AN ABSTRACT OF THE THESIS OF

ROY CHARLES MARTIN for the MASTER OF SCIENCE

(Name) (Degree)
in CIVIL ENGINEERING presented on (.k .(71
(Major) (Date)
Title: A TORSIONAL INVESTIGATION OF A CONCRETE I GIRDER
Redacted for privacy
Abstract approved:
' T. J. McClellan

This paper investigates a particular bridge construction tech-

nique which could result in a significant torsional moment. The
structural member under consideration is a prestressed concrete
girder for which no torsional consideration is made.
Torsional shear stresses can be adequately approximated for
an I-shaped girder. Laboratory model tests were performed to
verify an approximate analytical approach for the calculation of a
torsion constant. The laboratory model analysis resulted in moment
rotation curves from which the torsion constant could be obtained.
These tests confirmed the reliability of the approximate analytical
method.

The results of this investigation indicate that, depending on the

sequence of construction, the design shear stress may be in error
as much as 50 percent.
A Torsional Investigation of a Concrete I Girder
by

Roy Charles Martin

A THESIS

submitted to
Oregon State University

in partial fulfillment of
the requirements for the
degree of
Master of Science
June 1969
APPROVED:

Redacted for privacy

Profes(* of Civil Engineering

in charge of major

Redacted for privacy

Heao of Departmeyt(oktivil Erifikeering

Redacted for privacy

.uean of uractuate School

Date thesis is presented wr\m_ 3,tc10

Typed by Marion F. Palmateer for Roy Charles Martin
 ACKNOWLEDGMENT

The author would like to thank Professor T. J. McClellan for

his assistance and guidance. I would also like to thank Mr. Robert
Hafstad, employee of the California Bridge Department, for the
initial idea.
 TABLE OF CONTENTS

Page
INTRODUCTION 1

GENERAL THEORY 3

PROTOTYPE 15

MODEL STUDY 17

Test Procedure 18
Results 21
Comparison of Experimental and Analytical Results 27

INVESTIGATION OF PROTOTYPE 29

Loading Conditions 29
Analysis 30

DISCUSSION OF RESULTS 32

CONCLUSIONS 34

BIBLIOGRAPHY 35

APPENDIX 37
 LIST OF FIGURES
Figure Page
1 Torsional shear stress distribution on a rectangular
section. 5

2 Relative shear stress distribution for an AASHO Type

III girder. 11

3 Effects of fixed-end torsion. 14

4 Prototype loading. 15

5 Laboratory set up. 19

6 Model loading. 19

7 Typical failure of model. 20

8 Crack patterns of torsional failure. 20

9 Section view of model taken at loading point. 21

10a Moment vs. rotation - test 2. 23

1 Ob Moment vs. rotation - test 3. 24

10c Moment vs. rotation - test 4. 25

10d Moment vs. rotation - test 5. 26

11 Theoretical moment rotation curve for a prestressed

model. 28

12 Laboratory set up - test 4. 38

13 Laboratory set up - test 5. 40

 LIST OF TABLES
Table Page
1 Loading sequence - test 4. 39
2 Loading sequence - test 5. 41
 A TORSIONAL INVESTIGATION OF A CONCRETE I GIRDER

INTRODUCTION

In the design of most engineering structures, a torsional con-

sideration is of secondary importance. The majority of these struc-
tures are designed with no consideration for torsion, and if it is
present, may be assumed too small to endanger the integrity of the
structure. If torsion becomes the primary loading, the engineer will
probably use his personal past experience to arrive at, for the most
part, a conservative design. Little help is given to the engineer in
the design of torsional concrete members. The ACI Code 1 devotes
three lines to the entire subject of torsion.
This paper deals with the torsional properties of a particular
concrete I-shaped girder. It is the intent of the author to investigate
the stresses incurred in a prestressed I-shaped girder under a com-
bined torsion and bending load.
A widely used construction technique employed by various
agencies, including the California Bridge Department, can induce a
sizeable torsional moment in a bridge girder. This torsional moment
is produced during erection and stems from the manner in which the
overhanging deck is cantilevered from the exterior girders.

1
ACI Code, June 1963.
 2

The author will therefore try, through a model study and litera-
ture review, to determine the effects of this possible torsional
mom ent.
 3

GENERAL THEORY

Although of somewhat limited use, an exact solution to most

torsional problems can be obtained. Excluding the circular shaft,
the solution can become mathematically involved. The general theory
for non-circular sections subjected to torsion was developed over
100 years ago. While this basic theory is rarely used in design, a
series of simplifying assumptions has evolved which are commonly
used today. Presented in this section will be the general theory of
torsion as related to the following investigation.
The torsional properties of a circular shaft are well known.
The relatively simple and exact theory clearly gives the magnitude
and distribution of the torsional induced stresses. The torsional
properties of this section are related by the well known formulae

TTc
7:

and
6 T
JG

The validity of the above equations are based on several assumptions

(Pan lilio, 1963) which include:
1. The shaft is a right circular cylinder.
2. The material is homogeneous, isotropic and obeys Hooke's
law.
 4

3. The twisting moments lie in transverse planes.

4. Plane sections remain plane as the bar is twisted.
5. The diameters remain straight during deformations.
For the most part these assumptions are met. While concrete cannot
be classed as homogeneous material, few discrepancies should arise.
An understanding of the above theory gives limited aid when
dealing with non-circular sections. Many of the above assumptions
are violated, in particular, number four. Non-circular sections will
undergo a warping of originally plane cross sections during twisting.
Compared to the circular section, it is this warping that causes an
increase in shearing stress in some parts, with an accompanying de-
crease in other parts of the section. Some of the basic differences
can be seen in the well investigated rectangular section.
Although the shear stresses are again zero at the center with
the maximum occurring at points on the periphery, a different stress
distribution arises. The maximum shear stress will occur at the
points on the surface nearest the axis of the section. The stress dis-
tribution (Winter et al. , 1964) is assumed to vary linearly as shown
in Figure 1. A complete analysis of the rectangular section will be
given below.
The previously mentioned exact solution for many prismatic
sections was developed by Saint Venant in 1853. At this time he
formulated the general equation (Lyse and Johnston, 1936) for
 5

torsional shearing stresses by introducing the stress function "F, "

where
2F 2F
a a
2G(I)
2 2
ax ay

This stress function was visualized as a dome-like surface above the

particular cross section. Saint Venant then showed that for this
dome stress function, the shearing stresses at any point were propor-
tional to the slope of the stress function over that point, and that the
volume beneath this dome was proportional to one-half the twisting
moment. Although a stress function exists for every solid prismatic
section, the above equation has been solved for only a few of the more
common shaped sections (Seely and Smith, 1952).

Figure 1. Torsional shear stress distribution on a rectangular

section.
 6

In 1903 (Hartog, 1952) Prandtl developed the membrane analogy,

which is commonly called the soap film analogy. Prandtl discovered
that Saint Venant's differential equation for the stress function of a
given cross section was the same equation that described the surface
of an elastic membrane stretched over the opening of the section and
deformed by a pressure differential. The tremendous benefit real-
ized by this was not in the calculation for the shear stresses, but a
method bywhich the shear stress relative intensity and distribution
could be visualized.
By this aid (Seely and Smith, 1952) it can be seen that the shear
stress is zero at the center and corners of a rectangular section;
that the shear stress at a re-entrant corner is critical and theoreti-
cally goes to infinity; that the maximum shearing stress occurs at
one of the points of contact of the largest circle inscribed within the
section; that the torsional rigidity of a thin rectangular section is
much less than that of a square of equal area, plus numerous other
observations.
As seen from the two basic equations for a circular section,
the relation between torque and angular rotation is a function of the
material and the cross sectional shape. This is also true for non-
circular sections. For a circular section,

T = (I)JG
 7

while for a rectangular section (Lyse and Johnston, 1936)

T = EKG

where K, the torsion constant, is a measure of the torsional

rigidity of the section. Bach (Zia, 1961) was the first to make an
approximation for the torsion constant of a rectangular section. He

proposed that

bn 3
1
K
3

where b and n are the lengths of the long and short sides re-
spectively. This constant was proposed for relatively slender sec-
tions, but was thought to induce little error for non-slender sections.
Seely and Smith (1952) replaced the 1/3 factor with a varying param-
eter which is a function of b/n. The constant 1/3 is obtained only
when b/n tends toward infinity.
With certain modifications, the torsion constant can be approxi-
mated for almost any solid prismatic section. With the aid of the
soap film analogy, Lyse and. Johnston (1936) did extensive work in
this field for rolled steel sections. They added to the Bach formula
for a rectangular section and said:

K=
1

3
bn
3
- 2Vn 4

This expression is good for any rectangular section. The additional

 8

term accounts for end effects and can be visualized by the soap film
or membrane analogy as the volume lost when the membrane is
stretched down to conform at the ends. The term V is called the
end constant and is a non-linear function of b/n.
It is now possible to develop the torsion constant for a rectangle
of sloping sides. For a long rectangle

1 3
K -Van

and for any differential length dx

1 3
K n dx .
3

Therefore, considering the general case of a sloping rectangular

section as given below,

n
r m
 9

1
K S r3 dx
0
or
1 x
K 3- +
b
(m - n)1 dx
0

which gives

K = b/12(m +n)(m 2 +n2)

as the torsion constant. Including the end effects

K = b/12(m +n)(m 2 +n2) - V mm4 - n4

n

Assuming the torsional rigidity of a section can be determined from

the rigidity of its component parts, the torsion constant for the flange
of an I beam can be formulated.

T
Thus for the typical flange section shown above,

2 2
K
f
=
b12 -w
(m +n)(m +n ) + 3wm 3
1
- 2Vnn4
 10

and for the web of an I beam

(d - 2m)w 3
1
K =
w 3

where d is the total depth of the section. Assuming a symmetrical

section about both X and Y axes, K for the entire section can
be given as

b-w
K 6 (m +n)(m2 + n2) +
2
wm 3 +
1
(d - 2m)w 3 - 4V nn4
3 3

Although an apparent solution has been obtained, a look at the soap

film above the assumed cross section reveals a larger displacement
at the junction of the flange and web than had been taken into account.
Trayer and March (as cited by Lyse and Johnston, 1936) proposed
that the additional rigidity was proportional to the fourth power of the
diameter of the largest circle that could be inscribed at the juncture
of the flange and web. They proposed that an addition to the torsion
constant be made at each web-flange juncture of

K = a D4

where a is a function w, m, and the radius of fillet. Therefore,

for a symmetrical section, the term 2aD4 should be added onto the
above equation to make the solution complete. Once this value has
been obtained, a check for the critical shearing stress may be made.
This maximum shearing stress will occur at the middle of the side of
 11

the thickest component rectangle and is equal to

= Tn
(max) K

Through his tests on torsional stiffness of prestressed con-

crete AASHO girders, Tamberg (1965a) demonstrated the validity of
the soap film analogy in the distribution of shearing stresses. Re-
calling the assumption that the largest shearing stress will occur at
one of the contact points of the largest inscribed circle, Tamberg
found the relative stress distribution as given in Figure 2.

t Tb
w
9. 8G4) 12. 3G4 7. 6G4

Figure 2. Relative shear stress distribution for an AASHO Type

III girder.
Tamberg's results showed the relative order of magnitude of
shear stress distribution for a typical AASHO section under pure
torsion. A pure torsional loading for this type of section is uncom-
mon and a combination of bending and torsion is usually the case.
Since the transverse shear is almost completely taken by the web,
 12

any appreciable vertical shear will move the critical shear location
into the web. The summation of these two shearing stresses along
one side of the web will usually control the design for shear. Al-
though the addition of a bending load to a torsional member appears
detrimental, Cowan (as cited by Fisher and Zia, 1964) has shown that
accompanying bending moments will actually increase the torsional
capacity of a concrete beam. Many theories have been given for the
reason, and several foreign codes have realized this effect in granting
an increase of allowable shearing stresses up to 30 percent. Fisher
and Zia (1964) stated that the phenomenon was probably due to dif-
ferent modes of failure that occur in a pure bending or torsional
failure. It was recommended that no additional increase in shearing
stresses be allowed until further study was made.
Prestressing a concrete member that will undergo torsional
loading will greatly increase the structural capacity of this member.
Zia (1961) reported an increase of 281 percent in the torsional
strength of a member over that of plain concrete when f' = 6000 psi.
A concrete member subjected to a torsional load will fail in diagonal
tension. This failure can be delayed in either of two ways: (1) by
effectively placing reinforcement across all potential failure sur-
faces or, (2) by prestressing. It can be shown by Mohr's circle
analysis that this induced compressive stress will reduce the diago-
nal tension for a given shear stress. While additional web
 13

reinforcement will add some torsional strength, this additional com-

pressive stress will greatly increase the strength of the member.
Although prestressing contributes to the structural capacity of the
member, failure of this member under pure torsion is sudden and
explosive (Cowan and Armstrong, 1957).
In a concrete member, torsional stresses differ from vertical
shear with respect to the formation of diagonal tension. In torsion,
these stresses exist on all four faces of a rectangular section
whereas they only extend over two faces when subject to bending.
This additional stress is reflected in the type of reinforcement used
for a torsional concrete member. The addition of a horizontal com-
ponent of tensile stress requires corresponding reinforcement. U-
shaped stirrups and bent-up bars are therefore inadequate for torsion
and a closed hoop-type stirrup must be used (Winter, 1964).
To conclude the general theory for this investigation, mention
should be made of fixed-end torsion. If the ends of a non-circular
section are restrained from warping, pure torsion as described
above no longer exists. If an I section is completely restrained at
both ends, the effect of two fixed-end beams is produced. Seely and
Smith (1952) derives a method whereby the "effective" torsional
length is shortened by the amount "a" equal to

h ,/
a=
2
V EI /KG
 14

The effect of the fixed-end condition essentially disappears within

the length "a". That section of member denoted as 1' can now be
considered to be under pure torsion.
For an I section with sloping flanges, Lyse and Johnston
(1936) gives h as
h=d m+n
2

Figure 3. Effects of fixed-end torsion.

 15

PROTOTYPE

Although the prototype is a particular bridge in Northern Cali-

fornia, this type of construction is common throughout the state. The
construction technique is used when the temporary falsework cannot
be supported from the ground.
Figure 4 illustrates the method used to support the overhanging
deck. The span length is approximately 65 feet with the ends re-
strained by the abutments or bent caps. The girder is similar to a
type II Standard AASHO section with a depth of 36 inches, a top and
bottom flange width of 19 inches, and a web thickness of seven inches.

Figure 4. Prototype loading.

 16

For this particular type of bridge, no distinction can be made

between an interior and exterior girder. With an overhanging deck
of three feet, sufficient width is available to position the live loads
to give similar shear and moment values as found in an interior
girder.
The prototype torsional moment can be calculated assuming:
1. The torsional moment is uniformly distributed over the
length of the girder.
2. The effects of the small intermediate diaphragm are
neglected.
For the above conditions, a maximum torsional moment of approxi-
mately 19. 3 ft-kips occurs near both supports. Determination of the
effects of this torsional moment is the major purpose of this paper.
 17

MODEL STUDY

Scale model replicas of the above 2

I section were made. The
purpose of the model study was to verify the Lyse-Johnston approach
for the determination of the torsion constant. The basic data ob-
tained from this model study will be a plot of moment versus angular
rotation. As presented earlier,

K=EG
Therefore, with the slope of the above plot and a knowledge of the
shearing modulus of elasticity, the torsion constant can be obtained.
By comparing this value to that obtained by the analytical approach,
a possible range of error will be obtained that can be applied to the
prototype.

The cross sectional shape was a 1:6 scale model of the proto-
type previously shown. A casting tolerance of 0.10 to 0.15 inches
was maintained. A model length of seven feet was used for a longi-
tudinal scale of 1:10.
With the model's shape determined, an elastic design for a
uniformly loaded fixed-fixed beam was made. The design load was
75 pounds per foot and no consideration for torsion was made. The

2
The laboratory models were not prestressed.
 18

principle reinforcement consisted of #16 gage welded wire mesh.

The stirrups were of #16 gage wire, placed three inches on center.
The concrete was a sand-cement mixture of 2. 25:1 with a water-
cement ratio of 60 percent. A constant soaking by wet rags provided
the cure with all models being tested between 14 and 21 days. Three
standard mortar samples were cast with each beam. Following the
testing of each model, its corresponding compressive strength was
determined by the failure of these mortar cubes.
A seemingly significant model to prototype similitude is miss-
ing with regard to prestressing. Considering the desired results,
prestressing will have no effect upon the initial slope of the moment
rotation curve. For an uncracked concrete section, a given torsional
moment will produce a certain angular rotation.

Test Procedure

The laboratory set up and test procedure can best be explained

by Figures 5 and 6. The torsional moment was applied at three points
in a manner similar to that of the prototype. To measure angular
rotations, ten Ames dials were placed at five specific locations. By
positioning a dial on the top and bottom flange, the resultant angular
rotation was determined. The test procedure then simply consisted
of adding weights to the three loading arms and recording the dial
readings to failure.
 19

Figure 5. Laboratory set up.

Figure 6. Model loading.

 20

Figure 7. Typical failure of model.

Figure 8. Crack patterns of torsional failure.

 21

Results

Excluding prestressing, the author made every attempt to ob-

tain similitude between the model and prototype. This goal was not
achieved with respect to end restraint. While a total of five models
were tested, a truly fixed-end condition was never achieved.
Two types of movements were encountered at the supports. Be-
sides an end rotation, a horizontal translation toward the applied load
was observed. A correction for these movements was made based on
the structural action of the section. Inspection of the model at one
of its loading points

Figure 9. Section view of model taken at loading point.

indicates that the center of rotation lies within the distance "d".
From this observation it was concluded that any movement of the
 22

bottom flange at dial positions one and five was mainly a horizontal
translation. A linear horizontal correction was therefore made along
the length of the beam to correct for this type of end movement.
Completion of this horizontal correction still showed a sub-
stantial movement of the top flange at these extreme dial positions.
To obtain a true angular rotation, the relative rotation between dial
positions one and two, and that of four and five, were calculated.
These relative rotations were then converted into radians per inch
of length.

Since a fixed-end condition was attempted and partially

achieved, mention should be made as to the effect of fixed-end tor-
sion. This effect was neglected in the above study because:
1. A truly fixed condition was never achieved.
2. The distance "a" for the test model was only three inches.
Therefore, considering the dial locations and the data used, no cor-
rection was felt necessary.
The five laboratory models were built and tested in two stages.
Initially it was planned to test only three models, (the first being
used to obtain a working laboratory procedure and the remaining two
for the desired moment rotation curves). Due to laboratory error,
end movement corrections could not be made on the results of the
first two models; therefore, two additional tests were performed.
The results of the model study can be summarized by Figure 10.
60

50

40 0

30

--Theoretical
20
Experimental
/ Theoretical M/(1) = 15. 0 x 106 lb-in. a
2
10
Experimental MAO = 11.8 x 106 lb-in.
= 4100 psi

2 3 4 5 6 7

Angular Rotation (rad/in) x 105

Figure 10a. Moment vs. rotation - test 2.
60

50

40

30
Theoretical

20
/ / 0
Experimental
Theoretical Mic) = 12. 8 x 106 lb-in. 2
Experimental M/4 = 13. 6 x 106 lb-in. 2

f' = 3000 psi

10

3 4 5 6
Angular Rotation (rad/in) x 105
Figure 10b. Moment vs. rotation - test 3.
300

250

200

150

Theoretical
100
Experimental
Theoretical M/ct. = 12. 8 x 106 lb-in. 2
2
Experimental M/(1) = 12. 3 x 106 lb-in.
50
fl = 4400 psi

5 10 15 20 25 30 35
Angular Rotation (rad/in) x 105
Figure 10c. Moment vs. rotation - test 4.
300

250 O

O
200

150

Theoretical
100 Experimental
Theoretical M/4 = 12. 6 x 106 lb-in. 2
Experimental M/4 = 12. 3 x 106 lb-in. 2
50 f' = 4200 psi

5 10 15 20 25 30 35
Angular Rotation (rad/in) x 105
Figure 10d. Moment vs. rotation - test 5.
 27

Subsequent calculations will only use the results of the last two tests.

Comparison of Experimental and Analytical Results

The torsion constant for the above model can now be determined
by both methods. By the Lyse-Johnston approach, K was found to
be 8. 06 in. 4 From the model study,

K =G3.G

where the shearing modulus of elasticity can be approximated knowing

the ultimate strength of f'. The modulus of elasticity varies with the
ultimate strength (Winter, 1964) as

Ec= 33w I. 5 V7q

where w is the unit weight of concrete. Using an average ultimate

strength of 4300 psi,

Ec = (33)(143) 1. 5 VV 4300

= 3.72 x 106 psi

Using an assumed value of Poisson's ratio of 1/6 (Seely and Smith,

1952),
E
G
2(1 +1.1)

12. 3 x 106
6
-7.75 in. 4
1.59 x 10
 28

The above calculations are valid for an uncracked section where

a linear relationship exists between moment and rotation. These test
results show initial cracking taking place at approximately 50 percent
of the ultimate torsional moment. A similar prestressed I section
loaded to failure should have given a more linear plot but with the
same initial slope as found in the above study.

Slope = 12. 3 x 10 6 lb-in. 2

Angular Rotation (rad/in)

Figure 11. Theoretical moment rotation curve for a prestressed

model.
 29

INVESTIGATION OF PROTOTYPE

Loading Conditions

Two types of loading conditions could occur in the field. A pos-

sible but illogical sequence of construction could result in the pouring
of the overhanging deck followed by a waiting period. The usual deck-
ing operation progresses with the pouring of the entire bridge width,
the process usually starting at one end and moving longitudinally.
Each of the above construction techniques produces a similar
torsional moment. The twisting moment in the first possibility simply
consists of the weight of the overhanging deck, plus falsework, acting
through an average lever arm. The torsional moment for the more
commonly used technique consists of:

T T
1
+ T2 - T3

where
T1 7-- the moment described for the first possibility.
T2 = the moment due to the finishing machine.
T3 = the moment due to the support conditions of the first
interior bay. 3
Since T and T3 are approximately equal, T1 becomes the
maximum prototype loading and is equal to 19. 3 ft-kips.

3See
Figure 4.
 30

The angular rotation incurred during construction will be main-

tained with the hardening of the deck concrete. Therefore, the in-
duced torsional moment produces a permanent angular rotation which
must be considered.

Analysis

Consideration must first be given as to the location of the maxi-

mum combined shearing stress. With the deck and girder acting
compositely, the neutral axis was found to lie just above the junction
of the web and upper flange. The maximum vertical shear stress will
therefore occur at the upper web-flange junction. From the soap film
analogy and Tamberg's (1965b) theoretical stress distributions, the
maximum torsional stress will occur at points on the periphery of the
lower flange. Considering both vertical and torsional shearing
stresses, the maximum combined shear stress will occur at the
juncture of the web and upper flange.
As given in the General Theory, the maximum shearing stress
on a rectangular section is equal to

Tn
(max) K

This expression is good only for the shearing stress at the middle of
the long side of a rectangle. Since the point in question is at the
junction of the web and flange, an approximation for n must be
 31

made. Correlating Tamberg's (1965b) results to the prototype, an

average n of 7. 5 is obtained.
Using the Lyse-Johnston method for the determination of K

K = 12, 255 in. 4

Therefore,
Tn. (19, 300)(12)(7. 5) _
T 142 psi
K 12,255

From the model study,

K = 11,780 in. 4

resulting in a shear stress of 148 psi.

An analysis of the prototype girder as shown in Figure 4 con-
firmed the assumption that torsional effects had not been considered.
Using the design loading of an H2O-S16-44 and a stirrup spacing of
4
#4 at six inches, the net maximum design shear stress was found to
be 287 and 292 psi respectively. If a torsional stress of 145 psi is
included in the above, the allowable shear stress appears to have
been exceeded by 50 percent.

4
Prototype dimensions and specifications were obtained from
the State of California, Department of Public Works, for bridge
no. 2-32.
 32

DISCUSSION OF RESULTS

Two assumptions were made in the calculation for the torsional

moment. They were:
1. The torsional moment is uniformly distributed over the
length of the girder.
2. The effects of the small intermediate diaphragm are
neglected.
Although the prototype was loaded at specific points, it was
assumed the total torsional moment would be equally divided at the
two supports. This assumption should cause little error.
The preceding investigation neglected the possible presence of
an intermediate diaphragm. If the diaphragm is poured in the decking
operation, the preceding calculations are valid. It is not known what
effects an existing diaphragm would have on the above investigation.
The stiffening effects of one in-place diaphragm could effectively
decrease the torsional length by 50 percent.
The above study demonstrates the reliability that can be ex-
pected in the calculation of torsional stresses. In the approximate
relationship between shear stress and torque, most designers could
determine T and n with some confidence. However, they would
probably have some reservations as to the accuracy of their torsion
constant. The errors resulting from the Lyse- Johnston method
 33
should be no larger than that incurred in the approximation for an
average "n", or in the calculation of the applied torque.
 34

CONCLUSIONS

1. The Lyse-Johnston approach provides an easy and reliable

method for determining the torsion constant of a concrete I-shaped
girder.
2. All intermediate diaphragms should be poured before the
casting of the deck. Additional diaphragms to stiffen the exterior
girders are advised.
3. The torsional shear stresses should be considered. Since
the effects of the intermediate diaphragms are unknown, the magni-
tude of these stresses should be larger than that obtained by assum-
ing a full torsional restraint at the diaphragm.
 35

BIBLIOGRAPHY

American Society for Testing and Materials. 1966. Compressive

strength of hydraulic cement mortars (using 2-in, cube
specimens). In: 1966 Book of ASTM standards. Part 9.
Cement; lime; gypsum. Philadelphia. p. 70-76. (ASTM
Designation C109-64)
Cowan, Henry J. 1960. Design of beams subject to torsion related
to the new Australian Code. Proceedings (Journal) of the
American Concrete Institute 56:591-618.
Cowan, Henry J. and Stewart Armstrong. 1957. The torsional
strength of prestressed concrete. In: Proceedings of the
World Conference on Prestressed Concrete, San Francisco,
1957. San Francisco, Lithotype Process Company. p. 18.1-
18.12.

Fisher, Gordon P. and Paul Zia. 1964. Review of code require-

ments for torsion design. Proceedings (Journal) of the
American Concrete Institute 61:1-22.
Gersch, B. C. and W. H. Moore. 1962. Flexure, shear and torsion
tests on prestressed concrete I-beam. In: Bridge deck
design and loading studies. Highway Research Board, Bulletin
339:43-66. (Washington, D. C. , National Academy of Sciences-
National Research Council. Publication 1021)
Gesund, Hans et al. 1964. Ultimate strength in combined bending
and torsion of concrete beams containing both longitudinal and
transverse reinforcement. Proccedings (Journal) of the
American Concrete Institute 61:1509-1522.
Hartog, J. P. Den. 1952. Advanced strength of materials. New
York, McGraw-Hill. 379 p.
Lin, T. Y. 1963. Design of prestressed concrete structures. 2d
ed. New York, Wiley. 614 p.
Lyse, Inge and Bruce G. Johnston. 1936. Structural beams in
torsion. Transactions of the American Society of Civil
Engineers 101:857-865. (Paper no. 1941)
 36
Panlilio, F. 1963. Elementary theory of structural strength.
New York, Wiley. 478 p.
Seely, Fred B. and James 0. Smith. 1952. Advanced mechanics of
materials. 2d ed. New York, Wiley. 680 p.
Tamberg, K. G. 1965a. Elastic torsional stiffness of prestressed
concrete AASHO girders. Part 1. Proceedings (Journal) of
the American Concrete Institute 62:479-491.
Tamberg, K. G. 1965b. Elastic torsional stiffness of prestressed
concrete AASHO girders. Journal of the American Concrete
Institute. Sup. Detroit, American Concrete Institute. 22 p.
(Paper no 62-31, part 2)
Winter, George et al. 1964. Design of concrete structures. 7th
ed. New York, McGraw-Hill. 660 p.
Zia, Paul. 1961. Torsional strength of prestressed concrete
members. Proceedings (Journal) of the American Concrete
Institute 57:1337-1359.
APPENDIX
 37

LIST OF TERMS
Term Dimensions
a Length of beam shorten by fixed-end torsion in.
c Distance from longitudinal axis to point in
question in.
Ec Modulus of elasticity of concrete 2
lb/in.
F Saint Venant's stress function lb/in.
f' 2
Ultimate compressive strength of concrete lb/in.
G Shearing modulus of elasticity 2
lb/in.
I Moment of inertia about Y axis 4
y in.
J Polar moment of inertia 4
in.
K Torsion constant in. 4
1' Effective torsional length in.
T Torsional moment ft-lbs
Membrane analogy end constant dimensionless
0 Total angular rotation between two given points radians
I-1. Poisson's ratio dimensionless
T Shearing stress lb/in. 2
(i) Angular rotation per unit length of member rad/in.
 38

6.5 7.8
13.6 20.5 + 21. 4 14.2
00 00
L
0 2 3 4 0o

Dial Positions

00 0
r

0 0

15.1 20.5 21.4 15.8 .1,

5.0 6.2
Loading Positions

Figure 12. Laboratory set up - test 4.

 39

Table 1. Loading sequence - test 4.

Total load, pounds
Loading Position
Load No. A C
1 11.8 12.0 11. 9
2 23. 2 23. 1 23. 2
3 34.1 34.0 34. 0
4 44.7 44.7 44. 5
5 56.1 55.8 55. 5
6 67.4 67.1 67.1
7 78.4 78.9 78.7
8 89.1 89.4 89.6
9 100.5 100.9 101. 1
10 110.9 110.9 111. 2
11 121.4 121.6 122. 0
12 131.7 131.5 132. 4
13 141.3 141.0 142. 1
14 150.9 150.8 151. 7
15 162.1 162. 1 163. 0
16 173.6 173.8 174. 5
 40

...--1----
7.2 7.0
--1
14.9 21. 4 20. 5 13.0
1

O0 0o

Oo 1 2 3 4 5 0

Dial Positions

,-1..----)

O0 A B C

O0

..,..1..... 16.1 21.4 20.5

6.0
Loading Positions

Figure 13. Laboratory set up - test 5.

 41

Table 2. Loading sequence - test 5.

Total load, pounds
Loading Position
Load No. A C
1 11.2 11.3 11. 3
2 22.5 22.4 22. 6
3 34.5 34.2 34. 5
4 45.1 44.9 45. 2
5 56.6 56.4 56. 6
6 67.3 67.3 67. 5
7 78.9 78.8 78. 8
8 89.8 89.6 89. 7
9 101.6 101. 2 101. 5
10 112.0 111.7 111. 8
11 123.4 123. 1 123. 3
12 134.4 134.1 134. 3
13 144.4 144.2 144. 2
14 154.0 153.8 153. 8
15 165.2 165.5 165. 2
16 175.7 176.0 175. 6