ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 93-S47

Flexural Strength of Externally Prestressed Concrete

Bridges

by Angel C. Aparicio and Gonzalo Ramos

This paper presents the results of the analytical study of the f c′

behavior up to failure of 74 externally prestressed concrete bridges f ps = f pe + 10, 000 + -------- psi (2)
using a nonlinear finite element model. The aim of the study was to k ρp
obtain accurate values of the stress increase of tendons beyond the
effective prestress (Δfps). A code proposal is derived from the results where k = 100 for L/dp ≤ 35 and k = 300 for L/dp > 35, where
and is compared with existing current codes. L is the span length, dp is the distance from extreme compres-
sion fiber to centroid of prestressed reinforcement, and ρp is
Keywords: bridges; external prestressing; prestressed concrete; strength; the ratio of prestressed reinforcement (Aps /bdp).
unbonded prestressing. The AASHTO 1994 Code provides the following formula
based on the work of Naaman and Alkhairi7
INTRODUCTION
External prestressing is widely used in the United States dp L
f ps = f pe + Ω u E p εcu ⎛ ----- – 1⎞ ----1- ≤ 0.94 f py (3)
and Europe1,2 for medium- and long-span bridges. Externally ⎝c ⎠L
2
prestressed bridges are designed with prestressing tendons
placed outside of the concrete section, but yet remaining for T-section behavior
within the bounds of the cross section. Load transfer between
the tendon and the concrete is affected through the end A ps f ps + A s f y – A′s f ′y – 0.85β 1 f c′ ( b – b w )h f
anchorages and deviators. Because the tendons are c = -------------------------------------------------------------------------------------------------------
- (4)
0.85 f c′ β 1 b w
unbonded, external loading produces an increase in the stress
of the tendons beyond the effective prestress (Δfps) lower
than that of bonded prestress bridges. Therefore, to evaluate for rectangular section behavior
the nominal flexural resistance, the prestressing steel stress
at ultimate (fps) must be calculated from the effective stress A ps f ps + A s f y – A′s f y′
c = -------------------------------------------------
- (5)
(fpe) using the equation 0.85 f c′ β 1 b

fps = fpe + Δfps (1) (see notation appendix)

It is believed that neither European nor American prac-
Because there is no compatibility between the strain in the tices are consistent with the actual behavior of externally
prestressing steel and the concrete at every cross section, the prestressed bridges. For example, to ignore any increment of
increment of stress at ultimate must be evaluated by taking stress, as most European codes do, is too conservative. On
into account the whole structure, rather than performing the the other hand, American codes sometimes recommend
calculation at each section independently. unreasonably high increments of stress. To reconcile
An extensive review of existing equations for estimating between these extreme positions, a research program was
Δfps was conducted by Naaman and Alkhairi.3 The authors initiated at the Technical University of Catalunya (UPC) to
concluded that at the time of the study, none of the published study the behavior of bridges with external prestressing. The
prediction equations combined sufficient accuracy and program includes the development of a numerical analysis
simplicity to be recommended for code implementation. model and the performance of parametric studies, as well as
Therefore, most European codes allow no increase of the performance of laboratory model tests and full-scale field
prestressing steel stress at ultimate unless an accurate anal- tests. This paper presents results of the parametric studies
ysis is performed.4 On the other hand, American codes5,6
and some European codes recommend values or approxi-
mate formulas for estimating Δfps. For example, the 1989 ACI Structural Journal, V. 93, No. 5, September-October 1996.
Received August 17, 1995, and reviewed under Institute publication policies.
version of the ACI Building Code (revised in 1992) gives the Copyright © 1996, American Concrete Institute. All rights reserved, including the making
following equation to predict the stress at ultimate in of copies unless permission is obtained from the copyright proprietors. Pertinent discus-
sion will be published in the July-August 1997 ACI Structural Journal if received
unbonded tendons by March 1, 1997.

512 ACI Structural Journal/September-October 1996

 ΔLt also depends on the ultimate deflection. This, in turn,
Angel C. Aparicio is Professor of Bridges at the ETS de Ing. de Caminos de Barce-
lona, Spain. He received his MS in 1972 and his PhD in 1980. He is a member of depends on the span-to-depth ratio, the presence of conti-
ATEP, FIP, and IABSE. He has designed or consulted on the design of more than 200 nuity, the construction scheme (monolithic or segmental),
prestressed concrete bridges in Spain, South America, Korea, and the Middle East. and the cross section type (T-section or box girder).
His research interests include nonlinear analysis of prestressed concrete structures
and external prestressing since 1990. c. dp and a depend on the cross section.
The present parametric study evaluates the influence of
Gonzalo Ramos is Assistant Professor of Civil Engineering at the ETS de Ing. de
Caminos de Barcelona. He received his MS in 1990 and his PhD in 1994. He has been
these variables on the increment of prestressing steel stress
involved in the design of several prestressed concrete bridges, including two exter- at ultimate (Δfps). Values of Δfps are obtained by performing
nally prestressed ones. His research interests include nonlinear analysis of externally nonlinear analysis of several bridge decks.
prestressed bridges.

Bridge configurations
The parametric study includes 102 analyses on 74
developed to obtain the increment of stress of external different bridges. The geometric and design properties of
tendons at ultimate. The effects of several parameters are these bridges are presented in Tables 1, 2, and 3. Several
analyzed. These include factors such as the effect of span assumptions are made:
continuity, construction scheme (monolithic or segmental), a. Live load is considered centered on the bridge axis.
span-to-depth ratio of the deck, cross section type, tendon b. All the bridges are analyzed with free slip of the tendons
layout, etc. at the deviators. This assumption is valid even for nonreplace-
able tendons, as has been proven in Reference 12.
FINITE ELEMENT MODEL c. Joints of segmental bridges were always considered dry
The parametric study uses a finite element model developed (without epoxy).
in this study to perform the ultimate analysis of externally The mechanical properties of the materials are summarized
prestressed concrete bridges. The model is general, and is in Table 4. The average values of strengths and modulus of
applicable for the analysis of simply supported and continuous elasticity were used. Double T-section decks were designed
beams and monolithic or segmental construction, including with better quality concrete than box girder decks to avoid
free slipping or no slipping at the deviators, as well as external problems associated with excessive initial compressive
and/or internal prestressing. stresses due to prestressing. The stress-strain curves used in
The model includes three different kinds of elements: the analysis were adopted from Model Code 90.4
1. Beam-reinforced concrete elements with 6 deg of
freedom at each node. Simply supported bridges
2. Prestressing elements; bonded internal and unbonded A longitudinal section showing the bridge profiles,
external. including the span lengths and distribution of deviators for the
3. Joint elements for segmental bridges.
A detailed description of the model can be found in Refer- Table 1—Simply supported box girder bridges*
ences 8 and 9.
This model has been verified by comparison with existing L = 30 m L = 40 m L = 50 m
laboratory tests, mainly with C.E.B.T.P. tests,10 those h = 2.00 h = 2.00 h = 2.20
performed at UPC,11 and the only full-scale destructive test.12 As1 h = 2.20 h = 2.70
h = 2.70 h = 3.30
Monolithic
PARAMETRIC STUDY h = 2.00 h = 2.00 h = 2.20
Definition of variables As2 h = 2.20 h = 2.70
The nominal moment strength (Mn) of a flexural member Simply h = 2.70 h = 3.30
is given by ACI 318-89 as supported
box girder h = 2.00 h = 2.00 h = 2.20
AP1 h = 2.20 h = 2.70
M n = (A s f y + A ps fps ) ⎛ d p – ---⎞
a
⎝
(6) h = 2.70 h =3.30
2⎠ Segmental
h = 2.00 h = 2.00 h = 2.20
AP2 h = 2.20 h = 2.70
Once Aps is obtained from service limit state (fully h = 2.70 h = 3.30
prestressed or partially prestressed members), the nominal *See notation appendix.
moment strength depends on: Note: 1 m = 3.28 ft.
a. As, the amount of reinforcing steel for monolithic
construction (As = 0 at the joints for segmental construction).
b. fps, which depends on Δfps [Eq. (1)]. The value of Δfps is Table 2—Simply supported double T-section
obtained from bridges*
L = 30 m L = 40 m L = 50 m
ΔL t h = 2.00 h = 2.00 h = 2.20
Δf ps = --------E (7) As1
Lt p h = 2.20 h = 2.70
Simply
supported Monolithic h = 2.70 h = 3.30
where Lt is the length of the tendon between fixed points of double h = 2.00 h = 2.00 h = 2.20
the tendon (i.e., points with no slip). Only special mechanical T-section
As2 h = 2.20 h = 2.70
devices can prevent slipping at the deviators. Therefore, Lt is h = 2.70 h = 3.30
usually the length between end anchorages, even when using *See notation appendix.
nonreplaceable tendons, as has been proven in Reference 12. Note: 1 m = 3.28 ft.

ACI Structural Journal/September-October 1996 513

simply supported bridges analyzed in this study, can be (approximately 135 kip) distributed over three axles. The
observed in Fig. 1. When segmental construction is dead load and live load factor used to proportion cross
assumed, the segment length is 5 m (16.4 ft). All decks were sections is 1.5. As an example, Fig. 2 and 3 show the cross
designed as fully prestressed. For monolithic bridges, two section corresponding to a depth of 3.30 m (10.83 ft) for one
amounts of reinforcing steel were used: As1 and As2. As1 box girder and one double T-section deck.
corresponds to a minimum reinforcement needed to control
cracking due to shrinkage and temperature before prestressing Continuous bridges
(0.15 percent Ac). As2 is the reinforcing steel needed to It is herein assumed that a long viaduct formed by several
achieve the flexural strength at ultimate. As2 was calculated continuous spans can be modeled as five continuous spans.
prior to the analysis, assuming an increment of tendon stress at At the end supports, continuity moments due to dead load are
ultimate of 100 MPa (Δfps = 100 MPa = 14.5 ksi). applied. In the following examples, it is assumed that all
Segmental bridges were also provided with two different span lengths are equal to 50 m (164 ft).
amounts of prestressing steel: Ap1 and Ap2. Ap1 is the area of These bridges are assumed to have two different types of
prestressing steel necessary to achieve the flexural strength construction: monolithic and segmental.
at ultimate, assuming no increase of tendon stress (fps = fpe; Monolithic bridges
Δfps = 0). Ap2 is the area of prestressing steel necessary to Monolithic bridges are designed with two amounts of rein-
achieve the flexural strength at ultimate, assuming an forcing steel, with the same criteria previously used for
increase of tendon stress of 100 MPa ( fps = fpe + 100; Δfps simply supported bridges. Three different prestressing
= 100 MPa = 14.5 ksi). layouts were studied.
The Spanish live load model is used in the study, which is a. 1-span prestressing layout; the tendons are anchored at
very similar to most of the European models and a little every pier diaphragm.
higher than the AASHTO live load model. It is formed by a b. 2-span prestressing layout; the tendons are anchored at
surface load of 4 kN/m2 (0.58 psi) and one truck of 600 kN every other pier diaphragm.

Table 3—Continuous box girder bridges*

L = 50 m
1-span prestressing layout 2-span prestressing layout 3-span prestressing layout
Load 1 Load 2 Load 1 Load 2 Load 1 Load 2
h = 2.00 h = 2.00 h = 2.00 h = 2.00 h = 2.00 h = 2.00
h = 2.20 h = 2.20 h = 2.20 h = 2.20 h = 2.20 h = 2.20
As1
h = 2.70 h = 2.70 h = 2.70 h = 2.70 h = 2.70 h = 2.70
h = 3.30 h = 3.30 h = 3.30 h = 3.30 h = 3.30 h = 3.30
Monolithic
h = 2.00 h = 2.00 h = 2.00 h = 2.00 h = 2.00 h = 2.00
h = 2.20 h = 2.20 h = 2.20 h = 2.20 h = 2.20 h = 2.20
As2
h = 2.70 h = 2.70 h = 2.70 h = 2.70 h = 2.70 h = 2.70
Continuous h = 3.30 h = 3.30 h = 3.30 h = 3.30 h = 3.30 h = 3.30
box
girder h = 2.00 h = 2.00
h = 2.20 h = 2.20
Layout #1
h = 2.70 h = 2.70
h = 3.30 h = 3.30
Segmental
h = 2.00
h = 2.20
Layout #2
h = 2.70
h = 3.30
*See notation appendix.
Note: 1 m = 3.28 ft.

Table 4—Material properties

Material Compressive strength Ultimate compressive strain Tension strength Ultimate tension strain Elasticity modulus
f c′ = 34.3 MPa ft,k = 2.18 MPa
εc1 = 0.0022
Concrete #1 c.o.v. = 0.10 εcu = 0.0038 c.o.v. = 0.10 εct,u = 1.5E-4 Ec,m = 38,114 MPa
fc,m = 41.1 MPa ft,m = 2.62 MPa
f c′ = 39.2 MPa ft,k = 2.39 MPa
εc1 = 0.0022
Concrete #2 c.o.v. = 0.10
εcu = 0.0038
c.o.v. = 0.10 εct,u = 1.5E-4 Ec,m = 40,709 MPa
fc,m = 46.8 MPa ft,m = 2.86 MPa
fy,k = 499.8 MPa
Reinforcing steel c.o.v. = 0.05 εst,u = 0.01 Es = 205,800 MPa
fy,m = 544.6 MPa
fp,y = 1666 MPa
Prestressing steel Ep = Ep (εp)
fp,u = 1862 MPa
Note: 1 MPa = 0.145 ksi.

514 ACI Structural Journal/September-October 1996

 c. 3-span prestressing layout; the tendons are anchored at Δf ps ( characteristic ) = 110.7 MPa
every third pier diaphragm. (13)
= 16.05 ksi (confidence 95 percent)
Two load schemes were adopted. The first was chosen to
produce failure at midspan and the other one was chosen to c. For double T-section monolithic simply supported
produce failure at the pier section. Fig. 4 and 5 show both bridges, Δfps can best be represented by a lognormal distribu-
load schemes for a 2-span prestressing layout. tion with
Segmental bridges
Segmental bridges are usually built using the span-by- Δfps (average) = 139.1 MPa = 20.17 ksi (14)
span erection method. Therefore, the tendons are always
anchored at each span. Hence, only the 1-span prestressing standard deviation = 9.6 MPa = 1.39 ksi (15)
layout is considered.
For segmental bridges, two different positions of the devi-
ators have been studied. Layout Number 1 corresponds to a
distribution of deviators of 15 + 20 + 15 m (49.2 + 65.6 +
49.2 ft) (distance from end anchorage to first deviator, from
first to second deviator, and from second deviator to end
anchorage). Layout Number 2 corresponds to 20 + 10 + 20
m (65.6 + 32.8 + 65.6 ft).

Results
A summary of the results of the analyses performed in this
study are shown in Tables 5 through 12. The load process
adopted is as follows:
Load Level 1—Dead load + prestressing
Load Level 2—Dead load + prestressing + live load
Load Level 3—γDL dead load + γLL live load + prestressing
(1 < γDL = γLL < 1.5)
Load Level 4—1.5 dead load + γLL live load + prestressing Fig. 1—Longitudinal section for simply supported bridges,
(γLL > 1.5) showing span lengths and distribution of deviators (1 m =
Each load level is reached through several load steps. 3.28 ft; 1 kPa = 0.145 psi; 1 kN = 0.225 kip).
Simply supported bridges
From the results of simply supported bridges (Tables 5, 6,
and 7), it can be observed that the increase of tendon stress
does not change with the span length/depth ratio or with the
amount of reinforcing steel. Nevertheless, the increase of
tendon stress is affected by the cross section (box girder or
double T-section) and the construction procedure (monolithic
or segmental).
A statistical analysis using the Kolmogorov-Smirnov Test
was performed to determine the statistical distribution that
best represents the results. The results of the statistical anal-
ysis can be summarized as follows:
a. The distribution of Δfps for box girder monolithic simply
supported bridges follows a normal (Gauss) distribution with
a mean value Fig. 2—Cross section of box girder deck with 3.30 m
(10.83 ft) depth.
Δfps (average) = 120.5 MPa = 17.47 ksi (8)

and

standard deviation = 7.35 MPa = 1.06 ksi (9)

The characteristic value, defined by the Spanish Code as

the 95 percentile, would then be

Δf ps ( characteristic ) = 107.8 MPa (10)

= 15.63 ksi (confidence 95 percent)

b. For box girder segmental simply supported bridges, Δfps

can best be represented by a lognormal distribution with

Δfps (average) = 130.3 MPa = 18.9 ksi (11)

Fig. 3—Cross section of double T-section deck with 3.30 m
standard deviation = 12.6 MPa = 1.83 ksi (12) (10.83 ft) depth.

ACI Structural Journal/September-October 1996 515

Fig. 4—Load Scheme 1 for continuous box girder deck with 2-span prestressing layout (1 KPa = 0.145 psi; 1 kN = 0.225 kip).

Fig. 5—Load Scheme 2 for continuous box girder deck with 2-span prestressing layout (1 kPa = 0.145 psi; 1 kN = 0.225 kip).

Table 5—Monolithic simply supported box girder bridges: results from nonlinear analysis
Span, m Depth, m Prestressing As, cm2 Midspan deflection, cm Δ fps, MPa γLL
11.6 8.6 100.0 1.34
30 2.00 4 T 32 φ 0.6 in.
89.8 11.4 123.5 1.95
11.6 22.5 125.5 1.30
2.00 6 T 37 φ 0.6 in.
149 19.3 115.7 1.72
12.4 19.7 124.5 1.31
40 2.20 6 T 35 φ 0.6 in.
148 18.1 121.5 1.81
13.11 14.8 123.5 1.34
2.70 6 T 29 φ 0.6 in.
128 14.5 125.4 1.97
12.4 27 115.6 1.21
2.20 10 T 31 φ 0.6 in.
264 28.8 124.5 1.70
13.11 21.4 112.7 1.28
50 2.70 8 T 33 φ 0.6 in.
204 24.1 126.4 1.86
13.8 18.0 119.6 1.31
3.30 8 T 28 φ 0.6 in.
183 18.6 126.4 2.03
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 cm2 = 0.155 in.2; 1 MPa = 0.145 ksi.

Δf ps ( characteristic ) = 123.5 MPa of tendon stress for box girder continuous bridges, varying
(16) with the length of the tendon (tendon layout over 1, 2, or 3
= 17.9 ksi (confidence 95 percent)
spans) and the span-to-depth ratio. Fig. 6 shows the results
Continuous bridges for the minimum amount of reinforcement steel (As1) and
For continuous bridges, Δfps basically depends on the Fig. 7 shows the results obtained with the amount of rein-
length between tendon end anchorages and the span-to-depth forcement steel (As2) as previously described. There is no
ratio. Eq. (7) shows that Δfps is lower when the length of the significant influence of the amount of reinforcement steel,
tendon increases. This can be verified comparing any result even though with higher reinforcement (As2), Δfps is also
shown in Tables 8, 9, and 10. Fig. 6 and 7 show the increase higher due to the increase of the ductility of the bridge.

516 ACI Structural Journal/September-October 1996

Table 6—Segmental simply supported box girder bridges: results from nonlinear analysis
Span, m Depth, m Prestressing Midspan deflection, cm Δ fps, MPa gLL
6 T 28 φ 0.6 in. 14.8 147.0 2.10
30 2.00
6 T 25 φ 0.6 in 14.2 143.1 1.50
8 T 36 φ 0.6 in. 22.9 129.4 1.94
2.00
8 T 33 φ 0.6 in 22.9 128.4 1.50
8 T 34 φ 0.6 in. 20.0 174.0 1.98
40 2.20
8 T 31 φ 0.6 in 20.5 129.4 1.50
6 T 35 φ 0.6 in. 18.2 145.1 1.80
2.70
6 T 33 φ 0.6 in 14.3 121.5 1.47
10 T 43 φ 0.6 in. 30.7 115.7 1.90
2.20
10 T 40 φ 0.6 in. 27.5 104.9 1.52
10 T 35 φ 0.6 in. 25.6 125.5 2.10
50 2.70
10 T 32 φ 0.6 in. 23.5 122.5 1.55
10 T 29 φ 0.6 in. 21.6 139.2 2.16
3.30
10 T 27 φ 0.6 in. 23.8 148.0 1.72
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 MPa = 0.145 ksi.

Table 7—Monolithic simply supported double T-section bridges: results from nonlinear analysis
Span, m Depth, m Prestressing As, cm2 Midspan deflection, cm Δ fps, MPa γLL
35 105 1372 1.34
30 2.00 4 T 25 φ 0.6 in.
862 129 1568 1.94
35 205 1470 1.29
2.00 6 T 33 φ 0.6 in.
153 198 1480 1.76
34 192 1480 1.32
40 2.20 6 T 31 φ 0.6 in.
141 184 1490 1.85
36 131 1303 1.33
2.70 6 T 27 φ 0.6 in.
134 121 1303 1.98
51 292 1352 1.25
2.20 10 T 31 φ 0.6 in.
214 272 1313 1.48
36 209 1303 1.30
50 2.70 8 T 31 φ 0.6 in.
189 210 1372 1.85
38 160 1235 1.30
3.30 8 T 27 φ 0.6 in.
176 183 1421 2.03
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 cm = 0.155 in.2; 1 MPa = 0.145 ksi.
2

For segmental construction, there is no influence of the midspan sections when joint opening occurs. This happens
tendon layout because tendons are considered to always be with normal design before the pier section joints open,
anchored at each pier section. Assuming that Δfps follows a resulting in very different stiffnesses in both sections. For
lognormal distribution, the results can be summarized as instance, Fig. 8 shows the evolution of the bending moment
at midspan and at the supports for a monolithic box girder
Δfps (average) = 57.9 MPa = 8.4 ksi (17)
bridge with a 3.3-m (10.83-ft) depth.
The standard deviation is high, resulting in a low charac- Tendon layout influence
teristic value Table 12 shows the comparison between two tendon
layouts for the continuous segmental box girder bridges.
standard deviation = 13.8 MPa = 2.0 ksi (18) Layout Number 1 corresponds to a distribution of deviators
over the span of 15 + 20 + 15 m (49.2 + 65.6 + 49.2 ft), and
Δ f ps ( characteristic ) = 38.3 MPa (19) Layout Number 2 corresponds to 20 + 10 + 20 (65.6 + 32.8 +
= 5.55 ksi (confidence 95 percent) 65.6 ft).
Because the deviators in Layout Number 2 are closer to the
For both segmental and monolithic construction, the ulti- midspan area, the tendon produces a tendon stress increase
mate carrying capacity of the bridge is higher than that
Δfps higher than that obtained in Layout Number 1. The study
obtained from a linear analysis. This is caused by the fact
presented herein always uses Layout Number 1 and obtains
that the capability of bending moment redistribution of these
structures is significant. For monolithic structures, this conservative results.
capacity is due to the moment redistribution capacity of the
structure considering concrete and reinforcing steel PROPOSAL FOR CODE RECOMMENDATION
compressed by the prestressing. For segmental bridges, the From the results previously presented, simple values for
redistribution is due to the sharp loss of stiffness of the the increase of stress of the prestressing steel at ultimate Δfps

ACI Structural Journal/September-October 1996 517

Table 8—Continuous monolithic box girder bridges, tendons anchored at each on pier section: results
from nonlinear analysis
Span, m Depth, m Prestressing As, cm2 Load case Midspan deflection, cm Δ fps, MPa γLL
11.6 (span) Load #1 26.4 105.9 1.42
26 (support) Load #2 20.5 84.3 1.45
2.00 6 T 31 φ 0.6 in.
52 (span) Load #1 20.1 87.2 1.50
88 (support) Load #2 20.8 94.1 1.50
11.6 (span) Load #1 17.0 76.5 1.41
26 (support) Load #2 15.0 73.5 1.42
2.20 8 T 22 φ 0.6 in.
50 (span) Load #1 16.4 78.4 1.64
96 (support) Load #2 17.8 91.1 1.8
50
13.1 (span) Load #1 13.7 82.3 1.54
25.5 (support) Load #2 11.7 73.5 1.56
2.70 6 T 25 φ 0.6 in.
30 (span) Load #1 12.0 74.5 1.81
76 (support) Load #2 14.1 90.2 1.90
13.8 (span) Load #1 9.1 69.6 1.70
25 (support) Load #2 9.5 65.7 1.79
3.30 6 T 21 φ 0.6 in.
26 (span) Load #1 9.15 72.5 2.10
76 (support) Load #2 8.6 74.5 2.20
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 cm2 = 0.155 in.2; 1 MPa = 0.145 ksi.

Table 9—Continuous monolithic box girder bridges, tendons anchored every two on pier sections: results
from nonlinear analysis
Span, m Depth, m Prestressing As, cm2 Load case Midspan deflection, cm Δ fps, MPa γLL
11.6 (span) Load #1 22.6 50.0 1.37
26 (support) Load #2 19.9 51.0 1.39
2.00 6 T 31 φ 0.6 in.
52 (span) Load #1 19.4 43.1 1.50
88 (support) Load #2 19.3 53.9 1.61
11.6 (span) Load #1 15.6 35.3 1.38
26 (support) Load #2 15.6 45.1 1.40
2.20 8 T 22 φ 0.6 in.
50 (span) Load #1 16.8 41.2 1.58
96 (support) Load #2 15.3 46.1 1.65
50
13.1 (span) Load #1 11.2 33.3 1.47
25.5 (support) Load #2 12.0 45.1 1.50
2.70 6 T 25 φ 0.6 in.
30 (span) Load #1 13.3 43.1 1.80
76 (support) Load #2 13.0 52.0 1.90
13.8 (span) Load #1 9.1 36.3 1.67
25 (support) Load #2 9.6 45.1 1.70
3.30 6 T 21 φ 0.6 in.
26 (span) Load #1 9.3 37.2 1.80
76 (support) Load #2 8.3 36.3 2.03
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 cm2 = 0.155 in.2; 1 MPa = 0.145 ksi.

are obtained and can be incorporated into design specifications. Δfps = 39 MPa = 5.66 ksi (22)
These are summarized as follows:
a. For simply supported monolithic and segmental box These recommendations have been proposed by the
girder bridges, the Δfps recommended for design is Spanish Ministry of Public Works to the Permanent
Commission for Concrete to include them in the Spanish
Δfps = 108 MPa = 15.66 ksi (20) Design Code.

b. For simply supported double T-section monolithic COMPARISON WITH EXISTING CODES
It is interesting to compare the values of prestressing steel
bridges
stress increments given by current codes and those recom-
mended in this study.
Δfps = 122.5 MPa = 17.77 ksi (21)
A simply supported segmental bridge, spanning 50 m (164 ft)
with a box girder cross-section depth of 3.30 m [10.83 ft
c. For continuous monolithic box girder bridges, the incre- (Fig. 2)], was selected for comparison. The amount of
ment of tendon stress Δfps depends on the span-to-depth ratio prestressing is 10 T 27 φ 0.6 in. with the layout shown in Fig. 1.
and the prestressing layout, as shown in Fig. 9. The material properties used are shown in Table 4, corre-
d. For continuous segmental box girder bridges, sponding to Concrete Number 1. The effective stress in

518 ACI Structural Journal/September-October 1996

Table 10—Continuous monolithic box girder bridges, tendons anchored every three on pier sections:
results from nonlinear analysis
Span, m Depth, m Prestressing As, cm2 Load case Midspan deflection, cm Δ fps, MPa γLL
11.6 (span) Load #1 25.7 43.1 1.20
26 (support) Load #2 20.7 36.3 1.19
2.00 6 T 31 φ 0.6 in.
52 (span) Load #1 24.3 40.2 1.57
88 (support) Load #2 20.4 38.2 1.56
11.6 (span) Load #1 14.4 19.6 1.36
26 (support) Load #2 13.8 22.5 1.37
2.20 8 T 22 φ 0.6 in.
50 (span) Load #1 19.0 32.3 1.56
96 (support) Load #2 16.4 33.3 1.64
50
13.1 (span) Load #1 10.6 19.6 1.45
25.5 (support) Load #2 9.6 21.6 1.47
2.70 6 T 25 φ 0.6 in.
30 (span) Load #1 10.7 18.6 1.59
76 (support) Load #2 11.3 28.4 1.79
13.8 (span) Load #1 9.9 28.4 1.63
25 (support) Load #2 7.7 21.6 1.55
3.30 6 T 21 φ 0.6 in.
26 (span) Load #1 9.6 25.5 2.0
76 (support) Load #2 9.2 22.5 2.10
Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 cm2 = 0.155 in.2; 1 MPa = 0.145 ksi.

Table 11—Continuous segmental box girder

prestressed reinforcement is fpe = 1176 MPa = 170.55 ksi. For bridges: results from nonlinear analysis
this bridge configuration, different design codes recommend
Midspan
different values of Δfps. Below is a summary of these recom- Span, Depth, Load deflection, Δ fps,
mendations. m m Prestressing case cm MPa γLL
Load #1 14.3 48.0 1.75
2.00 8 T 28 φ 0.6 in.
ACI 318-89 Load #2 12.3 44.1 1.83
According to ACI 318-89 [see Eq. (2)], the bridge has a Load #1 14.4 57.8 1.85
span-to-depth ratio lower than 35, resulting in an increment 2.20 6 T 35 φ 0.6 in.
Load #2 13.0 58.8 1.97
of tendon stress (Δfps) of 412 MPa (59.75 ksi). 50
Load #1 10.6 51.9 1.82
2.70 6 T 28 φ 0.6 in.
Load #2 11.4 66.6 1.99
AASHTO (1994)
Adopting a Ωu coefficient corresponding to uniform load 3.30 6 T 24 φ 0.6 in.
Load #1 8.0 49.0 2.10
and using Eq. (3), the increment of tendon stress (Δfps) Load #2 11.8 87.2 2.48
obtained is 854.5 MPa (123.9 ksi). Since this value makes fps Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 MPa = 0.145 ksi.
higher than the maximum allowable value of 0.94 fpy , the
recommended Δfps is 390 MPa (56.56 ksi).
Table 12—Influence of prestressing layout on
13 continuous segmental box girder bridges: results
AASHTO (1989) from nonlinear analysis
The increment of tendon stress (Δfps) is 103 MPa (15 ksi).
This value is independent of span length, span-to-depth ratio, Midspan
Span, Depth, deflection, Δ fps,
tendon length, etc. m m Prestressing Layout cm MPa γLL
1 14.3 48.0 1.70
Eurocode N2 (Draft 1994)14 2.00 8 T 28 φ 0.6 in.
2 16.9 60.7 2.01
Unless a nonlinear analysis is performed, the increment of
1 14.4 57.8 1.85
tendon stress recommended is zero (Δfps = 0). 2.20 6 T 35 φ 0.6 in.
2 16.2 66.6 2.07
To study the effect of using different Δfps for the bridge 50
configuration chosen herein, they are compared with those 1 10.6 51.9 1.82
2.70 6 T 28 φ 0.6 in.
obtained from the nonlinear analysis of the bridge and from 2 13.0 70.6 2.05
the code proposal. Table 13 shows the different increments 3.30 6 T 24 φ 0.6 in.
1 8.0 49.0 2.10
of tendon stress recommended by the different codes, along 2 11.8 78.4 2.42
with the midspan deflection necessary to achieve them. For Note: 1 m = 3.28 ft; 1 cm = 0.39 in.; 1 MPa = 0.145 ksi.
the nonlinear analysis, free slipping has been assumed in the
deviators at ultimate. This assumption is consistent with the
results of laboratory tests, including the only full-scale shows the deflection obtained assuming a perfect hinge at
destructive test performed on a precast segmental box girder midspan for the different codes. The difference is due to the
bridge with dry joints and external tendons.12 The midspan neglect of the elastic deflection between midspan and the
deflection has been calculated assuming a perfect hinge at support when assuming all rotations concentrated at the
midspan and concentrating all the rotations at that hinge. The midspan hinge. The very high Δfps values given by the
deflection obtained from the nonlinear analysis is 23.8 cm AASHTO (1994) and ACI codes are due to ignoring the
(9.4 in.), but to be consistent in the comparison, Table 13 effect of the tendon lengths.

ACI Structural Journal/September-October 1996 519

Fig. 6—Tendon stress increment versus span-to-depth ratio
for continuous box girder decks with minimum reinforcement
(As1); results from nonlinear analysis (1 MPa = 0.145 ksi).
Fig. 8—Evolution of bending moment at midspan and at
supports for monolithic box girder bridge with 3.30 m
(10.83 ft) depth (1 kN-m = 8.85 kip-in.).

Fig. 7—Tendon stress increment versus span-to-depth ratio

for continuous box girder decks with reinforcement (As2); Fig. 9—Proposed values for Δfps for continuous monolithic
results from nonlinear analysis (1 MPa = 0.145 ksi). box girder bridges (1 MPa = 0.145 ksi).

The ACI 318-89 formula was obtained based on laboratory tendon eccentricity may be ignored without significantly
tests on simply supported beams, continuous beams, and affecting the results. This, however, may not be true for
continuous slabs prestressed with unbonded tendons. longer spans where high deflections produce large changes
Naaman and Alkhairi describe all the studies that lead to the in eccentricity.
formula.3 The tests were performed on short-span beam The AASHTO (1994) formula attempts to reduce the
models that were much shorter than normal bridge span analysis of beams prestressed with unbonded tendons to
lengths. Short tendons may produce high increases of tendon that of beams with bonded tendons through the use of strain
stress.11 In addition, the tests analyzed in Reference 3 were reduction coefficients.7 The value of the strain reduction
performed with internal unbonded tendons, neglecting the coefficient (Ωu) was determined using experimental tests
effect of loss of eccentricity of external tendons with high on 143 beam specimens. All the beams were partially or
deflections. For short-span beams, the effect of changes in fully prestressed with unbonded tendons, but they were

520 ACI Structural Journal/September-October 1996

Table 13—Code comparison CONCLUSIONS
Δfps, MPa
To obtain the tendon stress increment of externally
Midspan deflection, cm
prestressed bridges at ultimate, 102 load cases of 74 different
Finite element model 148.00 27.56
bridges have been analyzed with a nonlinear finite element
ACI 318-89 412.00 71.50 model. Simply supported and continuous bridges, box girder,
AASHTO 1994 390.00 68.06 and double T-sections with different amounts of reinforcing
AASHTO 1989 103.00 19.33 and prestressing steel have been studied, assuming mono-
Eurocode N2 0.00 0.00 lithic and segmental construction and several prestressing
Proposed value 108.00 20.25 layouts. Results are summarized in code recommendations
Note: 1 cm = 0.39 in.; 1 MPa = 0.145 ksi. that are compared with current bridge code recommendations.
Current codes give accurate values derived from
testing7,15,16 of the increase in tendon stress when applied to
unbonded prestressed slabs or laboratory unbonded internally
Table 14—Comparison of tendon stress increment, prestressed short beams, but these values are not always
MPa valid when applied to bridges.
FEM model The recommendations proposed in this paper are conser-
AASHTO Code
Tendon Test 1994 1* 2† proposal vative because: 1) they have been obtained assuming free
T1, lateral 176 390 12 223 111 slipping at the deviators; 2) they are based on characteristic
values corresponding to the 95 percentile of the values
T1, central 274 390 341 223 111
obtained from the analytical results; 3) for the analysis, code
T5 255 390 267 230 111 values4 for the ultimate strain of materials were adopted; and
T6 323 390 349 323 111 4) the tendon layout adopted is inefficient but appropriate
*No slipping at deviators for Tendon 1; free slipping for other tendons. when looking for a lower-bound of Δfps.
†Free slipping for all tendons.
Note: 1 MPa = 0.145 ksi.
ACKNOWLEDGMENTS
This paper is based on a research program sponsored by
internal to the concrete section and had very short span the Dirección General de Carreteras of the Spanish Ministry
lengths. The span-to-depth ratio of the tested beams ranged of Public Works, Contract C-1888, whose support is grate-
between 7.8 and 45. This range is too wide for an actual fully acknowledged.
bridge configuration, and a review of the practice of the
applicability of the results to bridge engineers is necessary. NOTATION
a = depth of rectangular stress block
A comparison between the AASHTO (1994) Code and the Ap1 = area of prestressing steel needed to achieve flexural strength,
recommendation made herein is performed for the only full- assuming Δfps = 100 MPa = 14.5 ksi
scale destructive test of a precast segmental box girder Ap2 = area of prestressing steel needed to achieve flexural strength,
bridge with dry joints and external tendons.12 The test span assuming Δfps = 100 MPa = 14.5 ksi
was designed in accordance with the specifications in Aps = area of prestressing steel
As = area of nonprestressed tensile steel
AASHTO (1983); results are summarized in Table 14. T1, As′ = area of nonprestressed compressive steel
T5, and T6 are tendons with different layouts. Among the six As1 = minimum reinforcement to control cracking due to shrinkage
different tendon layouts, only T1 did not slip freely at ulti- and temperature before prestressing
mate, but it did slip during the load process. For the nonlinear As2 = reinforcement needed to achieve flexural strength, assuming
analysis, slipping and free slipping of that tendon were not Δfps = 100 MPa = 14.5 ksi
b = width of section
considered. Results show very good agreement between the bw = width of web
test and the nonlinear analysis. Results also show that the c = depth from concrete extreme compressive fiber to neutral axis
AASHTO limitation of fps ≤ 0.94fpy governs the value of Δfps dp = distance from extreme compression fiber to centroid of pre-
for all the tendons analyzed. This indicates that the formula stressed reinforcement
for Δfps [Eq. (3)] may be too unconservative for normal Ep = modulus of elasticity of prestressing steel
fc′ = concrete compressive strength
bridge configurations. fpe = effective stress in prestressing steel
The low value of the code proposal compared with the test fps = ultimate stress in prestressing steel
result is due to: fpy = yield strength of prestressing steel
a. It is a characteristic value defined as the 95 percentile. fy = yield strength of nonprestressed tensile steel
b. It has been calculated with a different prestressing fy′ = yield strength of nonprestressed compressive steel
h = total depth
layout, one that is less efficient than the layout of the test. hf = thickness of flange
This assures us that it is a lower-bound of Δfp. L = span length
c. For the parametric study, code values for the ultimate L1 = length of loaded span or sum of length of loaded spans affected
strain of materials have been used.4 However, these mate- by the same tendon
rials achieve higher values of ultimate strains in laboratory L2 = length of tendon between end anchorages
Lt = length of tendon between no slipping points
tests, resulting in larger deflections, and, consequently, Mn = nominal moment strength
higher values of Δfps. This can be one of the reasons why the T = tendon
values of Δfps predicted by the AASHTO (1994) formula are β1 = ACI concrete compression block reduction factor
high. The bond reduction coefficient at ultimate nominal resis- εcu = strain in concrete top fiber at ultimate
tance Ωu was obtained from up to failure laboratory tests Δfps = stress increase of tendons beyond effective prestress
γDL = load factor for dead load
where the ultimate concrete and steel strains reached could be γLL = load factor for live load
higher than those given by the codes. More research appears Ωu = bond reduction coefficient at ultimate nominal resistance
necessary to adapt codes to actual high-strength concretes. ρp = ratio of prestressed reinforcement

ACI Structural Journal/September-October 1996 521

 REFERENCES 10. Fouré, B.; Rezendre, P. C.; and Hoang, L. H., “Problèmes de Sécurité
1. Virlogeux, M., “External Prestressing—Historical and Modern Appli- à la Rupture et de Modélisation du Comportement des Poutres en Béton à
cations,” External Prestressing in Structures, Association Francaise Pour la Précontrainte Extérieure (Behavior Modeling and Ultimate Safety Prob-
Construction (AFPC), Paris, 1993, pp. 13-42. lems of Externally Prestressed Concrete Beams),” Annales de l'ITBTP,
2. Corven, J., “Overview on External Prestressing in the World—Appli- Paris, Feb. 1991, pp. 46-95. (in French)
cation in North America,” External Prestressing in Structures, AFPC, 11. Aparicio, A. C.; Casas, J. R.; and Ramos, G., “Estudio Experimental
Paris, 1993, pp. 43-52. del Comportamiento en Servicio y Hasta Rotura de Estructuras de Hor-
3. Naaman, A., and Alkhairi, M., “Stress at Ultimate in Unbonded Post- migón con Pretensado no Convencional (Experimental Study of the Behavior,
Tensioning Tendons: Part 1—Evaluation of the State of the Art,” ACI In Service and Up to Failure, of Concrete Structures with External Pre-
Structural Journal, V. 88, No. 5, Sept.-Oct. 1991, pp. 641-651. stressing),” Final Report, Research Program PB90-0612, Education and
4. CEB-FIP, Model Code 1990, Comité Euro-International du Béton, Science Ministry of Spain, School of Civil Engineering, Barcelona,
Zurich, July 1991. Sept. 1994. (in Spanish)
5. ACI Committee 318, “Building Code Requirements for Reinforced 12. Takebayashi, T.; Deeprasertwong, K.; and Leung, Y. W., “Full-Scale
Concrete (ACI 318-89) (Revised 1992),” American Concrete Institute, Destructive Test of a Precast Segmental Box Girder Bridge with Dry Joints
Detroit, 1992, 343 pp. and External Tendons,” Proceedings of the Institution of Civil Engineers,
6. AASHTO, “AASHTO LRFD Bridge Design Specifications,” Ameri- Structures and Buildings, London, Aug. 1994, pp. 297-315.
can Association of State Highway and Transportation Officials, Washing- 13. AASHTO, “Standard Specifications for Highway Bridges,” Ameri-
ton, D.C., 1994. can Association of State Highway and Transportation Officials, Washington,
7. Naaman, A., and Alkhairi, M., “Stress at Ultimate in Unbonded Post- D.C., 1989.
Tensioning Tendons: Part 2—Proposed Methodology,” ACI Structural
Journal, V. 88, No. 6, Nov.- Dec. 1991, pp. 683-692. 14. CEN, Eurocode N2, ENV 1992-1-5, European Committee for Stan-
8. Ramos, G., “Estudio del Comportamiento en Servicio, Prerotura y dardization, Brusells, Oct. 1994.
Rotura de Puentes de Hormigón con Pretensado Exterior (Study of the 15. Yamazaki, J.; Kattula, B. T.; and Mattock, A. H., “Comparison of the
Behavior, In Service and at Ultimate, of Externally Prestressed Concrete Behavior of Post-Tensioned Prestressed Concrete Beams with and without
Bridges),” PhD dissertation, Technical University of Catalunya (UPC), Bond,” Report SM69-3, University of Washington, College of Engineering,
Civil Engineering School, Barcelona, Apr. 1994. (in Spanish) Structures, and Mechanics, Dec. 1969, 94 pp.
9. Ramos, G., and Aparicio, A. C., “Ultimate Analysis of Monolithic and 16. Mojtahedi, S., and Gamble, W. L., “Ultimate Steel Stresses in
Segmental Externally Prestressed Concrete Bridges,” ASCE Journal of Unbonded Prestressed Concrete,” Proceedings, ASCE, V. 104, ST7, July
Bridge Engineering, V. 1, No. 1, Feb. 1996. 1978, pp. 1159-1165.

522 ACI Structural Journal/September-October 1996