CURVED I-GIRDER DESIGN

Conrad P. Heins, Jr., Civil Engineering Department, University of Maryland

Design data that permit rapid determination of the forces and stresses in
curved bridges subjected to dead and live loads are presented. The data
were developed during a recent comprehensive study on curved bridges.
The data are applied in an examplE! design.

•DURING the past 5 years there has been a trend toward designing and constructing
horizontally curved highway bridges. This has created a need for methods by which
such systems can be conveniently designed. Several methods have been developed to
analyze curved bridge systems (!_, ~.; ~.. 2), but they are not design aids. Design aids
have been obtained by using 2 analytical methods tbat accurately represent the response
of the structural system under dead load (4) and live load (3, 5). The results are in de-
sign equations and graphs, which permits aesign of a curved composite girder struc-
ture. Thus the engineer can determine girder properties and diaphragm spacings
before a trial-and-error analysis.
Effects of impact will be included in the presentation of the design data.
The notations used in the formulations in this paper are as follow s :

Bi = induced bimoment (lateral flange bending effect),

b1 = flange width of steel girder,
C = distance from bending neutral axis to any fiber,
C1 = bending stress correction factor = S/S,
C2 =warping stress correction factor = (Wn/IJ/(Wn/I.),
Ee = modulus of elasticity of composite,
E. = modulus of elasticity of steel,
f = function,
F = maximum function,
Ge =modulus of rigidity of composite,
G. "' modulus of rigidity of steel,
h = centerline distance between flanges of a steel I-girder,
IF = flexural impact factor,
Ir = torsional impact factor,
Iw = warping torsional moment of inertia,
I. = primary moment of inertia of girders (actual stiffness),
K = design factor,
K1 = amplification factor,
& = distribution factor,
K:i = reduction factor,
Kr = Saint Venant's torsional rigidity,
L = span length,
m = GJG.,
M = induced bending moment,
Mb =maximum forces in 4-curved-girder system,
M. = internal bending moment,
n = E0 /E 1 ,

Publication of this paper sponsored by Committee on General Structures.

4
 5
R = radius of curvature of curved girder,
S = section modulus spacing or radial diaphragm spacing,
S =wheel load factor,
w =uniform dead load applied along the span of each girder,
~ = b, · h/4 = warping statical moment of steel girder section,
w0 1i. = referenced value,
w. = composite warping statical moment,
w..: =steel warping statical moment,
ab = normal stress (actual or corrected) in longitudinal direction of girder due
to bending,
ab = bending stress from referenced p1·operties,
aw = normal stress (actual or corrected) in longitudinal direction of girder due
to warping, and
a. = warping stress from referenced properties.

GENERAL DESIGN DATA

To analyze curved girder systems, one must know the girder section properties 1., Kr,
and I,. The designer, therefore, must estimate the size of the girders to compute
these stiffnesses. This size estimation can be achieved if the primary internal girder
forces M, and Bi, which induce normal stresses, are known. Then the basic bending
equation,

(1)

and warping equation,

(2)

could be applied by assuming a proportion of the design stress for ab and a, and com-
puting the i·equired (I/C) and (I/w..) properties. Estimation of these stresses and
forces has been obtained by performing a thorough system analysis of single-span
and multi span curved bridges (i, ~).

Impact Factors

The dynamic response of single-span curved girder bridges subjected to a sprung mass
vehicle was predicted by a Fourier series and lump-mass techniques (8). These tecb-
niques then were applied in determining the response of typical highway bridges. This
evaluated impact facto1·s. Partial results of these studies are shown in Figures 1 and
2 for plate girder bridges. Curves are given for various girder span lengths and cen-
tral angles, e = L/ R. The curves are for a vehicle speed of 60 mph (96 km/h); other
curves are available for velocities of 20 and 40 mph (32 and 64 km/h) (8).
Impact factors are applied as in conventional practice: -

(3)

(4)
6

Figure 1. Flexural impact factor .

• 34

.Jo

Note: 1 ft= 0.305 m.

90. lW. la. .1 o. lo •

SPJ\11 LEJIGTH - Fr•

1
Figure 2. Torsional impact factor .

• so

. 70

~ .60

.30

Note: 1 ft= 0.305 m

SPAJI Ll.'llGTH - Fr•

 7

Live Load Design

Design of any bridge element requires the establishment of forces in that element. If
the bridge has a straight alignment, the forces can be determined by distribution fac-
tors and simple beam theory. However, when the structure is curved, the interaction
of the bending and torsional forces creates an indeterminate situation. The following
equations therefo1·e were developed by using relationships between single curved re-
sponses and the system, to permit evaluation of the live load forces developed in a
curved composite-bridge I-girder system (3, 5, 6, 7). These forces then can be used
to establish the induced stresses and proper ilraer section.

Amplification Factor, K1

All the internal forces and deformations for a single curved girder and a single straight
girder have been evaluated by using various computer programs (6, 9). The ratio of
the reactions for these 2 girders gives the following: - -

K _ f(single curved girder

{5)
i - f single straight girder

This factor describes the immediate effect of curvature relative to a straight mem-
ber. A graphical representation and analysis of these data give the following general
equations:

Kmoment = o; 5
(L/R) + 1 (6)

2
Kbimoment = [ (35n)(L/R) - 15(L/R)J X 10 3 (7)

where n = R/100 [R > 100 ft (30.5 m)J.

Distribution Factor, &

The evaluation of true distribution of load to each girder, and realistic \lalues of inter-
nal forces, can be considered by analyzing the curved girder as a system. The num-
ber of trucks used in the analysis would depend on the number of lanes. The ratio of
these resulting maximum forces to those in a single curved girder gives

& _ f(system of curved girders (8)

- f single curved girder

A plot of this ratio versus Rand L will yield the following general equations:

Kmoment = (n + 3) (
0
·:L) + 0.6 (9)
8

K..
--uunomcru
~ Q_~ll{R/T.\
M ,_ .. ,
-.1 - - ·- . ,.. .
T.; '7n ff (?1 'I,._\
,., ,
~ ...., ......... (10)

(M - 1)
Kblmoment = 6 (R/L) L > 70 ft (21.3 m) (11)

where

n = R/100 [R > 100 ft (30.5 m)J, and

M = L/50 [L > 50 ft (15.25 m)].

These equations are valid for girder systems that contain 4, 6, and 8 girders and
girder spacings of 7, 8, 9, and 10 ft (2.14, 2.44, 2.75, and 3.05 m).

Reduction Factor, K:i

Because many bridge structures are continuous, it is desirable to obtain some factors
that can be applied to the simple span data to give preliminary forces in continuous
spans. This factor can be written as

~ _ f(system of curved girders)N (12)

- f{system oi curved girders)

where N =number of spans.

Using a compute1· program, we evaluated the maximum forces in a 2- or 3-span
curved bridge system of 4, 6, and 8 girde1·s wider various critical loadings. A study
of all the data and the resulting Ki values gives the values in Table 1. The data are de-
scribed relative to number of spans and a.re independent of number of girders. It
s hould be emphasized that the 2- and 3-span girder systems must contain equal span
lengths with a maximum given span length of 100.0 ft (30.5 m). For ex.ample, for a 3-
span system, t.Jie total maximum bridge length Wuuld then be 3 x L or 300.0 it (91.5 m).

EVALUATION OF GIRDER FORCES AND DEFORMATION

With the va1·ious factor equations available, one can evaluate preliminary forces in a
curved girder bridge , relative to the forces in a straight girder. (All Rs and Ls refer
to the centerline of the bridge system.)

1. Evaluate F for a single straight girder of length L that has been subjected to a
line of AASHO wheel loads . This iunctio11 would be Fbending• The function for Fblmoment
is assumed to be equal to 1.
2. Evaluate Ki equations 6 and 7 for L and R of the bridge system.
3. Evaluate I(i equations 9, 10, and 11 for L, number of girders in system, and R.
4. Select a reduction factor from Table 1 if system is a continuous span.
5. Determine F of curved girder system, that is

Maximum momentstutic = Mstraight X (K1 >< Kz >< Kz) (13)

and
 9

Table 1. Reduction factors for maximum bridge function.

Number
of Bending Saint Warping
Spans Moment Deflection Rotation Venant Torsion Bi moment Shear

0.75 0.70 0.70 0.75 0.45 0.35 1.00

0.65 0.60 0.70 0.75 0. 40 0.35 0.90

Figure 3. Section properties versus span length.

-3
xlO

2,

..... ,;
~ 1,

0
50, 70, 90. no. uo. 150,

,10

,08

,o6

j"~-j vs, Span I.e~th

'- Ix
,02

Note: 1 ft = 0.305 m.

80. 100. o. 140. 150.

L-Span (ft)
10

Maximum bimoment,1auc = 1.0(K1 X ~ X Ka) (14)

6. Multiply maximum moment by 1 + 4 from equation 3, and multiply maximum

bimoment by 1 + 1r from equation 4.

To account for S, one should multiply the resulting action of a single straight girder
subjected to a line of truck wheels by S. s
= 1.29 for S = 7 or 8 ft (2.14 or 2.44 m).
S = 1.57 for S =9 or 10 ft (2.75 or 3.05 m).

Dead Load Design

The dead load response of curved bridge systems has been predicted by Murphy (4),
and it realistically represents the response of the bare steel frame system . This tech-
nique has been computerized and then applied in the development of design curves. The
cu1·ves represent maximum ab and a. as a function of R, L, I,., and diaphragm S. The
cw·ves were based on typical girder properties obtained from a survey of bridge de-
sign as a function of girder length, as shown in Figure 3. If the actual design prop-
erties are different from the values in Figure 3, a correction factor is required to
modify the chart values. Bending stress is related by the following:

(15)

The warping stress is related by the following:

(16)

The design chaits, Figures 4 through 9, show the induced ai.s and a.s normalized
relative to the applied dead load per length. Thus any variation in dead load may
be considered in an actual design. The normalized stress versus R/L values are
plotted for span lengths of 100, 125, and 150 ft (30.5, 38.1; and 45.8 m). Thes~ plots
are dependent on I,., number of girders, and diaphragm spacing. The plots of ab versus
w
R/L values (Figures 4, 6, and 8) are independent of diaphragm spacing and I,., as de-
termined in the development of the curves (4). The plots of a. versus R/L values (Fig-
- w
ures 5, 7, and 9) are dependent on diaphragm spacing an.d I,.. These curves are limited
to a 4-girder system and are for specified stiffnesses. However, for other stiffness,
the following equation may be applied:

( ~) - (~)
W 0
W chart
x!u
I,.
(17)

where

( ~) • = referenced modified factor, and

f,,. = referenced chart value given in Figures 5, 7, and 9.
 11

Figure 4. Normalized bending stress for a span length of 100 ft (30.5 m).

G,s, = Girdor System

250.

wo.

150.

100,

50.

Note: 1 ft = 0 305 m.

Figure 5. Normalized warping stress for a span length of 100 ft (30.5 m).

Note: 1 ft =0.305 m. 1 in 4 = 41 .5 cm'.

soo.

400.

R/L = 1.5
l5? JOO.
"'
u

200.

24 . 22 . ~ . l • l • ..
DIAPHRAGM SPACIJK; S - l·'T,
12

300.

250,

200.

1o"k 150,

100.

50.

Note: 1 ft ~ 0.305 m.
o• .__~~~~~.......~~~~~~~~~~~~~~..-!t-~~~~~-....,\,--~~~~~~~..;.,,.
l. 2. s. 50.
fVL

Figure 7. Normalized warping stress for a span length of 125 ft (38.1 m).

Note: 1 ft= 0.305 m. 1 in~ = 41.5 cm 4 •

500.

300.
·I ~

200.

100,

4.0

24, 22. 20. 18,

DIAPHRAGM SPACill(l S - FT,
 13

Figure 8. Normalized bending stress for a span length of 150 ft (45.8 m).
400

J 50

300

WO

1 50

Note: 1 ft = 0.305 m.

Figure 9. Normalized warping stress for a span length of 150 ft (45.8 m).

500,
Note : 1 ft= 0 .305 m. 1 in.' = 41 .5 cm'

400,

JOO,

wo.

100,

w. 10.
o, o.
2.4. 22 , 20. 18. 16.
DillPHRAGM SPACillG S - FT,
14

To account for the behavior of 6- and R-gi:rde:r systems,

chart value. These chart values are modified by the following equation:

(a·) - (~) -a(~)

W exact W chart W
(18)

where

(19)

A(a•)'
w
=change in the value of the chart because of an increase in the number of
girders beyond 4 girders shown in Figure 10, and
f. = property given in Figure 10 •

The curves were developed for a girder spacing of 8.0 ft (2.44 m). However,
spacings up to 10 ft (3.05 m) are acceptable.

Torsional Properties

To determine design stresses or distortions, one must know girder section properties.
If the girder is subjected to torsion and bending, the torsional properties, in addition
to bending properties, will be required. The exact solution of the torsional properties
of composite sections has been demonstrated (10). However, by idealizing the compos-
ite section, one can develop a series of simplined equations (11).
Figure 11 shows a typical composite girder and its pertinent dimensions. If we
neglect the top girder flange and modify the concrete slab thickness, as shown in Fig-
ure 12, the dimensions are defined as follows: t 1 = nt., d1 = b 0 , t2 = W, d2 = dg + t,/2 -
t,/2, da = b,, and ta = t,. Using these dimensions, one can determine the resulting
torsional properties, '\1.'hich are shear center, normalized warping tw1ctions, warping
stiffness, and torsional constant. Shear center is

(20)

Normalized warping functions are

(21)

for slab, and

_ (d2 - OI) d
Wn 1 - 2 3 (22)

for beam. Warping stiffness is

 15

Figure 10. Normalized warping stress variation.

Note: 1 ft• 0.305 m. 1 in~ z 41 .5 cm 4

)00. Ix = 32,000. in4
L = 100'
250. I
I
200. I
1
181; I
150.
I
<J
I
ioo.
300. Ix = 80, ooo.
L = 125 1
in4
~\
250. 50. \
'"
200.
.5. 10.
150.
ltJ~ ;;...
<1 100. 1[J1~ 150. Ix = 125, 000. J.n4
<J L = 150 1
50. 100.
\
\
!V1 \
o. 50. \
o. 5. 10.
o. ' ~ ,._ VL
6 Girder• o. ~· 10.
8 Girders

Figure 11. Typical composite section. Figure 12. Idealized composite section.

bs i.-~~~~~d-1~~~~~~~~

I -.-'c::=====::=:i+ ti

~ ..
c.
II_. .,_ dg
16

,., 0\
\~UJ

Torsional constant is

(24)

With the evaluation of these torsional properties, one can evaluate the resulting nor-
mal stresses in the composite section due to bimoment. Normal warping stress for
slab is

(25)

Normal warping stress for steel is

(26)

DESIGN EXAMPLE

The design data just presented are sufficient to perform a preliminary design of a
curved bridge system. Using these data, we will design a single -span, 4-girder bridge
on which the girders are spaced 8 ft (2.44 m) apart at the centerline . Arc length is
98.0 it (29.9 m); cente1·1ine radius is 588.8 ft {179.0 m). Slab thiclmess is 8.5 in. (21.6
cm), and the composite girder is made of A36 steel.
For live load effects, the following values, derived from the basic equations, are
used:

1. R 01 = 588.0 ft (179.0 m),

2. S = 8.0 ft (2.44 m),
3. L max = 100.0 ft (30.5 m),
4. L 01 = 98.0 ft (29.9 m),
5. n = ~ = TI50588 = 5.88,
100
L 98
6. M = -mf=
50 = 1.96,
98
7. (L/R) 01 = "588 = 0.167, and
8. s = 1.29 ('.!)·
The evaluation of K1 from equations 6 and 7, gives

0
K1 moment = ~ 5 (L/R) + 1

= ~:~~ (0.167) + 1

= 1.004
 17

K1 bimoment = [(35n)(L/R) 2 - 15(L/R)] x 10 3

= [(35 x 5.88)(0.167) 2 - 15(0.167)] x 10 3

= 3,250.

Evaluation of Ka from equations 9, 10, and 11 gives

K2moment = (n + 3)(°·~L) + 0.6

= (5.88 + 3)(0.4 x 0.167) + 0.6

= 1.194

K2 bimoment = (M 6 l) (R/L)

= (1.9 ~ - 1) (6)

= 0.96

These values can be applied to the computed single straight girder moment, which
was obtained from the loading of a single set of AASHO truck wheels on a 98.0-ft (29.9-
m) girder.

Mstraight = (1,488.0 K'/2 X 12) XS

= 744 x 12 x 1.29

= 11,500 kips/in. (1300 kN/m)

Bi= 1.0

Applying equations 13 and 14, with Ka = 1.0, one arrives at the following result for
maximum forces in the 4-curved-girder system:

M_, = 1.004 x 1.194 x 11,500 = 13,750 kips/in. (1555 kN/m)

Bi= 3,250 x0.96 x 1.0 = 3,120 kips/in. 2 (21 528 MPa)

18

Allowable del'lign l'ltrAl'l~ for A36 i;t1;11;1l is 20 kips/in. 2 (138 MI>u). Assume t."lat 35
percent of stress is due to live load, and 65 percent is due to dead load. Then

S,equtred = l 3 ,~ 50 = 1,970 in. 3 (32 200 cm 3 )

We will now study a composite section based on moment (10). The web is 60.0 by
0.375 in. (152 by 0.95 cm); the flanges are 16.0 by 1.5 in. (41by 38 cm); and the slab
is 8.5 by 96.0 in. (21.6 by 244 cm). The composite properties are as follows:

1. I = 100,976.8 in. 4 (42 x 10 5 cm 4 )i{

2. Ix = 2,255,951 in. 6 (606 X 10 6 cm ),
Sw = 1,990 in .. (~2 600 cm ) , and
3 3
3.
4. W00 = 527.4 in. (3400 cm ).

The steel girder properties are as follows:

1. I = 51,949.0 in. 4 (21.6 x 10 5 cm 4 ),

2. { = 966,000 in. 8 (261x10 6 cm 6 ),
3. s• = 1,650 in. 3 (27 100 cm3 ) , and
4. W111 = 246 in. 2 (1590 cm2 ).

Therefore, the induced live load stresses are

ab = ~ = ;3 9~~o = 6.9 2
kips/in. (47.6 MPa)
'
_ BiWn _ 3,120 x 527.4 _ O 73 ki ;· 2 (5 05 MP )
a. - ---r,;- -2,255,951 - · ps m. · a

Check dead load stresses from Figures 4 and 5. For a 4-girder system where
600
L = 100.0 ft (30.5 m) and (R/L)max =WO= 6.0, then

(~) = 165

The dead load of the girder is computed as follows:

w = -b {[(60 x o.375) + 2(16 x 1.5)J 0 i~~o + (W x 8.o) 0.150}

w = 0.0875 kips/in. (1.53 kN/cm)
 19

Therefore,

ab= 165 x w

ab = 165 x 0.0875 = 14.4 kips/in. 2(100 MPa)

Where Sis the value given in Figure 3, the coefficient C1 = S/S. From Figure 3,
when L = 100.0 ft (30.5 m)

1/s = o.9 x 10- 3

S = 1,111.1 in. 3 (18 200 cm 3)

Therefore,

C - S/S - 1,111.1
l - - 1,650

C1 = 0.673

Therefore,

2
ab = 9.65 kips/in. (66.5 MPa)

The induced dead load warping stress is computed from Figure 5. If we assume a
diaphragm spacing of 20.0 ft (6.1 m) and if (R/L) m•x = 6.0, then

( ~)
w
= 100 x~
Ix

(~) = 100 x ~~:i~~

(~) = 61.9

(a.) = 61.9 x 0.0875

2
('lr".) = 5.4 kips/ in. (37 .3 MPa)
20

The coefficient C2 = (Wn/Iw)/(Wn/ Iw), where Wn/Iw is the design value and the
(Wn/ !w) vf!Jue il'l fotmd from Fi~ure 3_ If we use Figure 3 !!...T!.d if L _ 100.0 ft
(30.5 m), then (Iw/Wn) / Ix = 0.044 or (Iw/Wn) = 0.044 x 51,949 = 2,280. The design
ratio is (Iw /Wn) = 96:4~00 = 3,940. Therefore,

2 280
C2 = (Wn/Iw)/(Wn/Iw) = f.940 = 579

The warping stress is then equal to

a11 = 5.4 x 0.579 = 3.12 kips/in. 2 (21.6 MPa)

Total maximum normal stress is the sum of dead load and live load. The live load
stresses, however, must include impact, which is found from Figures 1 and 2. This
structure has an arc of 9 = 9.5 deg (use 15 deg). L = 98.0 ft (29.9 m), which gives
I, = 0.19, Ir = 0.52. The live load stresses are, therefore,

2
ab = 6.9(1.19) = 8.2 kips/in. (56.5 MPa)

a.,.= 0.73(1.52) = 1.11 kips/in. 2 (7.65 MPa)

The combined dead and live load stresses are equal to the following (1 kip/in. 2 =
6.9 MPa):

Bending Stress Warping Stress

Load (kips/ in. z) (kips/in. z)

Live 8,20 1.11

Dead 9.65 3.12
!: 17.85 4.23

The combined bending and warping stress is cr = 22.08 kips / in. 2 (153 MPa), which
indicates that a new section might be selected. However, a computer system analysis
using this selected stiffness and diaphragm spacing might be sufficient for an initial
trial.

ACKNOWLEDGMENTS

The information presented in this paper is in part the results of a research study spon-
sored by the Maryland Highway Administration and the Federal Highway Administration.
Their support and encouragement are gratefully acknowledged as are the work and
efforts of the various graduate students who worked on this research.
 21

REFERENCES

1. F. H. Lavelle and L. S. Boick. A Program to Analyze Curved Girder Bridges.

Univ. of Rhode Island, Engineering Bulletin 8, 1965.
2. Highway Structures Design Handbook. United States Steel Corp., Vol. 1, 1967.
3. L. C. Bell and C. P. Heins. The Solution of Curved Bridge Systems Using the
Slope-Deflection Fourier Series Method. Univ. of Maryland, Civil Engineering
Rept. 30, June 1968.
4. E. L. Murphy and C. P. Heins. Dead Load Analysis of Single Span Curved
Bridges. Univ. of Maryland, Civil Engineering Rept. 52, June 1973.
5. C. P. Heins. The Presentation of the Slope-Deflection Method for the Analysis
of Curved Orthotropic Highway Bridges. Univ. of Maryland, Civil Engineering
Rept. 15, June 1967.
6. K. R. Spates and C. P. Heins. The Analysis of Single Curved Girders With
Various Loadings and Boundary Conditions. Univ. of Maryland, Civil Engineering
Rept. 20, June 1968.
7. J. Siminou and C. P. Heins. Proposed Equations for Preliminary Design of
Curved Girder Bridge Systems. Univ. of Maryland, Civil Engineering Rept. 25,
June 1969.
8. K. M. Vashi and C. P. Heins. Impact Factors for Curved Highway Bridges.
Univ. of Maryland, Civil Engineering Rept. 32, Sept. 1969.
9. L. C. Bell and C. P. Heins. Curved Girder Computer Manual. Univ. of
Maryland, Civil Engineering Rept. 30, Sept. 1968.
10. J. T. C. Kuo and C. P. Heins. Torsional Properties of Composite Steel Bridge
Members. Univ. of Maryland, Civil Engineering Rept. 37, June 1970.
11. J. T. C. Kuo and C. P. Heins. Behavior of Composite Beams Subjected to
Torsion. Univ. of Maryland, Civil Engineering Rept. 39, Feb. 1971.

DISCUSSION
Douglas A. Nettleton, Federal Highway Administration

Charts are given in the paper for use in the design of curved I-girders, and a numerical
example is worked out. The example consists of a simple span on a 588-ft (179.0-m)
radius curve with an arc length of 98 ft (29.9 m), 4 girders spaced 8 ft (2.44 m) apart,
and an 8.5-in.-thick (21.6-cm-thick) slab.
Dead load stresses are based on the assumption that 4 girders act with diaphragms
spaced at 20 ft (6.1 m) and that the slab does not participate. Live load stresses are
based on composite action of the slab and girders, but no account is taken of diaphragm
spacing.
For live load, the slab is depended on for resistance to the rotational effect of cur-
vature on the girder. It is implicitly assumed that the girder will retain its cross-
section shape although the rotational and lateral restraining forces are applied at the
top flange only. This assumed action requires the transmission of radial forces from
the top flange to the bottom flange by means of the girder web. Such action would pro-
duce a bending of the web in a radial direction of such magnitude that the cross section
of the girder would be distorted; stresses would be of a magnitude that could exceed the
yield point. Properly spaced diaphragms would relieve the web of practically all of
this stress and distortion, and the lateral bending or warping stress in the bottom
flange would vary, approximately, with the square of the diaphragm spacing. By the
method in this paper, no account is taken of diaphragm spacing in the determination of
live load stress.
There is an error in the dead load calculation according to the values of 1/S from
Figure 3. This value should be about 0. 75 x 10-3 instead of 0.9 x 10- 3 • Use of the
22

corrected value changes the dead load bending stress from 9.65 kips/in. 2 (66.6 MPa)
to 11.6 kips/in. 2 (80.0 MPa).
Live load impact was calculated as 19 percent for bending stress, and as 52 per-
cent for warping or lateral bending stress. Because these stresses occur for the same
loading at the same point and are additive, the calculations are a decided departure
from past practice in design for impact.
Heins considers that failure to use diaphragms would be disastrous and that neglect-
ing the direct effect of diaphragm spacing on live load lateral flange bending would
seriously underestimate the lateral bending of the bottom flange.

AUTHOR'S CLOSURE
I appreciate Nettleton's comments and his observations concerning the effects of live
load on the diaphragms. His last paragraph, however, implies that I suggested that
no diaphragms were in the system during passage of the live load. This condition was
neither implied nor stated. Influence of diaphragms on design is considered only under
dead loading, but diaphragms are still present in the system during live load stresses.
Although the thoughts given in this paragraph are interesting, they are not pertinent.
The graphs presented are for preliminary design. The value for l/S from Figure 3
given as 0.90 x 10 - 3 was rounded from 0 .85 x 10- 3 •
The impact values that were used were based on maximum effects as given in the
curves. It was assumed that the worst case existed. Thus warping and bending
stresses were added.
It should be realized, as I stated in the beginning of the paper, that the information
presented is design data, which are helpful in selecting girder size and diaphragm
spacings before computer analysis.