Design of Precast Prestressed

Concrete Members Using

External Prestressing
This paper identifies some of the parameters and design considerations
for using external prestressing in prestressed concrete flexural
members. The behavior and design of these members are differentiated
from those prestressed with internal bonded tendons. The paper
elaborates on such issues as bonded versus unbonded tendons,
internal versus external tendons, the effects of friction and slippage at
tendon deviators, the behavior of members with deviators at different
locations, and various code issues. The paper also proposes a
Nihal Don Arlyawardena, modification to the Nebraska University (NU) girder series that would
Ph.D. reduce the web thickness of those girders by 3 in. (76 mm) (thus
Structural Design Engineer reducing their self-weight by 25 percent) and add external post-
Morrison Hershfield Corporation tensioning. Finally, a bridge design example of a precast, prestressed
Atlanta, Georgia
girder implementing the proposed modification combining —

pretensioned and external post-tensioned tendons is provided. The

—

design and analysis involve the use of a computer program, and the
validity of the proposed method is verified by comparing the numerical
results with published experimental data.

load applied to a prestressed The stresses and strains are also con

A concrete member with bonded

tendons induces equal strains in
the concrete and adjacent tendon. This
stant in an internally unbonded tendon
when friction is ignored.
When the tendons are unbonded, the
compatibility of strains between the analysis of a section based on strain
concrete and tendon is a basic design compatibility is, in general, inaccu
Amin Ghali, Ph.D. assumption in calculations for stress rate. Rather, the compatibility should
Professor of Civil Engineering and strain in prestressed sections. apply to the displacements of the con
The University of Calgary When prestressing is applied exter crete and the tendons at the anchor
Calgary, Alberta, Canada nally, the stresses and strains in the ages, and also at the deviators that do
tendon between the anchorages are not allow slippage.
constant, as are those in the tendon be The eccentricity of an external ten
tween the deviators if slippage is zero. don can vary with the level of loading.
84 PCI JOURNAL
In this paper, the concept is expressed
as the reduction in di., at ultimate load, Member top
Member top Deviators
where d,, is the smallest depth of the
tendon midway between deviators,
measured from the top face of the
member (see Fig. Ib). The reduction
in eccentricity and its effect in lower
ing the ultimate strength of a concrete
member are discussed below.
The parametric study and the design (a) (b)
example presented in this paper use a
computer program that performs a Fig. 1. Eccentricity variation of external tendon: (a) initial shape; (b) deformed shape.
comprehensive analysis, the details of
2 The
which are presented elsewhere.
’
1
3 assumes that the horizontal
program additional load that produces flexural tures. In this paper, however, only ex
and vertical translations of the con failure at a section. The ultimate mo ternal prestressing of new structures is
crete and the adjacent tendons are ment at the section will depend on the discussed. In recent years, designers
compatible only at the anchorages. increase in the stress 4f in the tendon have endeavored to reduce the web
The analysis accounts for slippage above its initial value prior to the in thickness in precast concrete bridge
and frictional forces between a tendon troduction of the additional load. girders to minimize their self-weight.
and the deviator. It also considers the At the failure section, zf and the The web thickness is frequently gov
change in member geometry as the ultimate moment will be greater in the erned by ease of production, rather
structure deforms. To trace the behav beam with bonded tendons. At other than by strength requirements.
ior of a member up to failure, the pro sections, the stress in the bonded ten When the web contains a post-ten
gram adopts nonlinear stress-strain re dons will be lower than at the failure sioning tendon, the web must be thick
lationships between the concrete and section. On the other hand, in the un enough to accommodate the prestress
reinforcement. bonded tendon, 4f will be constant ing duct and non-prestressed shear re
The program also considers the over the length between the anchor inforcement adjacent to each face of
time-dependent effects of creep and ages or between the deviators that do the web. External prestressing with
shrinkage of concrete and relaxation not allow slippage. tendons placed outside the faces of the
of prestressing steel. An internal ten Providing deviators and preventing web allows the web to be made thin
don is treated as an external tendon, deviator slippage (by bonding, for ex ner, resulting in a significant reduction
having closely spaced deviators. A ample) can enhance the ultimate in the girder self-weight.
bonded tendon is treated as an un strength of externally prestressed Another advantage of external pre
bonded tendon with negligible slip members. This aspect will be dis stressing is that it allows inspectors to
(due to a large amount of friction) at cussed below based on results of anal check the tendons in the event of cor
deviators. A sample analytical model ysis and published experimental data. rosion. The replacement or the addi
of an externally prestressed concrete tion of external tendons is relatively
structure is shown in Fig. 2. simple, particularly when the accom
ADVANTAGES OF EXTERNAL modation of future tendons is made in
Consider two prestressed concrete
members one with bonded and the
—
PRESTRESSING the design.
other with unbonded prestressed ten One well-established benefit of ex Potential disadvantages of externally
dons after creep, shrinkage, and re
— ternal prestressing is its ability to prestressed members include the rela
laxation have taken place, subjected to strengthen existing concrete struc tively lower ultimate strength and the

Concrete
member
Reference
axis

Fig. 2. Modeling of
externally
prestressed
External concrete structure
(Example of a
tendon box girder).

March-April 2002 85
 provides more load-carrying capacity
Stress (MPa) Stress (MPa) Stress (MPa) than prestressing a member entirely
with external tendons.
2300
-

METHOD OF ANALYS(S
A computer program is used here to
200 GPa 150 GPa
Stra Strain Li
Strain aid in the analysis of externally pre
-0.002 -0.0035
stressed concrete structures. The struc
tures are modeled as plane frames
Concrete Nonprestressed Carbon fiber composed of short, straight members
steel reinforced connected at several nodes. The pre
polymer (CFRP) stressing tendons are modeled as bars
3f (MPa)
Stress, having a negligible moment of inertia
and connected to the nodes by short
1830 arms (see Fig. 2).
The nodes are located on a reference
axis chosen at an arbitrary depth
1520 within the height of the member cross
:i90GPa section. Unlike the centroidal axes, the
position of the reference axis does not
0.008
Strain, change due to cracking or creep of the
0.035
concrete.
For < O.0O8,f (MPa)= I90000 The displacement components at
For > 0.008,f, (MPa) = 1848 —0.517/ (e— 0.0065) each node, translations in the x and y
directions, and member rotation are
Prestressed steel (Grade = 1860 MPa (270 ksi)1 determined by the solution of a series
’
2
of equilibrium equations:
6
Fig. 3. Stress-strain relationships used in the parametric studies and in the design
example. Note: 1 MPa = 0.145 ksi; 1 GPa = 145 ksi. [S]{D} =—{F} (1)

where [Sj is the stiffness matrix, and

wider cracks at overloading. The ulti crack widths can be reduced by provid {D} and {F} are, respectively, the
mate strength of externally prestressed ing supplementary bonded prestressed nodal displacements and nodal forces,
members can be increased, however, or non-prestressed 5
reinforcement. that can artificially prevent displace
by bonding the tendons to a sufficient Combining external and internal ments due to applied loads, prestress
number of 4 deviators. In addition, bonded tendons in the same member ing, temperature, creep and shrinkage
of concrete, and relaxation of pre
stressing steel.
The analysis involves iterative solu
F F tions of Eq. (1) to eliminate the out-of-
1 balance nodal forces resulting from
the variation of the stiffness matrix
jLJ i
J
2 ,i with the load level. A new stiffness
matrix is generated after each itera
(a) Tendon segments tion, based on the stress level and ge
ometry of the deformed structure.

Change of Forces in
Prestressing Tendons
(b) /1
2
1
z1 When slippage is prevented at devia
tors, the change in length of a segment
between adjacent deviators can be cal
(zl)/(lj culated from the translations at the tips
of the short arms shown in Fig. 2. The
(c) change in length divided by the origi
nal length of the segment gives the
Fig. 4. Typical change of strain in tendons due to load on externally prestressed beam: strain and, hence, the stress in the seg
(a) beam elevation; (b) tendons bonded at deviators; (c) tendons free to slip at deviators. ment, using the stress-strain relation-
86
PCI JOURNAL
ship of the material properties of the
tendon (see Fig. 3).
When the tendon is free to slip with
out friction at the deviators, the force
in the tendon is adjusted by the New
ton-Raphson iteration method using an
average strain given by the following
Segment
equation (see Fig. 4):
Fig. 5. Model for analysis of joint opening in precast segmental concrete structures.
Average strain = 1I
(tl) 1

(2)
deviators, with or without friction, re VERIFICATION OF
duces 4f and the ultimate moment ca ANAlYSIS
where pacity of the member.
length of ith segment of ten- The results from the computer pro
don gram are verified by comparing the
Joint Openings in Segmental predicted load and tendon stress at ul
1 = change in length of ith seg
(ill)
Construction timate with those obtained experimen
ment of tendon, calculated
by ignoring slippage To account for the effect of joint tally (see Table 1). The beams listed in
n number of segments openings in precast concrete segmen Table I are simply supported and have
The force F 1 in each segment and tal construction, a special element is internal or external tendons. The ten
the average force Favg are determined used to represent the joint (see Fig. 5). don profile is either straight or devi
from the known stress-strain relation The length of the element is equal to ated at the third points.
ship of the tendon material. The differ the depth of the member. Also, no Verification of the computer pro
ence Faug F 1 is eliminated by a series
—
non-prestressed reinforcement passes gram using additional experimental
of iterative operations. During each it through the joints between the seg data, including graphs of load-deflec
eration, initial tensile forces equal to ments. tion relationships up to failure, is pre
1 are assumed to exist in the The tensile strength of concrete is sented in Ariyawardena and Ghali.
2
Favg F
—

segments, with Favg and F, based on assumed to be zero. Ramos and Apari
the results of the preceding iterations. 7 used the same element and com
cjo PARAMETRIC STUDY
When friction at the deviators is pared its results experimentally. By AND DESIGN
considered, the forces in the segments considering a non-zero value for the
RECOMMENDATIONS
change due to the friction and are de joint’s tensile strength, the analysis
pendent on the direction of slippage.
2 can apply to the case in which the As an aid in deciding on the pres
At flexural failure, the slippage at the joint is filled with epoxy. ence of deviators, their type, and their

Table 1. Details and results of analysis and experiments on simply supported prestressed concrete beams.
Author
Tan and Ng’
3 Arduini et al.’ Du and Tao
°
2 Haraj Ii et al.’
Description of experiments
(1997) (1996) (1985) (1999)
and properties of materials
External straight External deviated Internal unbonded straight External straight
Type of prestressing
(beam designation) (TO) at third points (B) (A5)
Type of prestressed Aramid FRP Steel Steel
.
Steel
reinforcement
3000 3000 4200 3000
Span length, 1 (mm)
110 142 78.4 39.0
)
2
(mm
1297 366 810 935
Effective prestress (MPa)
402, 530 308, 650 308, 400 340. 612
),f. (MPa)
A, (mm
2
),f (MPa)
A (mm
2 201, 338 308, 650 —
—

34.6 33.8 30.6 40.1

Concrete strength (MPa)
Ultimate load, P (kN)
159 135 68 139
Experimental
160 135 70 137
By analysis
Increase in tendon stress,
zif, due toP (MPa)
368 486 505 450
Experimental
405 514 571 520
By analysis
1.01 1.0 1.03 0.99

1.13 1.16
(4,,p)pthm,,,t -

Note; l000nms=l m=3.28ft; s

3
=
2 qin.;
l.55x1
lmm 0 lkN=O.2248kip; IMPa=O.l45ksi.

87
March-April 2002
 outer segments would be inclined.
Axis of symmetry
External tendon For all beams, the span length 1 =
Segment length
5 =30 x iO mm
A 2 147.6 ft (45.00 m) and the depth of the
=4.5 m
(47 in
)
2 \P/2 tendon d = 56 in. (1.4 m); thus, the
/ ratio lId = 32. A relatively small
\ I’ II I value for d,, is selected such that the
I I I I I
prestressing force does not produce
15 ,n high negative moments at the sup
7.5m
‘IT’ ports. The cross-sectional area of the
49.2ft 24.6ft tendon is 47 sq in. (30000 mm ) and
2
the effective prestressing force is 8000
Segmental beam (two deviator case)
kips (36 MN).
The corresponding effective stress
lOm in the tendon is 170 ksi (1200 MPa).
0.2 The cross-sectional area of the non
1.8mj prestressed steel is A 3 6.2 sq in.
External tendon
—. Total = 4000 mm
2 (4000 mm) (or 0.1 percent of the
2
(6.2 in
)
2 gross area). The stress-strain relation
2m 0.6 0.6 2m ships of concrete and reinforcement
are depicted in Fig. 3. The self-weight
Cross-section of the beam is 6.85 kips/ft (100
Fig. 6. Details of the beams used in the parametric study. kN/m).
The analyses are performed for pre
cast monolithic beams or for beams
position, a parametric study is pre 2 with that of an identical reference composed of segments of length equal
sented below for externally post-ten Beam 1, where the prestressing steel to one-tenth of the span (without any
sioned simply supported beams, sub comprises straight bonded tendons. To epoxy between the segments). The
jected to two equal point loads, P12, at make the comparison possible, a computer program takes into account
the third points (see Fig. 6). The val straight horizontal tendon is assumed beams that have no deviators, two de
ues of the ultimate load P that cause in all beams, although in practice, in a viators at third points, or one deviator
flexural failure are compared in Table beam with deviators, the tendon in the at midspan.

Table 2. Load and tendon stress at ultimate for externally prestressed beams analyzed in the parametric study.
Ultimate load, Increase in tendon Ultimate load as percentage I Reduction of d5 Failure
Beam No. Beam description P (kN) Lstress, Af (MPa of P for reference beam at ultimate (mm) mode
I Monolithic; steel bonded
6540 628 100 Tendon
(Referencebeam) tendon 00
ylding
2 Monolithic; steel internal
unbonded tendon; 5560 428 concrete
.
85 0.0
.

friction ignored crushing

3 Monolithic; steel external
Tendon
. -

tendon with two deviators: 5035 494 77 124

yielding
.

4
Monolithic; steel external
tendon with two deviators;
. .

3790 195 58 concrete

.
92
free slip —
crushing
5 Monolithic; steel external
4630 350 71 concrete
tendon with one deviator .
.

0
crushing
iliti; steel externaiT
2600 100 40 concrete
tendon with no deviators
en, shing_
7 Monolithic; carbon FRP
external tendon;
.

5390 665 82 concrete

I 182
NodevrsLposli crushing
8
external tendon; 3685 210 oncre e
56 142
crushing
. .

9 Segmental; steel external

tendon with two deviators; 4670 448 tdon
.
71 92
no slip yieldine
10 Segmental; steel external
tendon with two deviators; 3470 206 concrete
.
53 114
freeslip crushing
Note: 1 kN = 0.2248 kip; 1 MPa = 0.145 ksi; 1 mm = 0.039 in.

88
PCI JOURNAL
 The prestressing tendon material is
steel or carbon fiber reinforced poly 7000.0 -

mer (CFRP). The stress-strain graph

of the CFRP is linear up to rupture
6000.0
(see Fig. 3). The modulus of elasticity
-

‘“‘Beam 1 (reference)
of the CFRP [22 x 10 ksi (150 GPa)1
is lower than that of prestressing steel, 5000.0 -
Internal bonded tendon
while its strength is higher.
As mentioned previously, the analy 4000.0 - ‘- Beam 3
sis involves a series of iterations to
External tendon with two deviators
satisfy equilibrium. The ultimate load
3000.0 (no slip at deviators)
is considered to have been attained
when the stress in the tendon exceeds
240 ksi (1650 MPa) or when the con 2000.0 -

\Beam 6
crete crushes. At this load level, con External tendon
vergence requires a large number of it
1000.0 - without deviators
erations, or else it will not occur.
The failure mode of each of the ana
lyzed beams is indicated in Table 2; 0.0 -

the term “tendon yielding” means that 0 500 1000 1500

the stress in the tendon at failure is Mid-span deflection (mm)
greater than 240 ksi (1650 MPa). In
Fig. 7, a comparison is made of load- Fig. 7. Load-deflection variations for three of the beams considered in the
deflection graphs for the beam when parametric studies.
the tendon is bonded (Beam I) and
when it is external with and without
deviators (Beams 3 and 6). steel in the prestressing tendons does LRFD Bridge Design Specifications,”
As expected, the ultimate load and not substantially change the ultimate give equations for determining the in
z1f are largest in the reference Beam I strength of the member. This is evi crease in Af of stress in prestressing
(monolithic beam with bonded ten dent by comparing Beam 3 with Beam tendons and the ultimate moment only
don). For the same beam, the change in 7 and Beam 4 with Beam 8. when the tendon is bonded or inter
eccentricity of the tendon at the The primary design recommenda nally unbonded. The equations of the
midspan section is zero. The smallest tion for externally post-tensioned ACT, CSA, and CEB-FIP codes yield
ultimate load and and the largest beams is that one deviator at midspan the results listed in Table 3 for Beams
reduction in d occur in the beam with or two deviators at the third points 1 and 2 of the parametric studies. The
out deviators (where d is the depth of should be provided. For two deviators, values of and the ultimate load P
the tendon at midspan). Providing the a significant loss of strength can be obtained by the computer analysis, re
beam with one deviator at midspan avoided if slippage at the deviators is ported in Table 2 and repeated in
(Beam 5) or two deviators at the third prevented. Table 3, indicate that the code equa
points (Beams 3 and 4) increases the
tions yield conservative estimates.
ultimate load substantially and elimi
ULTIMATE STRENGTH For internally bonded tendons,
nates or decreases the reduction in d.
CEB-FIP MC 90 assumes that the
Because of symmetry, no slippage BASED ON CODE stress in the tendon is 90 percent of its
can occur at a midspan deviator. A EQUATIONS tensile strength, divided by a partial
comparison of the ultimate loads of
Various codes of practice, including safety factor (equal to 1.0 for the beam
Beams 3 and 4 indicates a reduction in
ACI 31899,8 CSA A23.3-M94,
9 considered); but zif must not exceed
ultimate strength, when slippage oc
CEB-FIP MC 90,’° and the AASHTO 87 ksi (600 MPa). When the tendon is
curs freely at third-point deviators,
from 77 to 58 percent of the reference
Beam 1.
A comparison of Beam 3 with Beam Table 3. Values of ultimate load and corresponding increase in tendon stress
9 and Beam 4 with Beam 10 indicates for the beam in Fig. 6 calculated by equations of codes.
that a modest drop of the ultimate Increase in tendon stress, 4f (MPa) Ultimate Ioad,P(kN)
Internal Internal Internal Internal
strength occurs when the precast beam
Code bonded unbonded bonded unbonded
is segmental, instead of monolithic. ACT 3l8-99 559 209 6092 4424
Use of epoxy at the joints between the CSAA23.3-M94
8 519 175 5760 4210
segments can eliminate the difference CEB-FW MC90 474 0 5719 3387
in ultimate strength. Analysis 628 428 6540 5560
The use of CFRP tendons to replace Note: 1 MPa=O.l45ksi; I kN=O.2248kip.

March-April 2002 89
 externally unbonded, the CEB-FIP were less than 10 percent. Much the flanges and web should be re
Code assumes 4f is equal to zero un greater differences are calculated in duced by 3 in. (75 mm). Post-tension
less a more comprehensive analysis is the present parametric studies for ing would be accomplished by exter
performed. The ultimate strength cal beams having an lid,, ratio of 32. nal tendons having two deviators at
culated by ACT 318-99 is higher than Harajli et al.
14 also concluded from the one-third points of the span. The
that calculated by CSA A23.3-M94. It their analytical study, using simple deviators can either be made of con
should be mentioned that the amount beams with straight external tendons crete or fabricated steel elements, fas
of non-prestressing steel in the beams (l/d = 18.5), that the ultimate load tened to the faces of the web by
in the parametric studies is smaller was smaller when no deviators were through-bolts.
than what ACI 3 18-99 and CSA used, compared to beams with one or One layer of shear reinforcement
A23.3-M94 require. This, however, two deviators. They showed that the should be provided at the middle of
does not change the conclusions. Both reduction of ultimate load due to the the web. Welded wire fabric can be
codes allow designers to execute a absence of deviators was significant, used for this purpose. Alternatively,
comprehensive analysis in lieu of particularly when the area of pre the shear reinforcement can be double-
using the given code equations. stressing steel was small. head studs of larger diameter and
Pisani and Nicoli’
5 showed from greater spacings than commonly used
Published Experimental Data their analysis of simple beams that the in welded wire fabric. For ease of in
increase in stress, 4f, in the external stallation, a non-structural steel metal
The experimental and analytical
tendons due to loads close to ultimate element can hold a number of studs at
work in the literature, with one excep
was smaller than that in internally Un- the appropriate spacings. Such a stud
tion, confirms the conclusions of the
bonded tendons. The main reason for assemblage can be placed in the forms
parametric studies on the effect of de
this behavior was the reduction of ec prior to other reinforcement.
viators on ultimate strength of exter
centricity of external tendons with the With recent advances in concrete
nally post-tensioned beams.
increase of load. In practice, the effect materials and production methods,
Hindi et al.
4 concluded from their
of change in external tendon eccen casting the I-girders with 4 in. (100
experiments on three-span continuous
tricity with applied loads can be more mm) webs would be neither difficult
segmental box girder beams that the
significant than in the experiments nor costly. The strength of the con
ultimate strength can be increased by
mentioned above, because most pre crete can make the 4 in. (100 mm)
bonding the tendons at intermediate
stressed concrete members have a rel web thick enough to carry the required
locations in the span. However, the
atively larger lId ratio. shear in most applications.
analytical work of Muller and Gau
Lateral stability and transverse
2 on several precast segmental
thier’
bending of the web during handling
simple beams with different numbers
EXTERNAL. PRESTRESSING and shipping of the NU girders have
of deviators led to their conclusion
that the beams have almost the same IN PRECAST BRIDGE been considered by Seguirant’
7 and by
GIRDERS others. With the proposed thinner web,
ultimate strength, irrespective of the
the same concerns should be ad
number of intermediate deviators at I-shaped precast concrete girders
dressed, namely, temporary lateral
which the tendon is bonded. are frequently used for bridge super
support should be provided when nec
Tan and Ng13 tested several simply structures. In the following section, it
essary. Note that the girders require
supported beams with a span-to-depth is proposed that external post-tension
blocks for anchorage of the external
tendon ratio (l/d) equal to 15, each ing be used with the Nebraska Univer
tendons at the ends; for ease of form
with a straight external tendon. They sity (NU) girder series,’
6 a section that
ing, the blocks may be cast in a sepa
concluded that the change in tendon is gaining increasing popularity in the
rate stage.
eccentricity in the beams without de United States and Canada.
To illustrate the proposed design
viators resulted in a lower load carry The web thickness of post-tensioned
method, a numerical design example
ing capacity, compared to beams with NU girders is 7 in. (175 mm). With a
is presented in Appendix B.
one or two deviators. 1 in. (25 mm) of cover at each face of
Tan and Ng’
3 also suggested that the web, and two No. 5 (16 mm) stir
providing a single deviator at the rups, the remaining space for post-ten
midspan section leads to satisfactory sioning and the non-prestressed hori CONCLUSIONS
ultimate load behavior. Because of the zontal bars is 33/4 in. (95 mm). Based on the results of this investi
small lId,, ratio, however, the differ It is proposed that all the dimen gation, the following conclusions can
ences between the ultimate loads for sions of the NU cross section be be drawn:
the beams with and without deviators maintained except that the widths of 1. It is structurally advantageous to

90 PCI JOURNAL
combine pretensioned concrete mem dons can be easily inspected in the ACKNOWLEDGMENTS
bers, having internally bonded ten event corrosion occurs.
This research program has been sup
dons, with externally post-tensioned 4. It is proposed that the web thick
ness of the Nebraska (NU) girder be ported by grants from the Natural Sci
tendons.
reduced by 3 in. (76 mm), thus, reduc ences and Engineering Research
2. Future strengthening of concrete
members can be more easily accom ing its self-weight, and combined with Council (NSERC) of Canada and ISIS
plished by external post-tensioning external post-tensioning. Canada (Intelligent Sensing of Innova
than by internal prestressing. The re 5. It is recommended that in the design tive Structures).
placement or addition of new tendons of externally post-tensioned tendons, one The authors would like to thank the
can be easily accommodated. deviator be provided at midspan or two PCI JOURNAL reviewers for their sug
3. In external post-tensioning, ten- deviators at the third points. gestions and constructive comments.

REFERENCES
1. Ariyawardena, N., “Prestressed Concrete with Internal or Ex First Edition, American Association of State Highway and
ternal Tendons: Behaviour and Analysis,” Ph.D. Thesis, The Transportation Officials, Washington, DC, 1994.
University of Calgary, Calgary, Alberta, Canada, July 2000, 12. Muller, J., and Gauthier, Y., “Ultimate Behaviour of Precast
276 pp. Segmental Box Girders with External Tendons,” External Pre
2. Ariyawardena, N., and Ghali A., “Prestressing with Unbonded stressing in Bridges, ACI SP-l20, A. Naaman and J. Breen,
Internal or External Tendons,” .Jour,wl of Structural Engineer Editors, American Concrete Institute, Farmington Hills, MI,
ing, American Society of Civil Engineers, Washington, DC. 1990, pp. 355-374.
This paper has been accepted for publication. 13. Tan, K., and Ng, C., “Effects of Deviators and Tendon Config
3. Ariyawardena, N., and Ghali, A., PCF, Prestressed Concrete uration on Behavior of Externally Prestressed Beams,” ACI
Frames, Computer Program, American Concrete Institute, Structural Journal, V. 94, No. 1, January-February 1997, pp.
Farmington Hills, MI, 2002. 13-22.
4. Hindi, A., MacGregor, R. J., Kreger, M. E., and Breen, J. E., 14. Harajli, M., Khairallah, N., and Nassif, H., “Externally Pre
“Enhancing the Strength and Ductility of Post-Tensioned Seg stressed Members: Evaluation of Second-Order Effects,” Jour
mental Box-Girder Bridges,” Proceedings of the Workshop on nal of Structural Engineering, V. 125, No. 10, October 1999,
Behaviour of External Prestressing in Structures, Saint-Rémy pp. 1151-1161.
lès-Chevreuse, France, E. Conti and B. Foure, Editors, June 15. Pisani, M. A., and Nicoli, E., “Beams Prestressed with Un-
1993, pp. 153-162. bonded Tendons at Ultimate,” Canadian Journal of Civil Engi
5. Eibl, J., “Externally Prestressed Bridges,” External Prestress neering, V. 23, No. 6, December 1996, pp. 1220-1230.
ing in Bridges, ACt SP-120, A. Naaman and J. Breen, Editors, 16. Geren, K. L., and Tadros, M. K., “The NU Precast/Prestressed
American Concrete Institute, Farmington Hills, MI, 1990, pp. Concrete Bridge I-Girder Series,” PCI JOURNAL, V. 39, No.
375-387. 3, May-June 1994, pp. 26-39.
6. Ghali, A., and Neville, A. M., Structural Analysis: A Unified 17. Seguirant, S. J., “New Deep WSDOT Standard Sections Ex
Classical and Matrix Approach, Fourth Edition, B & FN Spon, tend Spans of Prestressed Concrete Girders,” PCI JOURNAL,
London, United Kingdom, 1998, 831 pp. V. 43, No. 4, July-August 1998, pp. 92-119.
7. Ramos, G., and Aparicio, A. C., “Ultimate Behaviour of Exter 18. Ghali, A., Favre, R., and Elbadry, M., Concrete Structures:
nally Prestressed Concrete Bridges,” Journal of the JABSE, Stresses and Deformations, Third Edition, E & FN Spon, Lon
Structural Engineering International, March 1995, pp. 172- don, United Kingdom, 2002, 604 pp.
177. 19. Arduini, M., Tommaso, A., and Giacani, S., “Modelling of
8. ACI Committee 318, “Building Code Requirements for Struc Concrete Beams Reinforced with External FRP Prestressed
tural Concrete (ACT 3 18-99),” American Concrete Institute, Tendons,” Proceedings of First International Conference on
Farmington Hills, MI, 1999. Fiber Composites in Infrastructure, Tuscon, AZ, V. 1, January
9. CSA Standard A23.3-M94, Design of Concrete Structures, 1996, pp. 481-490.
Canadian Standards Association, Toronto, Ontario, Canada, 20. Tao, X., and Du, G., “Ultimate Stress in Unbonded Tendons in
1994, 220 pp. Partially Prestressed Concrete Beams,” PCI JOURNAL, V. 30,
10. CEB-FIP MC 90, Model Code for Concrete Structures, No. 6, November-December 1985, pp. 72-91.
Thomas Telford, London, United Kingdom, 1993, 434 pp. 21. Canadian Portland Cement Association, Concrete Design
11. AASHTO, LRFD Bridge Design Spec(fications (Si Edition), Handbook, Ottawa, Ontario, Canada, 1995.

March-April 2002 91
 APPENDIX A — NOTATION
A = cross-sectional area of beam = length of ith segment of tendon
= area of prestressing tendon M = ultimate moment
b = width of compression face of member n = number of segments
c = depth of compression zone P = ultimate load
D = nodal displacements S = stiffness matrix
d = depth of prestressing tendon from extreme com t = instant of time
pression fiber w = uniform distributed load
F nodal forces = distance from centroidal axis to top face of beam
Favg = average force in each segment Yb = distance from centroidal axis to bottom face of beam
1 =
F force in each segment = coefficient in standard ACT compression block
f’ = specified compressive strength of concrete [see Eq. (B4)]
ice = effective stress in prestressed reinforcement Af = increase in stress in prestressing tendon
stress in prestressing tendon at ultimate (41)1 = change in length of ith segment of tendon
I = moment of inertia of beam section = prestressing reinforcement ratio
= span length = symbol for summation

APPENDIX B — DESIGN EXAMPLE

This example illustrates the method of calculating the ulti To expedite installation of the studs, the heads at one end
mate strength of the bridge girder shown in Fig. B 1. The of the girder may be clamped in a sheet metal trough, which
girder is composed of a modified precast concrete NU serves as a stud spacer. Welded wire fabric can serve as al
6 and a cast-in-place deck slab. The cross-sectional
girder’ ternative shear reinforcement. It can be verified, using equa
dimensions of the precast beam are the same as those of the tions given in the codes, that the web need not be thicker
standard NU girder, except that the widths of the flanges than 4 in. (102 mm) in order to provide the shear strength
and the web are reduced from 7 to 4 in. (180 to 100 mm). required in most practical applications.
The girder span is 131 ft4 in. (40.0 m). The specified compressive strengths of concrete for the
The bottom flange and top flange of the precast beam are modified NU precast girder and the cast-in-place deck slab
pretensioned with steel strands having initial prestressing are 7000 and 4000 psi (48 and 28 MPa), respectively. The
forces (immediately before transfer) of 800 and 94 kips (3.6 prestressing steel is Grade 270 (1860 MPa).
and 0.42 MN), the corresponding stress is —1.05 ksi (—7.25 The cross-sectional properties of the girder are as follows:
MPa), respectively. With this prestressing force, the precast A = 749 sq in. (0.484 m )
2
beam can carry the weight of the cast-in-place concrete deck I = 706 x 10 in. 4 (0.294 m)
4
without shoring. No cracking is expected to occur at this y = —43.55 in. (—1.105 m)
stage. Yb 35.2 in. (0.895 m)
After the deck slab has attained sufficient strength, the The cross-sectional areas and the locations of the cen
composite girder is post-tensioned with a force of 1020 kips troids of the prestressing reinforcement are indicated in Fig.
(4.54 MN) by external tendons with two deviators [see Fig. B 1(a). The weights of the precast girder and the deck slab
B 1(b)]. This configuration is different from the standard NU are 795 and 1040 lb/ft (11.6 and 15.1 kN/m), respectively,
girder, in which the post-tensioning tendons run through assuming 9 ft (2.7 m) spacing between girders.
ducts in the middle of the web. This change makes it possi It is required to find the magnitude of additional uniform
ble to reduce the web thickness from 7 to 4 in. (175 to 100 load that produces flexural failure of the midspan section.
mm), and thus reduce the self-weight of the girder by 25 Assume that the external tendons do not slip at the devia
percent. tors. The stress-strain relationships for concrete and pre
Steel studs of 3/4 or 1 in. (20 or 25 mm) diameter are pro stressing steel shown in Fig. 3 are used.
posed as shear reinforcement in the web. The studs are an The computer analysis involves the calculation of stresses
chored mechanically by heads at their ends, with the heads and deformations at the times t , t
1 , t
2 , and t
3 . The prestress
4
having a diameter of three times that of the stud. The spac transfer of the pretensioned tendons and the application of
ing between the studs can vary to provide the required shear the self-weight of the precast girder occur at t , when the
1
strength. age of concrete is 3 days.

92 PCI JOURNAL
 Cast-in-situ slab
108 in. / H
I
42.5 in. /
—

‘p
9 in.
-j

f Two pretensioned strands;

2 (280 m,n
area = 0.43 in )
2

Studs with heads; diam. 3/4

43.55 in.
or 1 in.; spacing var. Precast NU-Girder Yt
—..
(1.105 in)
78.75 in.
. 4 in. (2 in)
External tendons
Area = 5.43 ii x
Pretensioned strands; area
= 4.34 in
2 (2800 mm)
2
Yb= 35.20 in.
(0.895 ni)

5.25
1 in.

T Centroid of strands

(a)

65ft—8 in.
9 in cast-in-situ 4ck in.

External tendons
N precast NU-Girder —

Axis 1f
t Center line of Center line of
symmetry [_..viator support

23ft 42ft—8 in.

(7 m) (13 in)

(b)

Fig. 81. Example of composite bridge girder with pretensioned strands and external post-tensioned tendons: (a) cross section;
(b) half-span elevation of the precast part showing profile of external tendons.

The time-dependent effects of creep, shrinkage, and relax , under the effect of additional uniform
Failure occurs at t
4
ation are assumed to occur between girder ages t1 = 3 days loading of 5.20 kips/ft (76.1 kN/m), when the tendons yield
and t2 = 60 days. In analyzing these effects, the creep coeffi [their stress exceeds 240 ksi (1650 MPa)]. The correspond
cient of concrete is taken equal to 1.3 and the shrinkage is ing ultimate moment at midspan is 182,000 kip-in. (20500
equal to —100 x 10.6 (based on CEB-FIP MC 9010). A re kN-m). The last column in Table B 1 indicates the mode of
8 equal to —2.9 ksi (—20 MPa) is assumed
duced relaxation’ failure at t
.
4
for the prestressing steel.
, also at age 60 days, the external post-tension
At time t
3
Ultimate Moment by ACI 318-99
ing, the weight of the deck slab, and the superimposed load
are applied. At time t , additional uniform loading is intro
4 As mentioned above, the ACT 318-99 Code does not give
duced, causing flexural failure at midspan. Note that the equations for the ultimate flexural strength of members pre
time-dependent changes between t 4 are ignored.
3 and t stressed with external post-tensioned tendons or with a com
Table Bi gives the calculated stresses at top and bottom bination of these with pretensioned tendons. Nevertheless,
fibers at t
, t
1 2 and i3. It can be seen that no tensile stress oc Eqs. (Bl) through (B4), given in ACI 318-99 for use with
curs before application of superimposed dead and live load. unbonded internal tendons, are applied below. This gives,
The forces in the pretensioned tendons and the post-ten for comparison purposes, an approximate value of the ulti
sioned tendons are also given. mate moment M at the midspan of the girder in Fig. B 1.

March-April 2002 93
 Table Bi. Concrete stress and forces in the prestressed tendons at midspan in bridge girder of the design example.
1 immediately after
t 1, after time- 3 immediately after
t State of concrete
Time pretensioning transfer dependent changes post-tensioning and tendons at (4
Coi,crete stress at top fiber of precast part,
—1.05 (-7.25) —1.07 (—7.38) —1.34 (—9.24) Concrete uncracked
ksi(MPa) - —

Concrete stress at bottom fiber,

—1.05 (—7.25) —0.93 (—6.41) —2.13 (—14.68) Concrete cracked
ksi (MPa) - —- -

Force in pretensioned tendons,

769 (3.42) 722 (3.21) 696 (3.10) Tendons yielded
kips(MN) -- - .- - -

Stress in pretensioned tendons,

177 (1220) 166(1150) 160(1110) Tendons yielded
ksi (MPa) - . -—
-

Force in post-tensioned external tendons, —

kips (MN)
— 1020 (4.54) Tendons yielded
—

Stress in post-tensioned external tendons, —

ksi (MPa)
— 188 (1300) Tendons yielded

Note: I MPa = 0.145 ksi; I kN = 0.2248 kip.

Thus, treating the tendons in this example as if they were in Substitution of numbers in Eqs. (Bi) through (B4) gives:
ternally unbonded, the code requirements can be expressed as: f,,, =215 ksi (1480 MPa)
c =6.7in.(l7Omm)
+ L 215(9.77)[78.1 0.85(6.7)12]—

f,, =fse + l0ksi (Le + 6Oksi) (B 1)

loop’, 158,000 kip-in. (17850 kN-m)
A
pp
=
(B2) The uniformly distributed load that causes this moment is
bd equal to:
M =f,A’,,(d — f3
c
1 12) (B3) w =8MIP
= 8(158,000)/(13 1.33 x 12)2

= 0.5 10 kips/in.
c= (B4)
0.85 f’f3
b
1 = 6.11 kips/ft (91.0 kN/m)

where The combined self-weight of the precast girder and the

3 = 9.77 sq in. (6302 mm
A,, ) total area of prestressed
2 deck slab is 1.84 kips/ft; thus, the additional load that causes
steel in tension zone failure is (6.11 1.84) 4.27 kips/ft. This estimate is more
—

4, = 78.1 in. (1984 mm) = distance from extreme compres conservative than the 5.20 kips/ft calculated by the more so
sion fiber to centroid of compression reinforcement phisticated analysis discussed above.
b = 108 in. (2743 mm) = width of compression face of The parametric studies presented in the main body of this
member paper have shown that the ultimate moment with internally
c = depth of compression zone unbonded tendons is higher than with external tendons and
= stress in prestressed reinforcement at nominal strength smaller than with bonded tendons. In the above approximate
( specified yield strength of prestressing tendons) calculation, the two types of tendons in the beam in Fig. B 1
fse = effective stress in prestressed reinforcement (after are treated as if they were internally unbonded.
all prestress losses) = 177 ksi (1220 MPa), calcu Note that a higher estimate for M would be obtained by
lated for the pretensioned tendons summing up the contributions of the bonded tendons and the
f’ = 4 ksi (28 MPa) = specified compressive strength of external tendons. Eq. (B3) would be applied with an appro
concrete priate code value forf, for each tendon type.
i
13 0.85

94 PCI JOURNAL