Straight-Element Grid Analysis of Horizontally

Curved Beam Systems

HERBERT A. WEISSMAN

HORIZONTALLY CURVED beam systems undoubtedly find short-duration loads, three different analyses must
their greatest applicability today in the framing of normally be employed:
curved highway bridges. Although this paper will utilize
1. Dead load of steel and concrete 77 = 00
in its discussions the specific structural components of a
2. Superimposed dead load 377 = 30
steel /-shaped beam system with a concrete slab deck,
3. Live load 77 = 10
the basic principles are applicable to other horizontally
curved beam systems. THE PLANE GRID SYSTEM
M u c h research 1 has gone into the theoretical and
T h e horizontally curved highway bridge is not a true
practical aspects of curved beam analysis. A method will
grid system. Such a system can consist only of discretely
be presented here which will enable a designer to secure
connected beam elements (straight or curved) lying in a
sufficiently accurate results by using a method of analysis
plane and subject only to forces perpendicular to the
(more practically a computer method of analysis) with
plane a n d / o r moments acting in the plane.
which he is familiar, namely the statically indeterminate
T h e points where the beams intersect are referred to
analysis of plane grid systems with straight elements.
as joints. Because the elastic properties of the beam ele-
S T R E S S 2 is probably the most popular computer pro-
ments can be determined from their cross-sections, length
gram for the analysis of indeterminate structures, and
and curvature (if any), the elastic energy of the system
reference will be made to a computer program which
and the final state of stress can be found by dealing only
aids in the preparation of input data for STRESS.
with the forces and deflections at the joints. T h e beam
A single horizontally curved beam which is not
elements do not have to be prismatic as long as the
torsionally restrained at the ends is not stable u n d e r
manner of variation of the properties is known. T h e
vertical loads. I n curved bridge construction, diaphragms
beam properties involved are the shear area (Az), the
(steel b e a m or more commonly trussed members, normal
twisting moment of inertia (Ix), and the bending moment
to the main beams) provide the transverse load capa-
of inertia (Iy).
bility necessary for stability by resisting the torsion in
Thus, in accordance with the definition of a grid,
the main beams. Because the deck slab frames only into
there are three loads (and three associated deflections—
the top flange of the beams, it cannot effectively resist
degrees of freedom) at each joint. Each member has
torsional forces in the beam. However, it does provide
three load components acting on it. Figure 1 shows the
good lateral load distribution capability by tending to
equalize vertical deflections of adjacent beams.
I n normal bridge construction (curve or straight) ,i
the concrete slab is made to act compositely with the
main steel through the use of shear connectors. Because
the modular ratio, 77, between steel and concrete is as-
sumed to be different 3 for long-duration loads than for
^y

Fz ,
Herbert A. Weissman is the head of the Structural Department in the ^. ^_» ^x.
New York City Office of Goodkind & O'Dea, Consulting Engineers Mx.
and Planners.
Fig. 1. Load components in a grid system

41
A P R I L / 1970
 Acfua/ structure - curve the structure back into a grid. It should be noted that all
grid analyses—including those using curved elements—
must rationally account for these non-grid character-
istics.

EFFECT OF USING STRAIGHT ELEMENTS

In order to use conventional indeterminate analysis (and

conventional indeterminate analysis computer pro-
grams), it is necessary to replace curved sections of beams
Fig. 2. Curved bridge—Model "A" by straight lines. T h e number and arrangement of
straight line segments will affect the degree to which the
straight line structure simulates the curved structure.
three joint load components, where the X-Y plane is the T h e maximum spacing 3 of diaphragms in straight bridges
plane of grid. System coordinate axes will be indicated is normally in the order of 25 ft. Since diaphragms in
by capital letters. Member coordinate axes will be indi- curved bridges must be thought of as main members,
cated by lower-case letters with the member x-axis necessary for the stability of the structure, they are usually
coinciding with the longitudinal axis of the member and spaced more closely than on straight bridges. A logical
the member and system Z-axes being parallel. positioning of joints would appear to be at the inter-
Horizontally curved highway bridges deviate from a section of diaphragms and beams as shown in Fig. 2 and
true grid system in the following ways: designated as Model " A " (a simply supported single
span of three stringers is shown for simplicity). This
creates an angular discontinuity in the beam at its inter-
Diaphragms most often consist of trussed (axially section with the diaphragm.
loaded) members framing into the flanges of the
Consider first a portion of curved beam between
curved beams.
diaphragms as indicated in Fig. 3. Assume that the
T h e /-beam cross-section warps under torsional
system X-axis is parallel to the chord. Taking moments
loads, introducing additional degrees of freedom.
about the X-axis:
T h e concrete slab is a continuous member (not
discrete) and its elastic line is offset from that of the
beams. ^wz X -e + (MyL + MyR) sin - -

(MxL + MXR) cos - = 0 (1)

Each of the above deviations from true grid action will
require certain assumptions to be made in order to model
where e is the eccentricity of the load from the chord. For
short segments the angle 6 is small and cos 6/2 = 1,
sin 6/2 = 6/2. Equation (1) can be rewritten as

31wz X e + Myav. X 6 = 2 M , (2)

where

My„. = y2{M„VL + MVR)

and
VMX = MXL + MXR
Equation (2) indicates that between diaphragms the
torsional moment—necessary to resist the components
of the bending moment due to curvature (adjacent sec-
tions are not parallel) and the eccentricity due to curva-
ture—builds up as the distance S between sections in-
creases. T h e main beams are weak in torsion and the
diaphragms, which can resist these forces by bending, act
as elastic supports for the torsional loads acting along
Fig. 3. Curved beam segment the length of the beam.

42

AISC ENGINEERING JOURNAL

MXL •Mxe
AJC/L Mya
(a)

Fig. 4. Chord intersection point—without diaphragm

Consider now a portion of a curved beam which is

modelled by a series of chords. Figure 4 shows the forces
acting at an intersection point of the chords without any
d i a p h r a g m (vertical forces are not shown). Without loss
of generality, assume that the system X-axis bisects the
angle between the chords. Taking moments about the
(b)
X-axis:
f) f) Fig. 5. Straight element "building blocks": (a) Element /,
{MVL + MyB) sin - - (MXL + MxB) cos - = 0 (3)
(b) Element II

For small 6, Eq. (3) can be rewritten as:

a uniformly varying twisting moment on the straight
y2 (MVL + MVR) e = 2MX (4)
members will more closely simulate curved beam action.
where 2MX has the same definition as in Eq. (2). In most This sort of accuracy is not normally warranted.
of the areas of the bridge, the bending moment is much In Fig. 5b the torsional moment in the beam is in-
larger than the torsional moment, with the result that troduced as a concentrated effect at the angular dis-
MyL and MyR will be approximately equal. Equality of continuity. This torsion is a function of the longitudinal
the bending moments gives a good modeling to the bending moment at the discontinuity as indicated in
structure. T h e twisting moments will in general not be Eq. (4). In Eq. (2), it is a function of the average of the
equal on both sides of the joint since they depend on the bending moments at the ends. Still, there is a good cor-
rotations of adjacent joints. This produces a discontinuity relation between Eqs. (2) and (4).
which does not exist in the curved structure. Values of Figure 6a shows a diaphragm framing into a portion
the twisting moment at such a joint are not reliable and of curved beam. Consider the forces acting at the joint as
should not be used for design. shown in Fig. 6b (vertical forces are not shown). Here
Figure 5 shows two possible "building blocks" of
straight line segments which can be used to model a
curved beam. Each building block corresponds to the
arc length S used in Fig. 3 for a curved segment. T h e
difference between Figs. 5a and 5b is that in 5a the angu-
lar discontinuity occurs at the boundary of adjacent p—Diaphragm
segments; in 5b it occurs within the segment. Figure 5b
would, therefore, seem to be a better "building block".
(a)
This will become more evident when diaphragms are
introduced into the structure.
Compare Eq. (4) for Fig. 5b with Eq. (2) for the
curved segment. There is no term in Eq. (4) for the
eccentricity of the load acting directly on the segment. A/C/L Myo.
T h e twisting moment in each arm of Fig. 5b will be

r~!
MXJL
constant unless a uniform (more correctly a uniformly MxfZ
varying) twisting moment is introduced along the mem-
ber. It is interesting to note that if a segment of a curve Myo
is broken u p into five equal chords, the maximum offset (b)
of the curve from the chord is about 1/25 the offset of
the curve from a single chord joining the ends. If ten Fig. 6. Curved beam with diaphragm: (a) Plan, (b) forces at
chords are used, the offset ratio becomes 1/100. Placing joint

43
A P R I L / 1970
 Center of Curvature Acfua/ structure -<

\M/D MUD Wyo \Myo\

)( )( )(
I! II
V

Fig. 7. Cross-section of curved beam system -Mode/ -chore/

the bending moment components are parallel and, as Fig. 9. Curved Bridge—Model "£"
was previously noted, the diaphragm acts as an elastic
support for the torsional stresses in the beam. Calling the
bending m o m e n t in the diaphragm MyDi and taking
moments about the X-axis: loads which is common to curved beam systems (see
Fig. 7).
MxL + MxR - MyD = 0 (5)
Figure 8 shows the intersection of diaphragms with a
It is the beading moment in the diaphragms that stabi- curved beam modeled by the two "building blocks" and
lizes the individual curved beam. To keep the diaphragms free body diaphragms at the diaphragm intersections,
in equilibrium the curved beams must supply upward Figs. 8c and 8d (vertical forces again are not shown).
and downward reactions to the diaphragm ends. It is Taking moments about the X-axis in Fig. 8c and using
these reactions which cause the vertical redistribution of the same angle relationships as before:

^ W i + MvR)6 + MvD = 2MX (6)

In Fig. 8d, moments about the X-axis gives':
XMZ = MyD (7)

Eq. (7) is identical to Eq. (5). Equation (4) remains

(o) satisfied for Fig. 8b. For Fig. 8a neither Eq. (4) nor Eq.
(5) is satisfied. Clearly the use of Fig. 8b will more ac-
curately model the behavior of curved beams than will
the use of Fig. 8a.
Using Fig. 8b the three beam curved bridge structure
is shown in Fig. 9 and is designated as Model " B " . There
are about twice as many joints in Model " B " as there are
in Model " A " . T h e chord lengths are about the same.
(b)
If intermediate points are added on the curve between
those of Model " A " , a framing with the same number of
points as Model " B " will result. This will be called Model
MX 12 " C " and is shown in Fig. 10. T h e chord lengths here are
A/XL half those of Model " B " ; the diaphragms still are not
perpendicular to the beam as in Model U B " . A computer
program will compare Models "B""and " C " with a more
rigorous analysis 4 made for a two span continuous struc-
• MjC/3
ture with two beams.

THE NON-GRID ELEMENTS

ML/L. Mye
Diaphragms—In order to incorporate a trussed diaphragm
MyD
(such as the X shown in Fig. 11a) into the grid system,
(d) equivalent beam properties must be found for it. These
can then be used in the computer program in the form oi
Fig. 8. Straight-element model of curved beam with dia- a member stiffness matrix. 2
phragm: (a) and (c) Using element I, (b) and (d) using Figure l i b shows the position of the members anc
element II the forces acting after the left end (assumed to be framing

44

AISC ENGINEERING JOURNAL

into a non-deforming beam) is given a rotation 0. Let dctvo/ sfrc/cfc/re - curve no/ shown
Ac be the area of the top and bottom members, Ad the
area of the diagonals, and E the modulus of elasticity.
T h e forces in the members are:

ACE6
AcEdd
P =
2h
h d
AdE -0 -
I 2 AdE6hd
Q = Fig. 10. Curved Bridge—Model " C "
I 2/2

/z AcE6d AdE6h2d
HL = P + ~ Q = — h —
analysis, from which only an accuracy necessary for
design is required. However, if lower and upper bounds
Mr = Hrd
for the torsional stiffness could be obtained, two grid
analyses could be made and design values could be ob-
HR Q tained by comparing and evaluating the results. While

MR = HRd

I ^ /3

T h e values for a deflection A at the left end (Fig. l i e ) are:

AdE- A
R = I _ AdEAd
1 V~
fo)
h AdEAdh
HR = R =
7 ~F~
ML = MR = HLd mr P
\
o d' 2AdEAd>
Hi.
\S<^. He
V
\ ^ /
ML (
Note t h a t the shear for a unit rotation is: ) Me.
V _ AdEhds
/-/L He
G ~ Is
T h e m o m e n t for a unit deflection is the same: \v (b) ' V

ML _ AdEhdz
~A ~ P~ <v
HL He
F r o m these values the member stiffness matrix can be
formulated. Other types of trussed diaphragms can be
ML (
solved in a similar manner. ^^ >

Warping of Cross-section—The torsional stiffness of an HL

\s'S\
,4% s
^ s

He
/-shaped section is greatly influenced by the amount (if X A I
Vv
any) of resistance to warping caused by either the sup- \v t (c)
ports or the nature of the loading. Any theoretical
solution 5 to this problem is quite complex. It would cer- Fig. 11. X-diaphragm: (a) Dimensions, (b) end rotation,
tainly n o t be warranted in a straight-element grid (c) end displacement

45
A P R I L / 1970
 "center of gravity" of the two values o f / H - T h e twisting
moment of inertia can then be figured in a similar
manner.
If there is no restraint against warping, the resulting
torsion is referred to as Saint Venant torsion 6 and the
T-Pd
twisting moment of inertia (for an /-section) is ap-
proximately

Co) h = E \ b? (9)

where t is the thickness of plate and b is the width as

shown in Fig. 12c.
T h e Ix based on Saint Venant torsion is a good lower
bound and the sum of this value and (the m u c h larger)
value given by Eq. (8) is a good upper bound for use in
the grid analyses.

Concrete Slab—The action of the concrete slab poses two

problems in trying to fit it into a grid: its continuity and
the fact that it is attached only to the top flange of the
main beams. If no temporary supports for the beams are
used, the slab is effective only for the super-imposed dead
load and the live load. Thus its action is only to be
incorporated for the analyses at 3^ = 30 and 77 = 10.
Since it frames only into the top flange of the beam
Fig. 12. Torsion in a restrained I-shaped beam: (a) loads, the slab will not offer the transverse resistance to twisting
(b) deflections, (c) cross-section of the beam which the diaphragms do, but it will in-
crease the torsional moment of inertia, Ix, of individual
beams. An effective slab width, similar to that used in
not as desirable as an exact approach (if one could be figuring composite bending properties, could be used in
obtained) such procedures—utilizing upper and lower figuring composite torsional properties. T h e slab trans-
bounds—are not uncommon in engineering design. verse bending strength must be represented by a beam
Because even the upper bound of torsional stiffness is so element in order to fit into the grid analysis. T h e stiffness
m u c h smaller than the bending stiffness of the dia- to resist differential deflections of adjacent beams must
phragms and the main beams, the two analyses will give involve a stiffness to resist the bending rotation of the
similar results. 4 transverse members. T h e rotation of transverse members
Torsion in an /-section with fixed ends (fully re- is the same degree of freedom as the twisting of the main
strained against torsion) can be resisted by bending of beams. Thus a transverse member with bending stiffness
the flanges. 6 Figure 12a shows an /-section resisting a must resist torsion in the main beams. Therefore, if the
torsional load T = Pd. Figure 12b shows the horizontal beam element is assumed to frame into the main beams
deflection of the flanges assuming that there is no hori- between diaphragms, it will offer a resistance to the
zontal rotation at the ends. T h e deflection is twisting of the main beams which actually does not exist.
By placing the beam representing the slab at the dia-
PL*
8 = phragms results in there being no transverse resistance
12£/H
to torsion between diaphragms. It will, however, in-
where In is the horizontal moment of inertia of the flanges crease the resistance over that which actually exists.
(assumed equal in this example). T h e angle of rotation is Since the bending stiffness of the diaphragms alone is so
a = 8/(d/2) — 28/d. T h e twisting moment of inertia is much larger than the twisting stiffness of the main beams,
the resulting force distribution will not be greatly in
TL Pd2L 6EIHd2
/* = Ga (8) error.
2G8 GL2
T h e question arises as to what value of Iy and Ix
where G is the shear modulus of elasticity. If the flanges should be assigned to a concrete slab about 8 in. thick
have unequal / H , the center of rotation will be at the and about 20 ft wide, where the transverse spacing of

46

AISC ENGINEERING JOURNAL

beams is about 8 ft. Experimental evidence* exists con-
cerning the action of a concrete slab on straight stringers.
If such a structure were analyzed as a grid using a dia-
p h r a g m spacing of about 20 ft, an effective transverse
slab width in bending could be approximated. Using
one-half the effective width of the slab will give good re-
sults. T h e torsion moment of inertia should be taken
equal 7 to the bending moment of inertia. T h e properties
of the steel diaphragm and the concrete slab would be
added together (taking into consideration the different
moduli of elasticity) even though their elastic lines are
offset.
Slab continuity also poses a problem regarding the Fig. 13. Two beam, two span continuous curved structure
distribution of the applied loads. Most of the loads are
not applied directly over the modeled grid structure. Table 1. Loading 1
Factors used in the design 3 of the slab could be used in (1 kip/foot uniform load on each girder)
determining the distribution of loads from the slab to the Outside Girder Forces
grid elements. These factors involve an effective width
Analysis
for concentrated loads and continuity in the slab under
certain conditions. Point Force Model B Model G Rigorous

0.0L Vc
COMPUTER PROGRAM FOR STRESS INPUT
vt 31.2 31.2 31.5
Because there are a relatively large number of joints M .0 .0 .0
and members in the straight-element grid analysis of T0
9.4 6.4 7.7
curved beams, a large part of the input deck for S T R E S S
will consist of data describing the joint coordinates and
the member incidences. 2 Calculating the joint coordi- 0.1L v0 15.3 15.4 15.4
nates is a big task; coding them for key punching and vt 14.2 14.2 14.3
M 369. 371. 375.
punching the cards makes it even bigger. However, a
T0 -3.6 -0.1 -5.0
computer program can compute all the joint coordinates 14.8 8.1 14.5
Ti
and the member incidences and can punch the results
out on cards for use directly in the S T R E S S program.
T h e author has written such a program for radial sup-
0.2L v0 -1.7 -1.7 -1.7
Vi -3.1 -3.1 -3.2
ports. M 468. 470. 476.
I n p u t for the program consists of the number of -12.3 -5.5 -12.7
supports and their stations (along a line of given radius), 11.2 5.0 11.9
the number of stringers and their offsets from the sta-
tion line, the number of diaphragm spaces in each span,
0.3L v0 -19.0 -19.0 -19.2
and the number of intermediate points between dia- -19.8 -19.9 -20.1
Vi
phragms. Joints are numbered consecutively along the M 292. 294. 297.
stringers and then along the diaphragms (on radial T0 -13.7 -7.5 -12.8
lines). T h e program places an S after support joints as Ti 0.1 -1.5 1.8
required by STRESS. Joint releases must be prescribed
at supports to account for actual conditions of articula- 0.4L v0 -35.7 -35.8 -36.1
tion. vt -35.1 -35.3 -35.5
M -149. -149. -152.
T0 -6.1 -4.6 -i.9
COMPARISON OF RESULTS AND CONCLUSIONS Ti -15.7 -9.4 -13.4

Figure 13 shows the two beam, two span continuous

structure which will be analyzed using Models " B " and 0.5L v0 -51.0 -51.1 -51.5
Vi -51.0 -51.1 51.5
M -834. -836. -848.
T0 12.0 4.5 15.7
* Reference 3 lists some studies used to determine lateral distribution of Tt -12.0 -4.5 -15.7
wheel loads.

47

APRI L / 1970
 Table 2. Loading 1 Table 3. Loading 2
(1 kip/foot uniform load on each girder) (1 kip/foot uniform load on outside girder only)
Inside Girder Forces Outside Girder Forces

Analysis Analysis

Point Force Model B Model G Rigorous Point Force Model B Model G Rigorous

0.0L v0 0.0L V0
Vt 24.9 24.9 25.0 Vt 30.0 30.1 30.2
M .0 .0 .0 M .0 .0 .0
T0 T0
7.5 5.3 6.3 Tt 13.0 10.1 11.7

0.1L va 10.7 10.7 10.7 0.1L V0 14.1 14.2 14.2

Vi 11.8 11.8 11.9 Vt 13.7 13.7 13.8
M 254. 252. 256. M 351. 351. 356.
T0 -1.6 0.7 -2.5 T0 0.6 3.9 -0.4
Tt 10.8 6.0 10.4 Tt 16.0 9.5 15.7

0.2L v0 -2.4 -2.4 -2.4 0.2L V0 -2.2 -2.2 -2.2

-1.0 -1.0 -1.0 Vi -2.7 -2.7 -2.7
M 321. 319. 324. M 443. 444. 449.
T0 -8.0 -3.4 -8.2 T -9.9 -3.4 -10.0
X
0
Tt 7.7 3.4 8.1 9.3 3.4 9.9
T
0.3L v0 -15.2 -15.1 -15.3 0.3L v0 -18.6 -18.6 -18.7
vt -14.4 -14.2 -14.4 -18.8 -18.8 -19.0
M 206. 205. 208. M 274. 275. 278.
T0 -9.7 -5.3 -9.0 -14.3 -8.4 -13.5
Tt -0.1 -1.3 1.0 -3.5 -4.9 -2.1

0.4L V0 -28.5 -28.4 -28.7 0.4L v0 -34.7 -34.7 -35.0

Vt -29.1 -29.0 -29.2 Vt -34.4 -34.5 -34.7
M -99. -98. -100. M -151. -151. -154.
T0 -4.9 -3.7 -3.3 T0 -9.1 -7.7 -7.1
Tt -11.5 -6.8 -9.6 T, -17.6 -11.6 -15.7

0.5L V0 -43.3 -43.2 -43.6 0.5L V, -50.3 -50.3 -50.7

Vt 43.3 43.2 43.6 V, 50.3 50.3 50.7
M -614. -611. -621. M -825. -825. -837.
T0 8.4 3.1 11.2 T0 9.8 2.1 13.1
Tt -8.4 -3.1 -11.2 Tt -9.8 -2.1 -13.1

U
C " and then compared with a more rigorous curved loadings 1 and 2, outside girder and inside girder. V0
girder analysis.* For the comparison in Model U B , " points and V^ T0 and 7\-, are the shear and twisting m o m e n t
on and outside the curve were used (they could also at a section at the outside (away from middle support)
have been on and inside the curve or both outside a n d and inside (toward middle support). M is the bending
inside). For the beams Ix = 594 in. 4 , Iy = 12,290 in. 4 ; m o m e n t at the section. Since the results are symmetrical
for the diaphragms Ix = 2.3 in. 4 , Iy = 1500 in. 4 Shear de- with respect to the middle support, only one-half the
formations were not included. Loading 1 consists of a structure is shown.
uniform load of 1 kip/ft on both girders; loading 2 con- T h e results for shear and bending moment for Models
sists of a load of 1 kip/ft on the outside girder, no load on " B " and " C " are very close to those of the rigorous analy-
the inside girder. Tables 1 through 4 show the results for sis. T h e torsional values for both models are not in as close
an agreement with the rigorous analysis as are the values
for bending moment and shear. However, the values for
Model " B , " using Fig. 9, are in substantially better agree-
The values (and the sign convention) for the vigorous analysis are
taken from the computations made by Richardson, Gordon and ment than those of Model " C , " using Fig. 10. Although it
Associates; these computations are the basis for the results shown in is risky to generalize from one example (one that uses a
Reference 4. symmetrical structure with uniform, symmetrical load-

AISC ENGINEERING JOURNAL

 Table 4. goading 2 (1 kip/foot uniform load on outside girder only) Inside Girder Forces

Analysis Analysis

Point Force Model B Model G Rigorous Point Force Model B Model C Rigorous

0.0L v0 — — .— 0.3L v. .0 .0 .0
-0.9 -1.0 -0.9 0.2 0.3 0.3
Vi
M .0 .0 .0
v
Mt -19. -20. -20.
T0 • — • —
— T
x
-0.9 -1.1 -1.0
0
T{ 6.3 6.1 6.7 Tt -4.6 -4.2 -4.9

0.1L v„ -0.9 -1.0 -0.9 0.4L v0 0.2 0.3 0.3

v -0.5 -0.5 -0.5 Vi .0 .0 .0
Mt -12. -13. -13. M -16. -16. -17.
T0 6.7 6.3 7.1 T0 -3.6 -3.6 -3.7
Tt 2.6 2.7 2.8 Tt -3.1 -2.4 -3.0

0.2L V, -0.5 -0.5 -0.5 0.5L v0 .0 .0 .0

Vt .0 .0 .0 Vt .0 .0 .0
M -19. -20. -20. M -17. -16. -17.
T0 3.5 3.2 3.8 T0 2.1 -1.9 -1.9
T{ -2.1 -1.7 -2.2 Tt 2.1 1.9 1.9

ing), it is felt that Model U B' 5 can be used to adequately 3. Standard Specifications For Highway Bridges, American
simulate a horizontally curved beam system. Association of State Highway Officials, 1969.
4. Highway Structures Design Handbook, United States Steel
Corporation, 1965.
REFERENCES 5. McManus, P. F., and Culver, C. G. Nonuniform Torsion of
1. McManus, P. F., Nasir, G. A., and Culver, C. G. Horizontally Composite Beams, Proc. Am. Soc. Civil Engs., June, 1969.
Curved Girders—State of the Art, Proc. Am. Soc. Civil Engs., 6. Timoshenko, S. Strength of Materials, D. Van Nostrand Co.
May, 1969. Inc., 3rd Edition.
2. Fenves, S. J., Logcher, R. D., Mauch, S. P., and Reinschmidt, 7. Yettram, A. L., and Husain, H. M. The Representation of a
K. F. STRESS: A User's Manual, M.I.T. Press, Cambridge, Plate in Flexure by a Grid of Orthogonally Connected
Mass. 1964. Beams, Int. J. Mech. Set., Vol. 7, 1965.

MANUAL OF STEEL CONSTRUCTION—7th Edition

Preparation of the 7th Edition of the AISC M a n u a l of A brochure describing the new Manual, including an
Steel Construction is nearly complete, and copies are order form, will be mailed to readers of the Engineering
expected to be available in July, 1970. Journal within the next few weeks. We suggest you com-
T h e new edition is being extensively revised and plete the order form promptly to assure delivery of your
expanded to keep pace with the many new developments copy as early as possible after publication.
in steel construction since the 6th Edition was published
in 1963.

49

A P R I L / 1970
 Errata
April, 1970 (Vol. 7, No. 2)

Page 45
Page 44 Lines 11, 18, 20: T h e last term in the equations should
Fig. 7: T h e direction of moment MyD immediately to be d2 instead of d3.
the left of the center stringer should be clockwise instead V V
of counter-clockwise. Line 18: — should read —.
G e
Line 5, column 2: "Free body diaphragms" should Fig. 1 1 : T h e direction of shear V at the lower left hand
read "free body diagrams". corner should be u p instead of down.

104

AISC ENGINEERTNG JOURNAL