6

prevented since twist of the cross section

is also restricted. I the case of twin
beas with a diaphrag or cross fame
bet ween t he members, l at eral
displacement of the system is peritted at
the cross fame. Tis location, while able
to displace laterally, is still considered a
brace point because twist is prevented.
I general, bracing may be
divided into two main categories, lateral
and torsional bracing, as illustrated in
Figure 1.5. Lateral bracing restrains
lateral displacement of the top fange as
its nae implies. Te efectiveness of a
lateral brace is directly related to te
degree tat twist of the cross section is
restained. For te uniforr moment case
illustated in Figure 1.1, the center of
twist is located at a point near or outside
of the tension fange. A lateral brace is
most efcient in resticting twist when it
is located at the top fange. Lateral
bracing applied at the bottom fage of a
simply suppored bea is almost totally
inefective. A torsional brace can be
diferentiated fom a lateral brace in tat
twist of the cross section is restained
direct1y, as in the case of cross fames or
diaphrags located between adjacent
stngers. Braces can be continuous along
LATERAL BRACING
h
Beam "b" has lower load so it can
late rally brace the top flange of girder
a b
.. buckled shape if framing is weak

->-: buckled shape if framing is strong

.
" Girder T
op
Flange Framing
PLNVIE
Metal Deck Forms
TORSIONAL BRACING
Cross Frames
loor beam
Decking
r : SHE
SECTA-A
Figure 1.5 Types of bracing.
the beam span as wit metal deck fors or located at discrete points as with cross faes. Sore systems such a
concrete decks can act both as lateral ad torsional braces.
I te case of a wood or concrete deck resting on stee1 stingers, tere ca be restaint fom diferent
sources. Te fiction that may be mobilized at te deck-bea interface acts as a lateral brace, since it restains lateral
movement of the top fange as shown in Figure 1.6a if te deck has lateral stifess. As the bea ties to twist
during buck1ing, te deck plaks provide torsional restaint in two ways. For the plak supporting te wheel loads
there are contact forces Po on the beam as shown in Figure 1.6a. Tere can be a restaining moment M 6EI9/S
provided by te deck even when tere is no positive attachment between the deck and steel stnger. I such cases
M will produce contact force PI given by the relationship PIa 6EI9/S which is valid ifPI < Po. Wen the agle
e gets sufciently large so PI Po, one side of the flange will separate fom the wood plak ad the restaint
provided by the wood deck (6EIS) goes to zero. However, at this stage the force on the other fange tip is 2Po
which provides a benefcial restoring torque for lateral stability called "tipping efect." I reality, te bracing may
16
where
(2.6)
Te Tong and Chen solutions do not consider web
distortion which ofen has a sigifcant efect on te
bracing requirements. Teir solutions should not be
used unless substantial stifeners are provided at te
brace point. Equation (2.5) gives an ideal stifess of
1450 in-klradia shown by te solid dot in Figure 2.11.
Iftis stifess was provided, a 6 X 3/8 stifener would
be required to reach the maimum buckling load. If a
2.67 X 1/4 stifener is attached, te buckling load is
reduced by 14%; no stifener ad a brace stifess =
1450 in-klradia gives a 51 % reduction.
4
4 x 1/4 stif
-
/
3
"
/ yielding
L.----mldspan brace
1
--------------------
no brce
100 2000 300 4000
TOTAL TORSIONAL BRACE STIFFNESS (in-krad)
Figure 2.13 Continuous torsional bracing.
If continuous bracing is placed on the top
flage over the entire span with no stifeners, there will
be a increase in buckling stength as shown by te
heavy line in Figure 2.13 compared to midspan brace
cae. Te light lines are just a repetition of te sigle
brace at midspan solutions presented in Figure 2.11.
Even with continuous bracing, te 20-f beam canot
reach load levels associated with a single brace at midspan with a stifener to contol distortion. The continuous
bracing, however, does increase the buckling capacity sufciently so tat yielding, not buckling, contol s the bea
stengt. Te moment at frrst yield is shown in Figure 2.13. Figure 2.14 shows tat torsional bracing on te tension
fage (dashed line) is just as efective as compression fage bracing (solid line), even with no stifener. Ifte beam
has no stifeners, splitting the bracing equally beteen the two fanges gives a geater capacity tan placing all te
bracing onjust one fange. Te dot-dash curve is the solution ifweb distortion is prevented by transverse stifeners.
Te distortion does not have to be goss to af ect stengt, as shown in Figure 2.15, for a total torsional brace
18

20

",
1600 'i
' _
-
-
-
-
-
--

.

brace

1200
_
.
.
@
,

cr
,"
( in-k)
/
B00 _.'
t. W16x26 W "e
e 4Q-
400
,
8

_ e g
7

(kips)
4
o
3braces
0

0 400 800 1200 1600 2000

TORSIONAL STIFFNESS QEACH BRACE (in-kradian)
Figure 2.17 MuItiple torsional braces.
lTOP Flange Load
t W1ZX14- Z4R.

at 10 f. I the lateral brace case the three brace

system requires 1.7 times the ideal stifess of the
single brace system, as shown in Figure 2.4 and Table
2.1.
Te efect of diferent brace aragements ad
stifener depths are shown in Figure 2.18 for a W12 X
14 section with a spa of 24 f with top fage loading
ad one midspan braceo Seven torsional bracing
conditions were studied:
a)
b)
c)
d)
e)
f
g)
compression fage brace - no stifener
compression fage brace - 3/4 depth stifener
attached to the tension flage
tension fange brace - 1/2 dept stifener
attached to te tension fange
compression fange brace - 1/2 dept stifener
attached to te compression fage
centoid brace - 1/2 depth stifener not
attached to either fage
compression fage brace - 3/4 depth stifener
attached to the compression fage
compression fange brace - fll depth stifener
attached to bot flange.
1 Z o 4 Te results are typical for a number of diferent spas
1LClLPL PLL C1lLCC (in-kradian)
Figure 2.18 Torsional bracing details.
ad cross sections tat were analyzed. If te stifener
is not attached to the fage that is braced (curve b),
the behavior is very simila to the unstifened case.
Such a brace system is not capable of forcing the bea
to buckle between the braces. If a half-depth stifener is attached to te braced fange, the bea is able to reach the
ideal brace condition but te required brace stifess is almost twice tat for a stifener tat is at least 3/4 dept.
Tere is no diference in perforance between compression fange torsional braces and tension fange torsional braces
(curves c and d). 1 diaphrag placed at te centoid tat is 1/2 te depth of the bea to be braced will be almost
as efective as a braced fange with a fll dept stifener (compare curves e ad g). I general, fage connected
braces should extend at least 3/4 depth and should be attached to the braced flange. Connections for diaphrags
should stifen the web over at least 1/2 the depth for rolled beas. Te detail shown in Figure 2.10 shows a fll
depth web connection. Te stifeners or connection agles do not have to be welded to te fages when diaphrags
ae used ..
z.3.z Approximte Buckling Strength. Taylor ad Ojalvo (1973) give the following exact equation for
te critical moment of a doubly symmetic beam uder unifor moment with continuous torsional bracing
favorably with BASP solution in which the web
stifener is asumed to be 10 in. long or 11 in. long and
not attached to either fange. Te actual stifener was
an angle 11 inches long but it was bolted to te web
ad te distance between connectors was 9 in. Tere
fore, it is questionable that te total length of the
stifener is efective. Stifeners that do not contact the
braced flange are not recommended and tese tests
illustate the reduced stengt when compared with te
results in Figure 3.15.
3.6.4 Efect o/ Torsional Brace Location.
Teoretically, the attachment height of a torsional brace
should have 00 efect 00 the buckling load if te bea
m web does oot distort. Figues 3.18 and 3.19 show
values of critical load for tests with torsional bracing
placed on te compression flage, tensioo fange or
split evenly between te compression and tension
fanges (combined bracing). Figures 3.18 and 3.19
show that the combined bracing produced a slightly
higher critical load for beams with 00 stifener;
however, te beam wit a 4XI/4 in. stifener and a
brace stifess of 175 k-inlradia also showed an
increase i critical load. Based 00 tese tests, the brace
location . did oot sigifcatly afect the critical load
regardless of the cross-sectioo stifess of te test beam.
3.6.5 Forced Imerections. I te experi
ments, two types of imperfections were tested; natural
imperfections and forced imperfections. Since a natural
imperfection requires equilibrium of interal stesses
ad forced imprefection requires equilibrium wit an
applied exeral reaction, there is no teoretical basis
for assuming that both types of imperfection would
have the same impact on te efective brace stifess.
Based on Figure 3.20, a forced imperfection has an
c;
e

-
O

9
C
Z
:

O
:
m
8
7
6
5
4
3
2
1
Full4x1/4
Web
4x1/4x11
x \
1 4 x 114 stfener 1
Inlta Sweep ( In.)
o 0.2
D 0.16
x 0.0
0 20 40 600 80 100 120
BRCE STIFFNESS (pin/radian)
Figure 3.17 Web stifened tests.
7
Brac Loction
6
.CompFlg
C 5
l Ten Flg
a
;
4
DComblned
"
a
. 3
2
I 0.16" Imperecon I
l
O
175 666 Brc Stfnes (k-in I rd)
N
one
4"x1/4" Stfener Size
Figure 3.18 Efect of brace locatioo.
35
efect simila to a natural imperfection. Since lateral-torsional buckling involves both a twist and a lateral
displacement, the magnitude of the initial tist may also have an efect on the brace stifess.
3.7 Summar of Twin-Beam Tests
Te BASP progra ad bracing Eqs. 2.1 and 2.10 showed good corelation with te test results. Te tests
verifed tat cross sectioo distortion is very important for torsional bracing. Te W12X14 test beams could not reach
the load level coresponding to buckling betweeo braces unless web stifeners in contact wit te braced fange were
36
used. Torsional bracing at the bottom tension flage
was just as efective as bracing on the compression.
Lateral bracing at te top flange of simply supported
beams wa very efective and was not afected by cross
section distortion (stifeners are not necessa).
Te Meck technique for obtaining exerimen
tal buckling loads worked very well for lateral bracing,
but was not usefl for torsional bracing because of te
efect of local cross section distorion. I te torsional
bracing experiments, the results were sensitive to the
sligt lateral restaint provided by the test fe,
especially for very staight beams. Combined lateral
and torsional bracing is very efective in contolling
lateral buckling of slender beas.
Te initial sweep of the beam did not afect
the buckling load of laterally braced beams, but did
influence the results in te torsional buckling experi
ments. Te experimental buckling were al larger tan
te BASP prediction regadless of the level of initia
imperfection, so the apparent out-of-staightess efect
was related to lateral fe restaint.
7
6
.5
r
o.
:
-4
"
r
o
- 3
m
o
: 2
(
1
7
6
a
;s
-
.
4
3
::
.:
o 2
o
10.2" Imperecon/
175
4"x1/4"
175
None
Brac Loton
Ten Flg
CompFl g
o Combined
666 Brac Stf(k-in /rad)
None Stfener Slze
Figure 3.19 Brace location efects.
F Imperfection
0.12" Forced
0.22" Natural
O 0.31" Forced
175 k-in/rad Brace - 4" x 1/4" Stifener
Figure 3.20 Forced initial imperfections.
59
connection details are used to contol distortion, then te stifener size can be detenined fom Eq. (A9). For patial
depth stifening illustrated in Figure 5.11, the stifess of the various sections of the web ca be evaluated sepaately,
ten combined as follows:
(5.21)
where = hc' h5' or and
1 1 1 1 1 1
-=-+-+-+-+-
/
T
/
c
/
5
/t /
b
/g
(5.22)
Even fther refnement could be made by considering the portion of the web within hb to be infnitely stif. For
rolled sections, if the connection extends over at least one-halfte bea dept, ten cross section distortion will not
be important as demonstated in Figure 2.18 ad
/
b

/
T' Te location of the diaphrag or cross fame is not very
important; it does not have to be located close to te compression flange. Te depth of te diaphragm, h5, ca be
less tha one-half the girder depth as long as it provides the necessay stifess to reach te required moment by
Eqs. (5.19) or (5.20). For cross fames,
/
5
, should be taken as infmity; only and hc will afect distorion. Te
distortion efect will be considered in the desig examples.
5.2.3 Example Poblem I Exaple 3, a diaphrag torsional bracing system is desiged to stabilize
the fve steel girders during constuction as described in Examples 1 and 2 for lateral bracing. Te stength criterion,
Eq. Al l , is initially assumed to contol the size ofte diaphag. A C10 x 15.3 is sufcient to brace te girders.
Both yielding and buckling of te diaphag are checked. Te stifess of the ClO x 15.3 section, 195,500 in
k/radian, is much greater tan required but the connection to the web of the girder and the in-plane girder fexibility
also afect te stifess. I tis example, te in-plane girder stifess is very lage ad its afect on the brace system
stifess is only 2%. I most practical desigs, except for twin girders, this efect can be igored. If a fll dept
connection stifener is used, a 3/8 x 3-1/2 in. section is required. Te weld desig, which is not shown, between
the chanel and the stifener must tansmit the bracing moment of 293 in-k. Ifte more exact Eq. (5.19) would have
been used, the
/
T required would reduce to 14,300, a 22% reduction. Te diference between te simple Eq. (AlOa)
and Eq. (5.19) is caused mainly by the top flage factor of 1.2 used in Eq. (AlOa).
Te 40-in. deep cross fame design in Example 4 required a brace force of7.13 kips fom Eq. (Al l ). Te
factored girder moment of 1211 k-f. gives an approximate compression force in the girder of 1211 x 12/49 = 296
kips. Tus, te brace force is 2.5% of the equivalent girder force. Tus, common rule-of-thumb brace forces of
2% of the girder force would be unconservative i this case. Te faming details provide sufcient stifess. Te
3-in. unstifened web at the top and bottom flanges was small enough to keep
/
s
c
well aboye the required value.
For illustration purposes, a 30-in. deep cross fame attached nea te compression fage is also considered.
I tis case, te cross fae itself provides a large stifess, but the 14-in. unstifened web is too flexible. Cross
section distortion reduces the system stifess to 16,900 in.-k/radia, which is less tan te required value. If this
62
same cross fame was placed at te girder midheight, the two 7-in. unstifened web zones top and bottom would be
stif enough to satisf the brace requirements. For a fxed depth of cross fame, attachment at the mid-depth provides
more efective brace stifess tan attachment close to either flange as illustrated in Figre 2.18.
5.3 Bridge Deck Evaluation
As discussed in Section 1.5 ad Chapter 4, a bridge deck has te potential to act as a lateral and/or torsional
braceo Torsional bracing efects should only be considered when there is a positive attachment to te deck which
provides a moment connection between the /deck ad the supporting stingers. Wen there is no positive attachment
between te deck ad te stngers, torsional bracing by the deck is not reliable since the restaint would reduce to
zero when fange contact is lost as the beam twists. Te tests reported in Chapter 4 show that even when the deck
is not attached to te stingers, the deck can still fnction as a lateral brace at the wheel point due to fction at te
contact surface. For tese reasons, only lateral bracing by the bridge deck is discussed in this section. Te
application of the lateral bracing equations for short spa bridges must be based on te assumption that sufcient
fiction or shear connection exists to tasfer lateral loads associated wit stringer buckling to the bridge deck. Te
deck ca ten fnction as a lateral brace and/ or distibute lateral loads among te stringers, such that the most heavi1y
loaded stringers are braced by those with lighter
loading as shown in Figure 1.3. Te lateral stifess of
a short-span bridge should be based on the stifess of
the total system, to include contributions fom te deck
and all of te stngers.
F or the desig or rating of short span bridges
wit roUed sections, te goveming loading condition
for fexure is a wheel load near midspan, as shown in
Figure 5.12. To deterine the lateral bracing efect of
the deck at rdspa, L' for tis loading condition, the
Figure 5.12
general bracing equations can be simplifed to te
foUowing for the Load Factor Design method in the
AASHTO Specifcation (in-lb units):
Wheel
Load .
Deck Braclng
1
L
I
Loading condition for short-spa
bridges.
Mc = ( M; + 0.766 P
2
A )( 1 + A )

Ms

(5.23)
where
CHTER 6
CONCLUSIONS A RECOMNDATIONS
A efective brace must prevent twist of the cross section. Two types of bracing systems have been
considered, top fange lateral bracing systems ad torsional bracing systems. Tese systems must satisf bofu
stifess and stength criteria and a sum ary of fue requirements for each system is given in Appendix A. Te
analytical and experimental studies showed fuat fue bracing requirements for beams are related to fe shape of fue
moment diagas, fue maximum moment, and fe location of fe load on fe cross section (top flange loading is
fue most critical), ad fe desig mefuods developed consider all of fese factors.
For torsional bracing systems such as cross fames and diaphrags, cross section distortion is a very
importat paameter fat must be considered when evaluating or desiging fue bracing system (Eqs. 5.21 ad 5.22
or Eq. A9). Tese forulas can be used to desig adequate stifeners and connection details to contol distortion.
For rolled sections, diaphrag connections should extend at least tree-quaers of the depf. For plate girders which
have fuin webs, fue connection of fue diaphrag or cross fames should be as close to each flage as practica!.
Desig Examples 3 ad 4 illustate fue sensitivity of fue brace stifess to fue connection details. For most torsional
brace desigs, fe stengt criterion will contol the size of fue bracing members and fue stifess criterion will
contol fe size of fue stifeners and fue connection details.
Te test on a fll-size short-span bridge wif a wooden deck which was not attached to fue steel stingers
showed fe fiction at fe wheel location was sufcient to mobilize fue lateral stifeners of fue deck for bracing fe
beams. Te bridge capacity was doubled by the bracing efect of fue deck. Concrete decks ae very stif so fue
wheel location can be considered a brace point in such bridges. Wooden deck stifess ca vay considerably,
depending on fe constuction details. Wooden decks examined in fue cental Texas area had sufcient stifess to
prevent beam buckling at fe wheel location ad fue bridges could be rated based on feir yield stength. For rating
purposes;te lateral stifess ofte deck ca be obtained by considering only the lateral stifess of fe deck nailers.
The studies herein have also verifed fe new lateral buckling equation in fe 1990 AASHTO Bridge
Specifcation. Bofu experiments and feory show fuat the new formula can be used wif confdence. Te old
forula for lateral buckling gave overly conservative capacities for lage unbraced lengts. Te new formula should
be used in fe rating of bridges. A sample rating of an existing bridge (Bridge AA0539-001, Village Creek - Ellis
County, Texas) is given in Appendix C. Tat bridge has a current inventory rating of H3.2; the new rating based
on fue 1990 AASHTO Specifcation would be H6.2. Increased rating of short-span bridges should be expected if
lateral buckling contolled the cur ent rating. If fe bridge is not adequately braced at fue wheel location to force
either yielding or buckling between existing braces (like fue fll-size bridge tested), fen te buckling capacity should
be based on the sum of the lateral buckling capacities of all fue girders if positive vertical contact between fe deck
ad fue girders ca be assumed at te wheel location. In wide bridges it is probably more realistic to sum fue
buckling capacities of the girders within fue lane wif plus one or two girders on eifuer side of lane to account for
possible uplif of fue planks. A example of this lateral buckling concept is given in part 2 of Appendix C. Friction
at fe wheel location will force all fue girders to buckle simultaneously, so fue bridge capacity should not be based
on just fe most highly stessed sigle girder.
69