia-
Tension field panels gram are shown in Fig. 2. The design equations have been
Several of the equations that limit the web dimensions for plotted in the graphs in Figs. 1a and 1b. The black diamond
tension field panels are the same as those for anchor panels. indicates the chosen web depth and stiffener spacing for a
The equations for minimum depth and maximum depth are 10 mm thick web. The maximum panel lengths are listed in
the same. Equations [1] and [2] again provide the extreme Table 2. As seen from Table 2 and Fig. 1a, the anchor panel
upper limits for stiffener spacing. The maximum stiffener length in this example does not meet the elastic buckling re-
spacing will generally be governed by the elastic buckling quirements of CSA-S16-01 (CSA 2001). The reason is that
equation or by the fabrication and handling equation. The the resistance factor φ was not considered when determining
extreme maximum stiffener spacing for tension field panels kv in this example. When the φ factor is correctly included,
occurs along line 13 (Fig. 1b), the boundary between the the anchor panel length needs to be reduced from 1600 to
elastic buckling and fabrication and handling regions. 1400 mm to satisfy the elastic buckling requirements.

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:51 PM
Color profile: Disabled
Composite Default screen

164 Can. J. Civ. Eng. Vol. 31, 2004

Table 2. Case studies: review of stiffener spacing.


Stiffener spacing a required based on
Stiffener spacing a Elastic Fabrication and
Panel type provided (mm) buckling (mm) handling (mm)
Case 1
Anchor 1600 1400 4200
Tension field 3000 6820 4200

Case 2
Anchor 635 710 1340
Tension field 1092 2730 1340
Midspan 1219 1340

Case 3
Anchor 1067 1060 3890
Tension field 1105 1970 3890
Midspan 5487 3890
Note: The numbers in bold are those that govern.

As illustrated in Fig. 1b and Table 2, the length of the ten- Fig. 3. Case 2: load and shear force diagram. All dimensions are
sion field panel for the given web thickness and depth is lim- in millimetres.
ited by eq. [1], from the fabrication and handling equations.
The stiffener spacing of 3000 mm is very conservative, since
a spacing of 4200 mm limited by the maximum a/h ratio of
3 would satisfy all design requirements. However, for the
chosen web size no reduction in the number of stiffeners is
possible.

Case 2 (Galambos et al. 1980)


The girder details, loading, and factored shear force dia-
gram are shown in Fig. 3. The design equations have been
plotted in Figs. 4a and 4b and the maximum panel lengths
are shown in Table 2. The panel lengths for a web thickness
of 6.35 mm have been chosen correctly in this example. The
web depth is very close to the vertical buckling limit. The
maximum anchor panel length is limited by elastic buckling,
while the length of the tension field panel is limited by fabri- tions in web dimensions affected the flange area required to
cation and handling. The design meets all requirements. As support the factored moment. For thin webs, the flange area
can be seen from Fig. 4b, the length of tension field panels was increased to compensate for the reduced flexural
could be increased and their number reduced if a shallower strength due to web buckling. The variations in web dimen-
web were used. sions also resulted in different stiffener areas and different
numbers of stiffeners.
Case 3 (McCormac 1992) Unit prices in Canadian dollars obtained from the steel
The girder details, loading, and factored shear force dia- fabrication industry have been used as representative mate-
gram are shown in Fig. 5. The design equations have been rial and fabrication costs. The following approximate unit
plotted in Figs. 6a and 6b and the maximum panel lengths prices were used for the cost comparisons:
for the chosen 9.53 mm web thickness are shown in Table 2. Cost of materials and setup and welding of flanges and
The length of the anchor panel is at its limit. The length of web = $1.50/kg
the tension field panel is governed by elastic buckling and is Cost of materials for stiffeners = $2.00/kg
much shorter than required. Three tension field panels would Cost of setup for stiffeners = $50/pair
be sufficient. Cost of welding each pair of stiffeners = $100/m height
If the fabrication and handling limitation is to be fol- An addition of 3% has been made to the theoretical flange
lowed, an additional stiffener should be added to subdivide area and 10% to the theoretical stiffener area to accommo-
the portion of the beam between load points. date rounding of dimensions.
All stiffeners for a given beam were assumed to be identi-
Cost analysis cal; the differences between bearing stiffeners and interme-
diate stiffeners were ignored. Stiffeners were assumed under
Cost studies were carried out for stiffened plate girders each concentrated load, but the end stiffeners were not in-
with different web thicknesses and web depths. These varia- cluded.

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:52 PM
Color profile: Disabled
Composite Default screen

Loov and Parthasarathi 165

Fig. 4. Case 2: h versus a: (a) anchor panel, and (b) tension field panel.

Fig. 5. Case 3: load and shear force diagram. All dimensions are Case 1
in millimetres. The cost study considering web thicknesses of 8, 10, and
12 mm indicated that the most economical girder depth was
1.7 m for a 12 mm thickness. At their optimum depths the
8 mm web needed a total of 7 stiffeners, the 10 mm web
needed 5 stiffeners, while the 12 mm web needed only 3. A
graphical representation of the cost versus girder depth for
different web thicknesses is shown in Fig. 7. The estimated
cost of a 1.7 m deep girder with a 12 mm web is CAN$7647
as compared with the estimated cost of the original 1.4 m
deep girder with a 10 mm web, which is CAN$7977. In this
case, the estimated cost saving using the optimum dimen-
sions is about 4%.

Case 2
The original design was in Inch–Pound units, but this re-
view of alternatives has been based on metric sizes. The op-
timum occurs for a web depth of 1.7 m and a web thickness
Cost comparisons were made for each of the three cases.
of 12 mm. This solution is estimated to be 15% cheaper be-
It should be kept in mind that all the examples were devel-
cause only 4 pairs of stiffeners would be required compared
oped by their designers to show the procedure for designing
with 15 pairs needed for the original web dimensions. This
stiffened girders, and therefore none of these examples were
is an example where the original design did not have a depth
necessarily designed to be optimum. Nevertheless, the results
and panel length falling near the line separating the elastic
indicate that these experienced designers often produced de-
buckling and fabrication and handling zones, so the potential
signs close to the optimum determined by this study.
for cost savings was large.
To assess the cost sensitivity, the computations were re-
peated with the costs of stiffeners 50% higher and 33%
lower than the initial estimates while the cost of the web and Case 3
flange material remained the same. The results show that The optimum depth was found to be deeper than that
such major changes in the stiffener cost had only a marginal chosen by the designer. A depth of 2.0 m with a web
effect on the optimum depth. With increased stiffener cost thickness of 12 mm is estimated to be about 8% cheaper
the optimum depth was the same or up to 5% smaller, while than the given design. As indicated in Fig. 6b, the larger
with reduced cost the optimum was the same or up to 5% depth allows the stiffener spacing to be increased substan-
deeper. tially.

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:52 PM
Color profile: Disabled
Composite Default screen

166 Can. J. Civ. Eng. Vol. 31, 2004

Fig. 6. Case 3: h versus a: (a) anchor panel, and (b) tension field panel.

Fig. 7. Case 1: cost comparison. All costs are in Canadian dollars. For all the cases studied, the cost of girders with the depth
obtained using eq. [14] was virtually identical to the cost us-
ing the optimum depth obtained by considering a range of
depths as plotted in Fig. 7.

Summary and conclusions


(a) It has been found that of the 13 equations that could
restrict the design only 3 are likely to have any influence on
a typical design. The anchor panel is almost certain to be
governed by elastic buckling as limited by Cl. 13.4.1.1(d),
while the tension field panel for an optimum design is likely
to be close to the lower edge of the fabrication and handling
region governed by Cl. 14.5.2 but may be limited by a/h < 3
as defined by eq. [1].
Optimum depth (b) The parametric plots show the influence of each shear
As an incidental offshoot to this study, it was observed design equation, so a visual confirmation that the chosen de-
that the optimum depth for the chosen examples was close to sign meets all restrictions is immediately apparent. There is
eq. [13], the boundary between the elastic buckling region therefore no need for any further design or check calcula-
and the fabrication and handling region for the tension field tions.
panels. The length of the tension field panels is maximized (c) The girders investigated in this study had a constant
along this line. The resulting reduction in the number of web thickness and flange size for the full length of each
stiffeners reduces the girder cost. member. Where applicable, flange areas were increased to
Equation [13] is almost horizontal, so the variation in counteract the strength reduction resulting from flexural web
depth along this line is very small. This leads to a simple buckling. Flexural web buckling did not occur when the op-
method for estimating the optimum depth. If it is assumed timum web thickness was used.
that the optimum a/h ratio will be about 2, and φ = 0.9, then (d) Equation [14] provides a useful estimate of the opti-
eq. [13] can be solved to determine the optimum depth h. mum depth of beams such as those investigated in this study
because the number of stiffeners is then minimized.
Vf
[14] h ≈ 955
93 + Fy
References
where h is in millimetres, Fy is in megapascal, and Vf is in AISC. 1995. Manual of steel construction LRFD. Vol. I. 2nd ed.
kilonewton. American Institute of Steel Construction, Chicago, Ill.

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:53 PM
Color profile: Disabled
Composite Default screen

Loov and Parthasarathi 167

CSA. 2001. Limit states design of steel structures. Standard CSA- Fcreelastic critical plate-buckling stress in shear
S16-01, Canadian Standards Association, Toronto, Ont. Fcriinelastic critical plate-buckling stress in shear
Galambos, T.V., Lin, F.J., and Johnston, B.G. 1980. Basic steel de- Fyspecified minimum yield stress — web
sign. 2nd ed. Prentice-Hall, Inc., Englewood Cliffs, N.J. Fyf specified minimum yield stress — flange
Kulak, G.L., and Grondin, G.Y. 2002. Limit states design in struc- hclear depth of web between flanges
tural steel. Canadian Institute of Steel Construction, Willowdale, kvshear buckling coefficient for stiffened girder
Ont. kvushear buckling coefficient for unstiffened girder
McCormac, J.C. 1992. Structural steel design LRFD method. Mf bending moment under factored load
Harper & Row Publishers, N.Y. Vfshear force in a member or component under factored
Parthasarathi, N. 2000. A parametric study of shear provisions for load
stiffened plate girders. M.Eng. thesis, Department of Civil Engi-
w web thickness
neering, The University of Calgary, Calgary, Alta.
wie web thickness on dividing line between elastic and in-
Parthasarathi, N., and Loov, R.E. 1999. A parametric study of
elastic stress zones
shear provisions for stiffened plate girders. Proceedings of the
27th Canadian Society for Civil Engineering Annual Confer- wvb minimum web thickness to avoid vertical buckling of
ence, Structures, Regina, Sask, 2–5 June 1999. Edited by C.L. compression flange
Borbely. Canadian Society for Civil Engineering, Montréal, wy minimum web thickness based on yield stress limit
Que. Vol. I, pp. 115–124. φ resistance factor

List of symbols
a centre-to-centre distance between transverse web stiff-
eners

© 2004 NRC Canada

I:\cjce\cjce3101\L03-096.vp
February 5, 2004 3:27:53 PM

Vous aimerez peut-être aussi