Lehigh University

Lehigh Preserve
Fritz Laboratory Reports Civil and Environmental Engineering

1961

Strength of plate girders in shear, Proc. ASCE, 87,

(ST7), (October 1961), Reprint No. 186 (61-13)
K. Basler

Follow this and additional works at: http://preserve.lehigh.edu/engr-civil-environmental-fritz-lab-

reports

Recommended Citation
Basler, K., "Strength of plate girders in shear, Proc. ASCE, 87, (ST7), (October 1961), Reprint No. 186 (61-13)" (1961). Fritz
Laboratory Reports. Paper 70.
http://preserve.lehigh.edu/engr-civil-environmental-fritz-lab-reports/70

This Technical Report is brought to you for free and open access by the Civil and Environmental Engineering at Lehigh Preserve. It has been accepted
for inclusion in Fritz Laboratory Reports by an authorized administrator of Lehigh Preserve. For more information, please contact
preserve@lehigh.edu.
F\)L~ Library
l

WELDED PLATE I~DE

I
 STRENGTE; .OF PLATE GIRDERS IN SHEA~

Submitted to the Pla~e Girder Project Committee

for approva~ as an ASCE Publication

by

Konrad Basler

F~RITZ ENGINEERING
LABORATORY LIBRARY

Lehigh Univers-i ty
Fritz Engineering Laboratory
Report No. 251~20
December 1960
1':"'-
11

TABLE OF CONT~TS
Page
Foreword iii

Synopsis iii

1'0 Introduction 1
2 0 The Ultimate Shear Force 5
30 Intermediate Stiffeners 16
40 Detailing of Gird:er Ends 24
50 The Influence of Strain Hardening 28
60 Allowable Shear stresses 31
70 Appendix: Review of the Shear strength Theo~y 34
Acknowledgments 40
Nomenclature 41
Table and F~gures 43
References 56
 ili

FOREWORD

This paper is the 'second in a series of three reports

on the study of the strength of plate gird~rs carried out
at Lehigh Universityo The research project was guided by
the Welded Plate Girder Project Connnittee whose members
~re listed in the Foreword to the first papero

§XN0PSIS

A study of the sh~ar strength of plate girders is pre~

sentedo In utilizing the post-buckling strength o~fered by

the transverse stiffening of girders, new design rules are

proposedo. The new approach 1s checked with ultimate load
tests carried out at Fritz Ehgineering Laboratoryo
251-20 -1

It' INTRODUCTION

lIn current civil engineering practice the shearing

stresse's in webs of plate girders are'·:analyzed according
to the classical beam theory established by N~vler and
sto Venant o Acco~ding to this theory the shear force is
re~1sted by a state of shearing stresses as pictured in

Figo lao The principal stresses at the neutral axis ~re

of the: same magnitude as the shear stress and act' at 45°

wi th the longi tudinal a.xis (Figo Ib) 0' Such a shear

oarrying action may be called "beam actiontio To satisfy

the condi tion of small deforma tiona p on which this ,"~.,

beam theory is based, tran"sverae stif.feners must be spaced

close enough so that instability due to shear is excludedo

Ever since plate girders came into use J it has been

·r~cognize.d th~t beam action alone is not the only way that
shear can be carriedo References 1 to 5·reflect·the ex~

tensive disoussion of the probl~m of web stiffenlng"oarried

on just before the turn of the last centuryo Intuition led
to the opinion that the action of a plate girder: was-similar
to that o~ a Pratt truss. (1) Turneaure (3) concluded from
a girder test "that there 1s as much' reason to suppose that
the web stresses rollow down the web and up the stiffeners,
as in a Howe Truss~' as to suppose that they follow the lines

of a Pratt Truss ll • However, model stu~ies (2,5) and a gir~er

test' (4) clearly indicated the importance of the web as
tension and the stiffeners ~s compression elementso The
goal was~ at that time~ to assess the nature of the stress
flo1rl in ~he web rat!?-sr' than to estimate the carrying

capacityo Thus~ more qua11~ative than quantitative results

could emergso But at the beginning of th~s century it led
to a rather l:tberal American 'design of girder webs,Sl with web

depths greater than 170 times the web thickness 0 Meanwhile,

a web buckling theory was developed to determine safe limits
for the desigrl of plate girders e, A possible truss action

was advanced merely to jllS'tify a somewhat lower factor o£

safety against web buckling than that required against
other stability cases such as column failurso Later tests
conducted in this country (6 1 7) were mainiy concerned with
the establishment of a slenderness limit for unstirfened
websc

In 1916,9 the Norwegian HeRo Rode wrote an outstanding

dissertation (8) in which one chapter deals with webs of
plate girders 0 It appears that he may have been the first
to mathematically formulate the effect or a tension field
or truss action which sets in after the web loses ita .
~igidity due tq buckllngQ He proposed to evaluate its in~

fltlence by considering a tension diagonal of a width equal

to 80 times the web thicknesso

 with the development of aeronautical science the shear~

carrying capacity of membrane~like structures got new atten-

tiona The ~aramount requirement of aircraft design (to
minimize the weight of the structure) led to ex~remely thin
webs 0 Since such structures were built of aluminum alloys~

the modulus of elasticity and henoe the web buckling stress

were correspondingly lower than f~r steel girderso By
.,

neglecting beam action completely in such structures,~ and

considering the web as a membrane resistant only to tension 9

Wagner formulated the "Theory of Pure Diagonal Tension" (9)
While pure diagonal tension is one limiting case of the state
of stress in a thin web~ pur~ shear is the other~ occurring
only in sto,cky webs 0 As seen from Ref 0 10}J extensive e·xp.eri-
mental studies were undertaken by the National Advisory
rCommi t-tee for Aeronautics to cover the trans! tion range ~
where uincomplete diagonal tension" o~curaso

There are several reasons why the civil engineering

profession has not applied ~he highly specialized practice j

of aeronautical engineering to the analysis of their struc~

r

tures 0 One reason is the reluctance to c o'nslder the flanges

as transversely ,loaded by the tension fi~ld forces and aoting
as beams supported by the intermediate stiffeners 0 The
flanges,,'- utilized ,primarily in carrying,' "the gird~rt s bending

mo~ent, would have to be specially designed to serve this

secondary purpose 0, With a continuous skin, such as arol.md
 ~4

a. fusela.ge or a wing of an aircraft 9 the- b,Qundar.y ~ondi tiona

are much more favorable for membrane action than in a welded

plate girder where no angle sections between web and flange
plate are used~ which would provide a certain d~gree of
rigidityo

'The purpose of the sUbsequent study is to derive a

simple but gE?neral formula :for the· ultimate ,shear strength
of steel girders with flanges not resistant to membrane
tension but with webs so slender that a certain tension
ac,tion might" develope> utilizing, s,ome idealizations, a
~~ctional dependance of th~, shear strength upon the princi~

pal param~~ers will be establishedo

 ~ ,TH~ ULTIMATE e~~ FORCE

Assuming that the ultimate shear force, Vu » of a trans~

versely stiffened plate girder could be e~pressed in a

formula, it would certainly have to be a function of the
following dimensional variables~ the stiffener spacing a~

the girder depth b~ the web thickness t» and the material

..
properties cry and E, where cry ia the yield stress 'and E
th~ modulus of elasticity. The shear force for which un-
re,str'ieted shear yielding occurs shall be termed n,p;lastic
shear force", Vp , analogous to the term "plastic moment",
Mp , used in plastic analysis 0 Since Vp has the, dimension
of a foree, it must be possible to express the ultimate shear
force Vu in the form Vu = Vpf(a,b,t,cry,E), where the function
f is nondimensionalo Therefore, whenever cry or E should
occur~ they can be brought in the form €y = cry/Eo The re-
maining variables a, band t, with length as their.
,
dimen~

sions, can only occur in ratios; where a = a/b.and ~ = bit

are sufficient to expr,ess all possible relations between
these three variables 0 'Th~s,_'p a formula for the ultima'te

shear load must be of the-form

(1)

From Mises t yield condit,ion for plane stress, which

1s illustrated; in Figo·· ,1.d, it is seen· that the shear yield
stress ~y equals cry/fro The full plastic shear force is
reached wh~n yielding occurs throughout the web depth b~

hence
(2) \

The shear contribution due to a possible tension field

is considered nexto Assuming that a field of 'uniform tension

stresses crt flows through a web l s cro,ss "s\ection bt (Figo 2&) p

the. 'resulting shear force V depends on the inclination ~ of

the tension stresses o From Figo 2b it is seen that the
maximum value is obtained when ~ is 4SO and thus

(3)

Whether a tension field, which is a membrane stress

field p can develop depends on the boundaries of the plateo
With regard to membrane stresses, a panel in a girder web
'.'

has two very"different pairs of boundariesjl those along the

£langes and those along the transverse sti~£~nerBo A
rlange of a oonventionally built welded plate girder has so
little bendingr1gid~~y in the plane of the web that, it
cannot effectively resist v~rt1cal stresses at its junction
wi th the web 0 Such :flanges ~ therefore ~ do, not ,,,serve 'a.s
anchors for a tension stre;Bs fielqQ The situation 1s
different at the panel boundaries along the transverse
stiff'eners (Figo 20) 0 There 'the tension strips can trans"
mit the stresseso Thus~ only a part of the web contains a
pronoun~ed tension field which gives rise to a shear force
~ Vcr ,.= 0tosotosin<p where s~ th~ field widthj) also depends

on <po

When a thin~web plate girder panel is subjected' to

shear 9 it will reach a stage where the compressive stress
02 indicated in Figo la ceases -to increase because the web
deflectso For the stress in the tension diagonal direotion,
no suc.h evading of dut,y"'existso Upon increasing the shear

£orce, yielding initiates along the tension'diagonalo A

further increas'e of the app+i'ed shear force causes a wider
portion of tne web t9 yield. Since the increase in field
width is gained by virtue of a decrease in the inclination
of the tension stress with respect to the girder ax!sjl' an
9ptimum value of the tension field contribution AVa to ~he

shear force Va is reaehedG It is reasonable to postulate

that at ultimate shear load the inclination of the tension
field is the one which furnishes the grea.te'st total shear
component AVa of this tension fieldo With the notation as
defined in Figp 20 this inclination is obtained from the
condition
d d (cryosotosin<p) =.: 0
d<p (6V 0) == dq>

[dS{~) sinq> + s· COS<p ] :::: 0

or CJyt d<p ~
 c>8

it ,reduces to ~

b tan 2 l<p + 2a. tanq>

Q f:'>I b :. 0

which gives~

tan<p

cos<p
)

Th~ strip corresponding ~o the optimum assumes a slope

between 450 and 0 0 ~ when a takes on values from 0-, ,to 00 0

It is also seen from the equations that 'the tension stx;ip

inclination is less than the inclination of the panel diagonal p
and the strip width a little wider than half the girder d~ptho

In Figo 3 the derived strip geometry is superimposed upon a

photograph of a thin~web girder panel subjected to shear
(girder G7, Hefo 11)~ The dark, yielded bands in the
diagonallybu.okled zones> alternating with at-rips of unyielded

or only slightly yielded metal p are due to the combined effect

of bending a~d membrane strasseso The formation of the
buckles 'produces plate bending stresses which are orthogonal
to the diagonal tension (membrane) stres~es and which are
maximtml at the surface of the web plate~, Yielding is pro=
nounced along the concave surfaces where compression bending
st'resses are superim,posed upon the membra.ne tension st,resses j)
 ~9

and reta~ded along the convex surfaces where the bending

produces tension stresses~ (Refo Fig$ lc)~

The photograph also illustrates the nature of the

anchorage provided in the neighboring panels, where the
horizontal tension component is transferred to the rlange
by shearo It indicates that the stiffeners must sustain
axial forceso The magnitude of this shear and stlfrener
~ .
force can be derived using the idealized tension rield, the
inclination of which is fixed by Eqso 40
A succession of equa.;L·. web panels all subjected to the.
same shear force is assumed as shown in Fig~ 40 cutting
along the sections A, B, and C~ a free body is obtained as
sketched in the same figure Q ., At the face A in the web an
unknown resultant is acting. It can be decomposed into a
normal component Fw and a snear force component which, be~

cause:of the symmetry of the chosen cut, must be Vcr/2o The

force acting in the flanges is denqted as mfo At section B
the stress pattern in the web is the same as in aectionA,
therefore the same components will occur as explained beforeo
T~e flange force will change in the amount of AFfo At
seetion C the stiffener force Fa. and tension field st~esses

0tare aoting~ These tension stresses are under~the 1n-

c~inatio~~whose trigonometric values are given in Eqso 40
Formulating moments around point 0 and considering equili~

brium in the horizontal and vertical directions furnishes

251~20 -10

three equations out of which the three unknowns AFf~ Vcrs and
Fa can be computedg

b
-;-AFf =

0ttb ( a
2

In plate girders with slender webs neither a pure beam

action <~) ,n'oI? a. pure tansion field action (a) occurs alone 9

but rather' the sum ot botho Therefore the ultimate shea.r

load Vu is

In.o~der to compute these two shear contributions~ two

more ,assumptions are requiredo The first one~ is :the postu~

late that the superposition o~ the stresses ~esulting from

both carr¥1ng actions is limited by the state o~ stress which
fulfills the yield condition~ as shown in the sketch on the
lefthand a ide o'f Fig o 1e () The s'econd is the assumption that p

up to 'CCI4 ~~hear is carried in a beam type lnannerjl but that,

.from then on,? Vrr; remairls constant (an assumption r'eviewed in
the Appe~dix)~ Any postbuckling benerit must be contributed
by tension field actiono

Then»
(8)
 -=11

IAccord1ng to the first ass1llllption

j
j the tension fie,ld
stress crt can·be expressed explicitlyo It is defined as
the s"t~esa which can be added to the state of shear stress

a.t the point of: bifurcation ofequil1bri;um (where "'xy equals

~er) ~uch that unrestr~cted y~elding occurs in th~.tensio~

field o For a short derivation attention must be given to

the subscripts used (Fi,gc>: Ie) 0 The .fixed coordinates .are

x and Yo The cartesi~n coordinates ~ and·v are generated

out of x and y by counterclockwise rotation in the magnitude
<po When, q> equals 45° thes,e are ca.lled axis 1 and 2 0 By

means of Mohr's circle?, a'hown in Figo lb j) it is seen that the

state of shear stresses ('t'xy:= 't
Ol
.,) compined with the diagonal
tension str'ess crt under the inclination cp 'can be G'xpressed asg

au = ~crs1n2~ + crt
CJ v =~ 't cr sin2<p (9)

't'uv = 't'ercos2'~

Introduoing this set or stresses in Mises t yield condition

(10)

the following result for the tension field stress crt is

obtainedg

= (ll)
 ~12

With this exp~ession the elements of the ultimate load

computation are eompleteo According to Eqs¢-5~ 7~ and 8
the ultimate shear load 1s.g

{J crt
vu == VP [ -
'tor
n;y
+
2 -cry 1"
., 1+a 2 i
] (12)

where (J·t/cry is given by Eqo 11.

Two simplifications which will ease the numerical com-

putation considerably are possibleo The first one is an
approximation of the yield condition which~ for the case or
plane stress j) is pictured in Fig. ld,o It can be seen that
states of stress anywhere between pure shear and pure tension
only lead to points on the ellipse between A and B 0 ,;Hence

the straight line cr1 := cry + {IT = 1) 0'2 passing throlJgh A and
B is alfair approximation of ~he yield conditiono For the
limiting case of ~ equal to 45~ 0u and av in-~qso 9 become
principal stresses~ 01 = ~cr + 0t$ and 02 =~ ~cr~ respeo~

tivelyo If they are intorduced in the"app~o~imated form of

the yield condition» the following simple relation resultsg

1 ~ (13)

The second simplifica.tion is tha.t the va.l~e for crt/cry

be computed from this relation ,when ~ is not equal to 45°9

This leads to a smaller tension fi~ld stress 'than Eqo 11

would give~ and the underestimation increases the more ~

decreases from 45°, that is, as a becomes larger 0 But, for
panels with large a ratios larg~r shear displacements are
required in order to develop a tension fieldo This second
approximation should not p therefore~ be considered merely
as a simplification made on account at the economy but al~o

as an allow~nce for compatibility conditionSa

In order to check the influence or these two assumptions?

the ultimate shear force, Eqo 12, was first eom1?uted'us~ng

at from Eqo 11 and termed Vu(ll), then using Eq~ 13 and·

termed Vu,(13) 0 The deviation would be defined~~s

5% = Y'u(!.2l - Vu(ll~~x 100; it is plotted in Figo 5 against

Vu(ll)
the variable "Ccrlr:y for various values of 0.0 Combining, the
effects or both simplifications yields results which dit~er

la'ss than 10% from the shear load computation fulfilling

Mises t yield condition exaetlyo

Thus~ using Eqso 12 and 13 the ultimate shear force can

be computed ,~as follows g

where~ Vu = the ultimate sheaI' rOl1ce

Vp , = the plastic shear force - ~ybt
rc y ~
the yield shear stress =- cry/V
 - 2.,. . 2-

1: == the cr'itical shear stress - k(a) 12' 'I-v

1t ,E (t) ,
(----tz2;~)~ b'
cr Q

When not otherwise ~tated» k(ct) in this last

expression is taken according to Ref~ 12 as
k = 5034 + 4~~ fora ~ 1 and k = 4.00 + ~.
, a a2
for a ~ 10 Whenever the thus computed value
~cr exceeds the proportional limit» it has been
considered only as an ideal value ~cri to be re~

duced according to Eqso 738 and 739 at Refa 12 in

connection with Table 2-7 of that reference o

The reduction concept employed here does not allow a

critical shear stress gFeater than ~y a~d thus does not
account for any stra1n~:p.a.rdening effect. It wl11,'p therefore !I
be too conservative ror low web slenderness'ratios. An
estimate of this effect» based on experiments~ is made in
Secb 5c>· '

Using Poisson's ratio \I := 003jl't cr/'t y equals 1057 k(a)/e y [32 o

SUbstituting this in Eqo 14 furnishes a function f on the right=
hand side which depends on ajl pjl and 8 y o~ly.. This function
r(a.'",{3J' ey)" is the one anticipated in\ Eqo 1 9 and it is seen
that the Itequirement concerning 'the paramete:r,~ is fulfilled o

Th:e ul timate shear forc~,' according to Eqo 14 is plotted

against the web slenderness parameter ~ in F1g~ 60 With the
nondimensional representation choaen p this chart gives the
predicted ultimate force of a transversely stiffened thin~web

pla.te girder ~s compared to t4e plastic shear force of the

same girder (> The influence of transverse s t'iffenin~ is
readily seen from the graph 0 When stiffeners ,are not present;.9
the value of a = ~applieso If the spacing were infinitely
close» the girder could always develop its rull:strength g
,--

represented by the plastic shear loado The e~f1ciency of

transverse stiffeners depends very much on the slenderness
ratio 0 The strength of a very slender web with a bit rat'io
around 300 can be increased several times using~atiffenerBo

This does not mean, that a very elose stir~ener spacing~ say
a == Oo5~ should be choseno It would be better "P9 _ reduce the
number of stiffen~rs and place add! tional material" in the
webo This would not only result in a lower slenderness ratio
and hence permit a. little higher allowable stress for any
given value of a but.9 because of the added web area? the 'com~

puted shear stress for a given shear force would also diminish 0

For the m,os't~· efficient girder desi-gn 1 t is necessary to

specify the st1frener size required to achieve the derived
tension fie~d actiono This will be discussed in the next
sectiono
 3 Q INTERMEDI4.TE STIFFENER.§.

In contrast ~to loading stiffeners ,intermediate

C" ' '

stiffeners are transverse elements through which no ex~

"tarnal forces a~e introduced into the girder 0 Their

function is two-fold~ to preserve the -shape of the girde~Vs

cross section and to insure postbuckling strengtho Dis-

regarding details at the rlanges (which are thoroughly dis~

cussed in Refo 13)$ the first function ,will require a certain

minimum stiffnesso The second function demands a minimum
strength, that is~ a certain minimum cross sectional area o

By dividing the girder's carrying capacity into two

parts'p simple beam a~t1on up to 1: :::t +t er and tension .field
action up to yielding in the web, it is possible to deter~

mine stiffness and strengtp requirements for the transve~se

intermediate stiffeners separatelyo Because a ~tate of shear

stress corresponding to simple beam action (Figo 1a) oauses
no axial load in the stiffener whatsoever~ the stiffener is
only required to be rigid en~ugh to foree~ at its. location,
a nodal line in the lateral deflection mode· of the webQ The
st~ffness requ~~~~en~~ of the current design sp~~1~1eations

are based on such considerat~ons and ensure the required

rigidityo In a tension field, however J the stiffener must
take the vert;1cal component of the diagonal stresses out of
the web at dne end· and transfer them to its othe~ endo
 ~17

Therefore» this section will be devoted to the second re~

quirement of an intermediate stiffener p the ability to

sustain compressiono

The stiffene,r force F s is derived in Eqo .6e> When the

maximum shear ~oree is reaehed~ the value for crt can be
taken from Eq<> 130 The st··~ffener force to be expected at

ultimate load is expressed by Eqo 15a belowo Depending on

whether a is greater or smaller than l~ and whether ~cr is
beyond the proportional limit or in the elastic range p
different analytical expressions for ~cr must be takeno An
expression for the zone where a ? 1 and ~cri = ~cr is given
in Eqo 15bo It is seen that in 'nondimensional form the re~

sult is again only a function or the three independent

variables a p B9 and 8yo

{15a)

By restricting" the investigation to a.., particular steel j)

Eq'o 15a can be plotted against the two coordinates a a.nd ~o

This is done in Fig" 7 for e y := O"OOl19:where the same k~

'values and r,eductions in the inelastic range are used as

presented under Eqo 140 There exists a maximum stiffener
force which,? sur.prisingly ~ is somewhere ir! the regiO!l where
most girders. are builto Its location can be found b:9~ I)ar"Jtial
differentiation of Eqo 5b with respect to the two variables
a and ~o This will yield a = 1.18 and ~ ~.l87; with a
maximum value of the stiffener force Fa ~ 5.0 x 10=4 Oyb 2 o

For unspecified ey~ Fs(max) occurs at a = 1018 and

2
8y {3 ~ 380 60

The physical explanation for this maximum is as follows~

Comparing all possible plate gir-ders wi th 50 inches web depth'j)

the one with a = 1 18 and 0 ~ = 187' would require a st·i1~1."bener

strong enough to carry a 41 kip axial load in order to develop

the maximum possible tension fieldo For all the other girders
of the same depth the stiff~ener force corresponding to a full
tension field could only be smaller~ becauseg
~ by increasing the stirrener spacing

the tension £ield action becomes less and less

effective and finally diminishes comp~etely ror
a = 00 Z

~ by decreasing ,the sti~fener spacing

the stirfener density increases and therefore the
sha.re assigned to a single sti.ffener reducesi
~ by ihcreasing the web slenderness

the web area decreases and with it the ultimate

shear ·force~ which is an upper bound for the
stiffener forcez and
 ~19

~ by decreasing the web slenderness

the web thickness increases and, although the web

area becomes larger~ more and more of the shear
:force will be carried in beam action. leaving s.

lessened capacity for tension field action.

With this finding» a simple expression :for specifying the

. minimum required are~ of intermediate stiffeners can be
derived as outlined nexto

When the tension. field has formed~ par-t of the web is

already at the stage of unrestricted yielding and no addi-
tional stresses can be assigned to it~ Therefore» the
stiffener force F s can only be resisted by the actual area
of t.he stiffeners,$ As 0 (As is the sum of the ·'areas of both

stiffenerso) In ~his case the required area is simply Fs!cru $

where au is the ultimate axial stress in the stiffenerstlo If
local buckling is avo1ded~ 0u is equal to the pr1marw buck~

ling stress of ,t~is ~tocky post, therefore practically equal

to the stiffenerfs yield st~esso If the latter is assumed to
be the same as for the web" the fol'lowing result is obtained~
Required As ? 0.0005 b 2 (16)

In order to get a .feeling for this area requirement 9'

~t is assumed that the stiffeners are built of rectangular
plates with an outstanding leg width UglY which is 12 times
the plate thickness Uhf? (F.igo 8) 0 Equation 16 then leads
 2$1-20 ~20

. to g? b/18. That is p 8.. 9 foot deep plate girder would rae:>

quire stiffeners o:f the size 2 Pls6 - ,6 1t X 1/2 fY f} while a
girder with a web depth of only 4' c= 6 n would require 2 Plso
~ 3ft x 1/4t? 0 A stirfener pair proportioned as ju_st derived

(g = b/lS? h = g/12) exhibits a moment of inertia Is ~ 53Xl0~8b4?

while Is ~ 16xIO= 8b 4 is the value required by-paragraph 26 (e)
of the AISC Specifications (14). Therefore? a stiffener size
as derived here would" simultaneously ful'fil1 the minimmn

rigidity requirement for ~ntermed1ate st1ffenerso

Another stiffener arrangement consisting of a single'

plate with cross=sectional area of A~ 'is presented in Fig. 90
It will be referred to as a one-sided stiffenero The plate
is loaded along one of its eqigeso The state o:f stress in

this plate 9 at the moment when yieldipg sets in p is shown. in

Figo 9ao The axial ~oad which causes this stress d1stripution
is F.8y :::: ,0 25cr'TA~
¢
J ~
c . If this load were increased and ~he,~;
stiffener plate propo~tioned such that unrestricted yie14~ng

is possible prior to plate buckling, the. load would ~eacn,a

limiting magnitude of F sp = O.414cryA~ and cause a stress·

distribution as shown in ,Figa 9b. Equating each or phe
c~rrying capacities with the expected maximum. va~.ue of
Fa = 500 x
,10~ 4oyb :2 givea~
2
Yi_~ldi"pg along
...~he loading edge 2-- FA'~ -·000020 b
Yis·lding allover cross sect;1on~( A~ l:!: 0.0012 b 2
251=20 =--21

Thi.s i~ 204 to 4 times more than the area whic,h a doublee=

sided ~rrangement requireso It is interesting to note that
an as sumed participati,<;>n of the web does not affect, this

relationo If participation would be assumed$}" equal '-effec~iv'e

web areas should be 'given to the one~sided and the douQle-

sided arrangement 0 But then p by jus,t deducting the amount
which the web carries from FSi the remainder of the force
would cause the same relation between the one and the two-
sides cases 0 In other words p 'no matter what min1mum"re-=-

quired area would be specified, a one--sided solu·tion. would

11se at least 204 times the sum o.f the areas of stiffeners
made in pairs p provided the stiffeners are made of rectang-
ular plateso

A one~sided stiffener~plate may th~refore have to be·

about five times as heavy as' either of the ,plates in a twOJ~

sided"anrangement1when ten:;3ion field action is the basis Qf

designo This is more than would be ex.pected,o The reason

for this is' th~t currently stiffener sizes are generally

det'ermined using only a stiffness criterio:q.o' Taking··--momf3nts

of inertia about the axis at the interface between stiffener

and web, 'it l's re~dily s'een that St if the outstanding: leg of
a one=-sided stiffener is o.nly 26% greater than the w:!ath, of

one o~ a pair, it would provide as much lateral stiffnesso

Therefore~ w~th respect to atiffness J the 6rie~sided ~stiffener

would r~quire only 63% -of the area of a two~sided arrangement~

while 240% might be required to implement a tension fieldo
 =22

If an equal Ie g angle is used as _a one.-,sided stiffener» .-

the excentricity would be less severe and a $imilar, reason-

a 2
ing as given above would lead to As = 000009 b 0

In order to enstlre an adequate s11eE:lr transfer from t'h'6

web .pla.te into the stiffeners" the connectors (rivets .,or

welds) will ha've ,to be sufficient in nlll11bers and", size 0 No
matter whether the arrangement is one or two~sided the con~

neotors,have to build up~ over half the girder depth~ a force

that may be as much as Fa ::.: 0.0005 b 2 (Jy. It is not expected
that in actuality the shear transfer is.exactly constant per
unit length~ such that a linear increase ,of the axial force
over half the girder depth occurs Q But a re,quirement that

the stiffener force Fa be built up over a third of ,the depth

should provide enough tolerance for nonuniformity in shear
flowo The connectors then need to be proportioned so as to
provide,. at ul~imate load, a shear flowiper unit length of
intermediate stiffener of 3F sib = o. 0015 bay 0 For A7 steel
w~th a yield point of 33000 Ibs/in
2
and factors of safety
of 1065 or 1083 the required shear flow for which ,rivets or
welds have to be designed at the commonly applied allowable
stres~es is 30 b or 27 b$ respeetivelyp where b is in inches
and the shear flow is given in l~s per linear incho
..
The ar,e~ requirements of this section are deri1va.d, under
the assumption that the stiffener should not fail before the
ultimate shear strength of the adjacent panels is reached9
251~20

In girder seet~ons p~edominantl~ subjected _t~_b~~di~~!

failure due toshe'ar cannot occur and, .the required stiffener

area may be reducedo Assuming the shear force ,V to be
carried in beam action until the critical shear'rorce Vcr
is reached~ only the excess V «=- Vor carried in tension "'~ield

man.ner causes axial forces in the transverse sti:ffenerso

Hence the required value of stiffener area ·1,8 obtained by
reducing the above speci~ied values~ which' are obtained
unde~ the ultimate shear force VU9 in ,the ratio (V~Vcr)+

(Vu~Vcr)~ A simplification to the conservative side would

be a reduction ratio V/Vuo In terms of allowable stresses
this would be ~/'"tW9 rt b,eing the highest shear stress that

can occur under any comb'ina.tion of life and dead Ifbad in

one of the adjacent panels, and 't w the as'sociated maximum
permissible shear stress in that panelo
2$1=20

For the end. panel of a girder the boundary conditions

are different than in an inte~mediate panalo When web
yielding sets in~ there is no neighboring plate serving as
anchor for a tension stress fieldo Bearing stifrener at
the end of a girde~, together witA an, extending portion of
the web~ may offer partial restrainto Also~ a more or less
pronounced gusset ,plate action in the upper corner of the
end of the web» as indicated in Fig o 14p may help to develop
a partial tension actiono But the degree of this eontribu~

tion is uncertaino Since beam action does not depend on

tens1on~reslstant boundaries, the computed shear forces Vcr
and Vu are the limits between which the ultimate shear
strength orHan end panel li~so

Prema tur e failure of girders due to t:q.~.; failure o:f end

posts was experi-ence,d and ,r:~ported in Ref 0 ·11. A picture of

a failed girder end appears in Figo 100 The web and end
st1~fener eom~ination was not strong enough to r~slBt the
horizontal component of the tension field stress. An
appreciation of the curvature imposed on it is obtaln~d by

noting the plumb line hanging over the girder end and the
compres$ive yielding at_ the outer edge of the stifrener
flange and the ,web toea
251=20

In order to exclude the possibility or a premature end

panel railure in an unrramed girder~ as for example a bridge
girder suppo-rted on masonry,.9 the're a;re two basically different
approaches possibleo The simplest and generally most econo~

mical solution (one which would have to be used on framed

girders) is the choice o~ a stiffener spaoing ror the end
panel such that the computed average shea.ring stress does not
greatly exceed ~cr/Np where N is the factor of safetyo A
provision for such a limit is already established with the
spacing requirement of"Seco 26(e) of the AISC·specifications(1L~)
and Art. 1.6.80 of the AASHO Specificationa(15).

To illustrate tnis p let it be assumed that the smaller

dimension of the web plate be denoted as na ft and the larg~.~

as "bno The limiting" shear stress is then

where the value k[b/a) varies only bet~een 9o~4 (a square

plate) and 5034 (an infinitely long plate) (12). Assuming
an ave~age value of k = 7034 and a factor of safety N = lQ65
the previously given condition becomes

.~ (17a)

If the modulus of.elasticity E,an~ the shearing stDess ~

are expressed in pounds per sqt1.are -t~1nch and the inequality

is solved for a p the well known expression of Arto 26(e)>>

Hefo 14~ is obtained
(17b)
 e=26

While the first possibility to elimJ.nate plqemature

end panel ~ailure exists in avoiding the development of a

tension fi-"eld in that panelj) an alternate solution is to
make the end post resistant to membrane tensiono T,his
could be done by bending down the' top flange at the end
or the girder or by welding an independ$nt plate to the .endo
The required end post dimensions could be estimated as
follows 0

For simplification let it be assumed corlservatively

that the tension stress field would act under an inclination
of 45° and be uni:rormly distributed over the entire girder
depth (Figo lla) 0 Then the tension field would subject the
end post to a vertical compressive force of Va = V~Vcr (the
excess not taken by beam action) and a horizontal load of
the same magnitude 0 Thus the maximum bending moment to whioh
the end post is subjected amounts· to Va~b/8o That is~ a
compressive and a tensile force of Vcr ob/8e are introduced in
the end ,plate and bearing stif.fener res.pectively» (Figo llc)~

The des~gn· of the bearing stiffener by the limi ta.tion of the

be~ring pressure is still adequate since the new force com~

ponent induced is tensile and is superimposed on compression~

Figo lIb 0 Equating the re,~isting force AeO y offered by the

end plate with the force component of the bending moment leads
to Ae(Jy = (V=V cr )b/8eo In terms of allowable stress and by
substituting EqQ 17a for 'tcr/N $ the following expre'ssion .for
 -27~,

end post proportioning is obtained:

(17c)
 50 THE INFLUENCE OF STRAIN HARDENING

For all compression elements made of mild steel there

exists a range of low slenderness ratios within which the
actual ~ailure stress exceeds the yield stressu This is
explained by the fact that yielding is confined to slip
bands 0 The steel next to.,,:these bands is only on the verge
, - -

of yielding whereas that within t~e bands alr,ea9-Y ~aa strain-

hardenedo ThuB,9 the mem.ber neyer loses all its~1g1ditYj}

and it is a mistake to assume that all buckling curves must

end at the ·yield level as .the slenderness ratio .approaches
zeroo HaaiJer and Thiirlimann (16) have determined the
range of slenderness ratios in which compact columns and
plates SUbjected to edge compression will atrain~harden be~

fore reaching their limiting buckling stress~ So far there

exists no similar theoretic~l treatment.for the case, of shear,
although it appeaFs to be of signiricance since the webs of
all ~olled sections are p~oportloned on the assumption that
the limit of their shear carrying capacity lies in the
t
stra1n~

ha~dening range"

In order to obtain ~n estimate of the shear strength in

this low web slenderness range, recourse can be made to ex~

perlmental w?rk~ Fortunately there exists a series of tests

carried out by Inge Lyse and H. J. Godfrey (7) which cover
the range of web depth~to-th1ekness ratios from 50 to 700
 =-29

The tests included welded plate girders without intermediate

transverse stiffenel~S.9 whose test data: 'are s,unnnarized in the
upper left quadrant of'Table 10 How far the e~perimentally

ex
obtained ultimate shear. force~ V-u 9 exceeded the plastic

shear force Vp is seen from Figo 12~ w~ic~ uses the same
coordinates introduced before with Fig o 60

It becomes apparent in compa.ring these test results

witn Figo 6~that the conventionally applied reduction pro~

cedure as stated in connection with Eqo14 is too conc=

servativ60 Let it be assumed that a correlation of the simple
form ~cr = C·~crin exists above the proportional limit with
C and n to be determined. Sinoe the id~al critical stress
't" cri and the "actua.l or i tical stress 'tor have to be equal at
the proportional limit ~prj the value c~ is de~ermined as
C = ~pr(l~n)o Trials to fit now the Lyse~Godfrey test data
indicate that~ for ~pr equal to 008 ~y, 005·is the best choice
for the' exponent no Thus, the reduction formula becomes

(18)

which affords a much more realistic estimate of the shear

strength in the inelastic and strain~harden1ng range 0

Since the difference in shear strength given by Figs 6. ,0

and 12 are based upon the p~esenee of local strain=hardening p

which must be preceeded by shear yielding j
the shear resisting
 ~30

capacity of t~e web cannot be assumed as being significantly

augmented by the development of a te~~ion field when~cr~~yo

Hence~ when substituting Eqo 18 in Eq~ ~14~ the second term'

in the righthand side must be omitted if't'er>rr: y <>

Because Figo 12 can accomm~date but two parameters~

namely the stiffener spacing-a and the web slenderness ratio

13, the curves can be exact for only one value-of cri, inthia
case taken as 33 kal. Such a value for cry is approximately
correct for the beams of Group B in Table 1" For c,ompar1soDS'
the curve for Group A beams where cry equals 47 kai is shown
as a dashed line 0 For all other test results the ratio
vu/vp was adjusted for a normal yield stress of 33 ksi, thus
showing graphically the deviation from the theoretical pre-
dictiono

Figo 12 also gives a survey of all the predominantly

shear-type tests ever carried out at Fritz Engineering
Laboratory on welded pla'teg1rders 0 The necessary test
information is summarized in the left half of Table 1; to
the right of this table the theoretical predictions are pre-
sented ror each individual testo The last column gives the
correlation betwe·'en theory and experiment 0
251-20 ~31

6 ¢ ALLOWABLE SHEAR; STRESSEg

With Eqo 14 and the modification for stra1n~hardening

presented in the previous section, the f~ction ~f(a,~~By)

anticipated in Eqo '1 is established and plotted in Figo 12

for the special case of' e y = 0"0011,, Using Eqs .. 1 atld 2jl.the
ultimate shear force can be expressed as given in EqG 19ao
Dividing bo'th sides by a factor of sa.fety Nu against ultimate

load,9 the allowable shear force,9 Vw_~ appears in Eq,o 19b4>

Wi th the conventional she-ar stress com,putation (shear :force
divided by the web area) the allowable stress is ex,pressed
in Eqo 19co
(19a)

(19b)

(19c)

In ass1.Ul1ing a constant factor of safety Nu == 1~65 and

a yield stress of 33 ksi, the allowable shear stresses for
various values of a, and 13 .,are plotted in Fig. 130 Sf'nee'
the shear yield stress is''ty ;:: 33/{Y :=: 19 ksi~ the nominal
factor of, safety aga.inSt yielding is 19/13 := 1 .. 46 in the
presently used AIS·C Specifioations Q T'here appears no reason
to change this margin which is merely a ~argin. against de ..
formations greater than those associated with elastic shear
 ~32

strain and not against catastrophio failureo Therefors l

the maximum allowable shear stresses are limited to 13 ksio
With the same reasoning the allowable shear stresses for
the AASHO t3pec1fications could be fixedo - The proposed

factors of safety against ultimate load and yielding are

Nu = 33/18 = 1 083 and Ny = 19/11 = 1073~ respective1yo

In the range of high web slenderness ratios» the

stiffener spacing should not be arbitrarily largso Although
the web might still be sufficient to carry the shear 9 the
dlstort1on~ could be almost beyond control in fabrication
and under loadc Currently.9 the AASHO and the AlSO specifi~

cations limit the maximum stiffener distance to 6 ft and 7 ft~

respectivelyo With a. minimum web thickness of 5/16"9 this

distance according to the AlSO Specifications could not exceed
270 times the web thicknesso It is suggested that such a
relative measure 9 rather than an absolute one~ be used to
specify the maximum stiffener spacing in the range of high
web slenderness ratioso The rule that the shorter panel
dimension does not exceed 270 times the web thickness when
a < 100 was used to te~minate the curves of F1g~ 13Q The
',justification for such a rule can be found in the fact that
the resistance of a rectangular plate to transverse loading
is essentially governed by the ratio of shorter span to plate
thickness 0·
 ~33

In the medium range of web depth~todthi~kneB~ ratios~

the cutt=off curve is arbitrarily taken as a straight ,",line

between the points a =~, ~ = 170 and a = 1, ~ 0 270~ These
limits may be too liberal for certain easeso The designerts
judgement~ however 9 is still needed to determine the density
of transverse stiffeners when rigidity and stiffness criter-
ions p or fabrioation and er,"eetion aspects, are governing
rather than mere strength considerationso

Flnally~ in the low range o~ web depth~to~thickness

ratios~ detail considerations generally determine the loca~

tion of stiffenerso A direct application of load onto the

web can be made for webs which are pro,portioned to allow a
certain degree of stra1n-hardening(that is~for webs whose
allowable shear stress is 13 ksi)~ provided the co~pressive

stress at the web toe of the fillets are suitably limited

in order to avoid web,.cripplingo For' all other cases the
load application must be made by means of transverse
stiffeners, unless it can be shown that the transverse
pressure is small enough not to cause vertical buckling of
the flange into the web, as discussed in Sec·o 201 of the
previous paper on bending strength of' plate girders (18) ()
251-20 -=-34

~7_o~A~E_PE_N_D_I_X~~~R~E_V_I_~_._O_F~T_9~-
~~~~·~H~~

STRENGTH_THEORX

In order to determine the shear strength of a ~irder

it was assumed th~t the girder acts aecord:tng to beam the.ory

• II " . , " ' ,...

up to the critical load and thereafter in a tension field

manner up to the point of web yieldinga Considering initial
web distortions and residual stresses due to weld1ng$ the
probability ofa clearly defined boundary between these two,
types 'of beh,avior a.ppeaJ:ls questionable 0 The assumption
should, therefore, be ~aken as an estimate for the amount
of shear to be carried in compression rather than a pheno~

menon that can be actually observed.

If a plate subjected to shearing atresses is assumed to

be built of two sets of strips orthogonal to each other and
acting in cOl11:preasion and tension respectively~ the com-
pression' strips are then elastically supported by the tension
strips 0 " This is the reason why the or1 tical stress of a.
shear panel is muoh higher than that of an equivalent isolated
compression strip extending from pa.nel bor'der to panel border (t

If more tension is superimposed on the tension stripsg it is

obvious thatj) until yielding occurs, the conditions for
stresses in the compression strips are improved since the
Uspring constants n o:r the ela.stic sup,ports increaseo But
when yielding sets +n, the ratio of shear carried in beam and
in tensiori field action changes againo From the recorded
251-20 -35

load def~ect1on curves in Refo 11, which generally exhibit

a pronounced yield plateau, it is seen that the reduction in
beam action is about compensated by the gain in tension field
action due to the increase in field.wditho Arter all, the
derived ultimate shear force expression is riot very sens'itive
to a slight error in this assum,ptiono If the contribution
~~ in Eqo 8 would be overestimated$ the yield condition would
allow less tension field action and vice versa.

Some remarks as to the geometry of the tension field are

in ordero The inclination of the tension field was obtained
fran a. maximum value c,ondition as they "were assumed to exist
on a model girdero The assumption was made that the edges
of the effective field would run twough the .panel corners
(Fig~ 4)~ and the two web traiangles not included in the
tension strip were not thought to be idleo They were supposed
to digest the membrane stresses oaused by the neighbori~g

panels t tension fields and ultimately transfer them to the

flanges 0 The shear force thus created is given by Eqo 5 and
the resulting stiffener force by Eqo 60 Let this derivation
be called Approach Ao

It can be shown that the established expression for Vcr

is not very strongly dependent upon the above derived tension
field geometryo Another approach, Approach B, would lead
to the same resulto Let it be assumed that the girder carries
the load in a trusa~type manner p that is~ that a tension
251.--20

diagonal forms whose centerline coincides with the panel

diagonal and whose effective width, so far unI:n.o~~. _i"~ . ~"'~

certain fraction of the 'girder depth, say ~b~ This would

lead to a shear cO,ntribution Va z debtO't· sin<p'$) where

, = co~-l~~ Hence,
V(J ::1

When a approaches 00, Va dimin1shes to-zero. At the other

limit, a = Q, the highest possible ·tension field contri~

bution as derived in Eqo 3 should result o This fixes the

constant ~ ~s equal to 1/2, and it is observed that Val • 'J'

derived this way, is exactly the same as in Eq.: 5 of Approach Ao

Using this secohd approach, however, the s,t'iffener forc~

Fs would ~qual the shear force Vao This would mean that even
for very cl'ose stiffener' spacing each stiffener would have to
carry the full amount . of, Va"' which is obviously inoorrect 0 The
reason for this defioiency 1s the fact that for small values of
a ,I, this truss assumpt"!on violates the yield condition because
the strips of nelghboJ:l1ng 'panels over:+ap and use p,ort10,ns of l•

the web twice 0 "But the longer the panel" the closer is the
agreement between the stiffener forces of the two approaches" 0

For a,» 1, Eqso 5. aI,1d 6 yield the same result 0 I~de,ed~, it

will be indicated next that for la~ger valu.es ora and lower
slende~ness ratios the tension diagonal. of Approach B is not
an impossible one, althOUgh its centerline, rather than it~
ef.feetive edge,P is assumed to run thl'oue;h the panel corn'erc
 (:>37

For simplification j the plate corn~r may be considered

to be built of two sets of orthogonal strips9 one set under
tension stresses 0tp the ~ther under compression stres~e~ crc~

as illustrated in Figo 140 The .postulate that no membra.ne

stresses orthogonal to the plane of the' tqp flange can occur
leads to the relation cr c ~ atotan2~o In approaching the
corner the cbmpression strips can pick up stress because the
strip length shortenso According to the above given relation
this will be accompanies by a gradual increase of tension
stresses as sketched in the figurso If this gradual increase
of tensile stresses is idealized p the tension field may be
thought to 'have an effective width intersecting the top flange
at point F~ some distance away from the corner A of the panel 0

It is seen that this distance increases with the web thickness

and decreases to zero for an extremely thin webo Figure 15
is a photograph of a panel with an aspect ratio a = 1050 It
is taken from a girder with a sturdy web (~ = 133) and shows
that the "edge" of the tension field is shifted away from the
panel cornersQ Both the tension field (Approach A) and the
. tension diagonal (Approach B) are shown in Qutlineo The
tension field actually developed can, in this case" be con~

sidered to be between the predictions of the two approaches 0

Finally, an Approach C might elimina'te completely the

idea of tension field inclinati'on and t;:lffective' widtho This

approach seeks merely for the simplest set of coordinate

 functions which yield a resu+t that mig~t commonly be ex~

pected from a ten$ion field typeactiono In order to f'ree

the functions from dimenslons~ Vcr is to be expr~~sed ~s

Vmaxf(a)~ where Vmax is the highest p03~i~le shear ~orce to

be carried by tension stresses ~lone(Eqo 3) and is here
considered to be the uamplitude 1t () The pur,e coordinate
function f(a) must be normalized such th~t f'(a=O) := 10 -For

larg~ values of a 9 it should approach ze~o with the power

of 1/0,0 The mos't" obvious choice for a simple .function fule:.
filling ttiese two limits is
~~~ ~
f(cd =: (l+an )n..~: n > 1 (20)

The result previously derived in Approaches A and B ,would

be obtained for n = 20 Since no physical significance was
noted in this approach~ it does not give any detail infor~

mation at allo For instanc6$ since no stress flow is

pictu.:red,9 an equilibrium condition cannot be· a.pplied t·o get,
a stiffener forc8 o

The ":outoome of this re.view can be sunttnari,zed as follows '0

. "By extracting the proper. parameters and'" q.onsidering the lim! t-

ing oases,S' a close guess as to the nature of the expression

for the ultimate ··shear force is possible 0 As soon as a more
specific problem is cons1dered~ such as the stiffener rorce,
a higher rung in :the ladder of condit1ons(.proper parameter p
equilibrium condition, yield condition~ compatlbilit~
~51=20

condition)must be usedo Approach C uses only the correct

parametera~ Approach B also fulfills the equilibrium ,condi~

tion but violates the yield condition~ Approach A fulfills

all three but ,is still short of the last conditiono Therefore$'
when a detail problem is encountered$ such as the determination
or the weld size between stiffener and web~ not even the re~

:{'Uined model used in Approach A c'an gl.ve more than an' average

valu~o The stress flow would have to be pictured more

accuratelyS', but this could no longer be done with tension
strips unde~ constant inclination and constant effective widtho
It should be kept in mind that this idealized model is an
engineering tool which provides the means ror a good jUdgement
of a girder l s carrying capaeityo Its ability to account for
an overall carrying capaoity is very good but a~'local points
the actual state of stress could be t~ite different from the
assumed oneo However 9 the situation is not unique to the
problem of welded plate girders'o In practice, welda and de~

tails are proportioned on the basis qf idealized conditionso

 ACK~Oj[L~DGMENTS

This report is based on a dissertation (17) worked out

at Fritz En,gineering Laboratory)) Lehigh Un1versity'~

Bethlehem 9 Perinsylvan:las> or which William JoEney ia the

Head 0 The Dir eo tor of the Labora tory is Lynn. SO' Bef3clle ()

The investigation was jointly sponsored by the American

Institute of Steel Construction, the Pennaylvania Department
of Highways~ the Uo So Department of Commerce - Bureau or
Ptlbli,c Roads.9 and the Welding Research C.ouncil Q It was

supervised by t~e Welded Plate Girder Project Subcommittee

of the Welding Research Councilo The financial support of
the s,ponsors a!ld th,6'" guidance which the members of the sub ba

committee have given to this l~es,earch work are gratefully

acknowledged"

The author owes many thanks to Bruno Thllrlimann who

was professor in charge of the dissertation and former
Project Directoro His advice and suggestions are sincerely
apprec1atedo Thanks are also due Messrs 0 L~- S Q Beedle$!
Bo T. Yen, To R~ Higgins J and WQ Ao Milek for revi~wing the.

manuscript 0
251-20

NOMENCLATURE

a : Distance of transverse stiffeners

b g Depth of girder web
e : Distance, defined in Figo 11
k Buckling coerr1cient
s : Tension field wi'dth
t Thickness of web

A ~~ Area of cross section

A's : Cross sectional area of a stiffener ,pair

A's •., Cross sectional area of a one"'sided stiffener

E ~
4 Modulus of elasticity (30,000 ksi)
F Force
V Shear force

a :::; alb Aspect ratio, panel length to panel depth

(3 = bit : Sl.ende1 nes s ra tic, web depth to web thiokness
1
 f 0
0 Flange

p Qc
0) Plastic
pr Pro.portional
s 0
(I Stiffener
t 0
0 Tension
th 0
6 Theoretical
u 0
I) Ultimate

w Q
I) Web, working
y g Yielding

a • As carried in ten.sian
0

't' 0
0 As carried in shear
 Table 1 : Summary of Shear Tests on Welded Plate Girders
rExperimentaY\TalUeS
. a Q b - .
, 'r (Tv
-Theoretical
I) l"cr 2 )
values~
V u I) 4)V u
ex
Source Glrder a=- tJ=- A (T.. Vex -r:'I 113
=-..l- T .- -- -- V th - - ,
No. b t w Y u crl Ty Vp u V th
in2 ksi k ksi ksi' k U
-
Group A
WE ..1 3 56.5 3047 43.3 109 2500 4806 1.25 1 0 25 108 10-01
WE-2 3 5409 3.57 47.8 128 27.6 51a3 1 22
0 1.22 120 1007
Rero 7 WB-3 3 .$8.9 4.36 49.6 139 28.6 44.6 1.12 1.12 139 l-,~OO

WB-6 3 70.0 '4040 33.1 96 19 10 31.6 1.15 1.15 97 0099

Group B WB-7 3 60.b- 3088 33.7 95 19.5 42.3 1.32 1.32 100 0095
Rer o 7
WB-8 3 59.7 4.10 2907 100 17.2 43.5 1.43 1~43 101 0.99
WB-9 3 50.0 3~c 12 30.3 92 17.5 62.0 1.69 1.69 92 1.00
WB-IO 3 49.4 "~:t'14 30.3 94 17.5 63.6 1.71 1.71 94 1 000

G6-Tl 1.5 259 9.65 36.7 116 21.2 2.85 0.134 0.550 112 1 0 04
G6-T2 0.75 2.59 9.65 36.7 150 21.2 5.40 .254 .771 157 0.95
Part 3 G6-T3 0.5 259 9.6-5 -36.7 177 21.2 10.1 .477 .882 180 0.98
Rer. 11. G-7-Tl 1.0 2.55 9.80 36.7 140 21.2 3.85 .182 .682 142 0.98
G7-T2 1.0 255 9-.80 36.7 145 21.2 3.85 .182 0682 142 1.02
EI-Tl 3.0 131 19.1 41.7 278 24.1 9.00 .374 .545 250 1.11
Part 4 EI-T2 1.5 131 19.1. 41.7 290 24.1 11. 1 • 460 '. 719 330 0 • 88
Re:f. 11
G8-Tl 3.0 254 9.85 38.2 85 22.0 2.30 .105 .350 76 1.12
G8-T3 1.5 254 9.85 38.2 _117 22.0 2.95 .134 .550 119 0.98
G9-T3 1.5 382 6.55 44.5 : 79 2,5.7 1.31 .050 .506 85 0.93
Standard-deviation 0.60
. 1'f
2
E t26750r,';1- . 4~00 . 5.34
1) T'cr i == k 12 ( l;..V:J <'b) ~ == k {j 2 a.k s i.J,. with k == 5. 344- (i2 for Q>l, and k == 4.00+ (i"2 for a< 1

) T'cr = T'cr:t when Tcri ~

o. T Yr and Tcr :::';0.8t"yTc~i for Tcri > O.8Ty
3) Computed according to Eq. 14 when T cr E T y ; for T cr > T y , simply vu/v p =: Tcr/Ty
4) Values under 3) multiplied with TyAw
251-50 -44

1 States of Stresses

2 Tension Field Action

3 Yielded Shear Panel, 50 tf
x 50", 1/4 tf
Thick

4 Equilibrium Conditions Applied to Free Body

5 Approximation of Yield Condition

6 Dependence of the Ultimate Shear Force
7 Dependence of the Stiffener' Force on ~ and ~

8 Stiffeners Used in Pairs

9 One-Sided Stiffeners
10 End Post Failure
11 Detail at Girder End
12 Test Results of Welded Plate Girders
Subjected to Shear
13 Proposed Allowable Shear Stresses (A.I.S.C,)

14 Gusset Plate Action

15 Approaches A and B
 b)

c)

F~g. ~

. .L::tV=qbt.cos¢. sin cJI

b)

[Flg~ 2
 ~

Fig. 3 Yielded Shear Pan1el. 50" x 50". V4" Web

251-20 -Lt.?

Fig, 4

-20 ~~

--10
o ,~---
~------i

o ~/4

Fig~ 5
 2;;1-20 -48

~
uti> (J '4>-_

,
,J~

'" ,~

,
'"Ot

,_~(:t
if:.-o.t.1~~~,)
i ~~

,
(::~

~ro
~
t 1 <tL

,,I
It
a
1
J
lj
,~

I ,'"
0
0

/
rn

"

o
o
(\1)

_~·:r;Z:;:"'"-"",-·"...,.,,,~,.~","·',·,,,,,---,·,-"~",-,,=~····~·,"'~~'··-·"'-~{t-='='="""'-""'="""-·"'~-~-'·"'·_'''"·"==''='=''='·'·''''''''·'"'''·'''''i1'''~-~."'""-=-"""-- ..""""""..f ==dJ'''--=-,'-=.... ,~'''~''''''''''''''~... 0

o Q
 -49

o
/ C\i

o
/-:

o
/
t

o
+-+
0'
 Fig. I 0 End Post Failur,e

Fig. 15 Approaches A anJd B

251-20 -52

f-
~J

. , . ~1

~. ~' ... ~,
~. .•... _ ~ _ ~ - ~-:o:'-..f.

I Q~~>.~'
~"Gt''}-~-~--_.__._._~.~, n ~,_._.".,-~_._.__ ._-~~-~~-._.-----;l.:,
 -53

"'=>'-==0'",'=.' lE;) \f\feldedJ Beams Il Group A

~ \rVelded Seams ~ Group B\
"~ P ~ ute G;rders ~ PpW'~ 3~ 4

WB . . 7
WB-i
i.2 WB-2
WB-6
WB~3

u=o/b

0.8 ~/2

0.2

Fig, ~2
 ~54

Ey =OY/Ec 33/30000
Nu :: L65 l' Ny =1.46

~~\\\ \ \

{)()

t---+---t--+-----+----+---+--+----+---+----+-.-~--+----+- I
11I:
I
b
L
t
'

o "'""---..e.-...~~
rp~ I
-+-~.--!-----'----..J..-I-__J......__.t.____L,___..lr-.-,~~~~~~
....
1_·

o 100 200 300 p= bit

Fig. 13
251-20 -55

Fig. 14
251-20

REFERENCES

1. BeaohJ) Ho To
A THEORY FOR SPACING-STIFFENERS IN PLATE GIRDERS
'Erigineeraiiig News,9 Volo 39,9 Nco 20» po 322j)
May» 18980
Discussions
Engineering News~ Vo'lo 39, No~ 23,9 .Pi) 370;
Volo 40p Noo I, po 10; and Vol. 40~ Noo 6~
po 90; 18980

Wilson, Joso Mo
SPACING STIFFENERS IN" PLA.TE GIRDERS'
En§ineering News, Vol. 40, Noo 6, po 89,
August 1898.
Discus sion.s
Engineering News~ Volo 40, NOe 10, p. 154,
September)) 18980
Turneaure, Fo E.
TESTS OF THE STRESS IN PLATE GIRDER STIFFENERS·
Engineering.News p VOlo 40, Noo 12, po 186,
September 9 18980
Turner p Co Ao Po and Shinner~ Fo Ge
SPACING STIFF'ENERS IN PLATE GIRDERS
Engineering NewB» Voll) 40,9 No. 25" po 399 9
December$' 18980
Beach p Ho To
SPACING STIFFEN"ERS IN PLATE GIRDERS
Engineering News, VOlo 41p Noo 7~ pc 106,
February 18990
Discussion
Engineering News, Vol. 41, Noo 15, po 234, 1899.
Moore$ Ho Fa and Wilson, W. Mo
University of Illinois Bulletin 86~ 19160
Lyse j) I G and Godfr ey.9 Ho J o'
4

INVESTIGATION OF WEB BUCKLING IN ST~ B~MS

Transactions AoSoCoEo, pa.per NO(j 1907, .p. 675,1935.
Rode ~ Ho Ho
BEITRAG ZUR THEORIE DER KNICKERSCHEINUNGEN
Wilhelm Engelmann Verlag, Leipzig~ 1916 (Diaser,tation) p
and EisenbaujJ Volo 7, ppo 121, 15'7~ 2~0;9 2)9, 296,
19160 ' I '
251.e20 -57-

9 I) Wagner J) ,Ho II' ••

EBENE BLECHWANDTRAGER MIT SERR DUNNEM ST-EGBLECH--
Zeitschrif~'rur Flugtechnik und Motorluftscniffahrt
Yolo 20~ ppo 200, 2279 256 J 279 & 306$ 19290
100 Kuhn>, Po g PeterSOl:l)1 Jo PQ~ ~nd Levin» Ro L~
A SU11MARY OF DIAGONAL TENSION
N oA CoA 0» Technical, Note 2661» 1952,0
0

ty
110 Basler» K~ ~ Yen.9 Bo To Mlle~ler~ Jo Ao and Thurlimann:; Bo
j

WEB BUCKLING TESTS ON WELDED PLAT'E GIRDERS

Welding Research Counc11~ Bulletin No. 64~ Septo 19600
120 ;Bleich» ,Fo
BUCKLING STRENGTH OF METAL STRUCTURES
McGraw~Hill Book, Company ~ New 'York 1952 j I)

" fY
Basler p Ko and Thurlimann~ Eo
PLATE GIRDER RESEARCH
AoloS6Co National Engineering Conferenc~, Proceedipgs,
19590
14 Q Ao loS. C Q

STEEL CONST,RUCTION fY MANUAL

American Institute of Steel Oonstruction, New York,
19590
15¢ AQAoSoHoOo
STANDARD S-PECIFICATIONS FOR HIGHWAY BRIDGES
American Assoe,iation of state HighwayOrricials p
Washington» n.co p 19510
160 Haaijer.9 Gc and Thilrlimann. p Bo
ON INELASTIC BUCKLING IN STEEL
Proceedings~ AoSoCoEo 9 Paper NOQ 1581~ April~ 19580
17" Basler;> Ko
STRENGTH OF PLATE GIRDERS
Dlsrsertationj) MicQ 59=-6958 p University Microfilms
Inco, Ann Arbor 9 ~ichigano
:y~
18 ~ Basler IJ Ko ~and Thu!~limann~" B. ",
STRENGTH OF PLATE GIRDERS IN, BENDING
Fritz -Engineering La.bor.a~tory~ Rept ~ No 0. 251-19,
Lehigh Un1versity~ Novo 19600