You are on page 1of 12

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 99-S54

Stresses in Unbonded Prestressing Tendons at Ultimate:


Recommendation
by Antoine E. Naaman, Ned Burns, Catherine French, William L. Gamble, and Alan H. Mattock

The use of unbonded tendons is common in prestressed concrete where fp = the stress in the prestressing tendon, in general, at
structures such as two-way slab systems, nuclear vessels, and the loading considered; and fpe = the effective prestress.
external prestressing. The analysis of members prestressed with The stress in the tendon for any level of loading beyond
unbonded tendons and subjected to loading up to their ultimate the reference state can be expressed as follows.
behavior is more difficult than that of those with bonded tendons.
Indeed, the stress increase in unbonded tendons under load, unlike For F + MD + ∆M,
bonded tendons, is member-dependent instead of section-dependent.
Thus, the stress in unbonded tendons depends on the deformation fp = fpe + ∆fps (2)
of the entire member and is assumed to be uniform at all sections. It
cannot be directly determined from the analysis of a beam cross where ∆fps is the stress increment in the tendons due to ∆M.
section by conventional strain compatibility, as can be done with Equation (2) also applies for any loading beyond MD up to and
bonded tendons. Rather, the stress increase in the tendons must be including the nominal bending resistance, in which case the
determined from the analysis of the loading-induced deformations stress in the prestressing tendons is termed fps by the ACI Code
of the entire structure. This is true for the elastic, inelastic, and
ultimate limit states. fps = fpe + ∆fps (3)
To determine the nominal moment resistance of beams pre-
stressed with unbonded tendons, and thus verify the ultimate- Most codes recommend prediction equations to estimate the
strength limit state as is required in most codes, there is need to value of fps from Eq. (3), which amounts to estimating as accu-
predict the stress fps in unbonded tendons at ultimate or nominal rately as possible the value of the stress increment ∆fps. In partic-
bending resistance. This can be done using nonlinear analysis and
numerical solutions techniques. However, a simplified method or a
ular, the ACI Code recommends a prediction equation originally
prediction equation for fps is needed for code purposes. Herein, the developed by Mattock, Yamazaki, and Kattula (1971) and later
authors, who are members of the Subcommittee on Stresses in modified to account for the effect of the span-depth ratio, based
Unbonded Tendons of Joint ACI-ASCE Committee 423, Prestressed on a study by Mojtahedi and Gamble (1978)
Concrete, recommend a procedure believed to represent the state
of the art and the best tradeoff at this time. f c′
fps = fpe + 10,000 + ---------
- psi (4)
k ρ ps
Keywords: bond; prestressed concrete; stress.
where k = 100 for L/dp ≤ 35, and k = 300 for L/dp > 35.
RESEARCH SIGNIFICANCE
Equations to predict the stress in unbonded tendons at REVIEW OF PREDICTION EQUATIONS FOR fps
nominal bending resistance of flexural members are given in To keep the length of this paper reasonable, prior investi-
various codes. They are based on past experience with steel gations cannot be reviewed in detail. Instead, the authors
tendons and do not account for a number of important param- refer the reader to the main references listed at the end.
eters, such as the use of fiber-reinforced polymeric (FRP) Therefor, the following provides only a brief summary.
tendons, partial prestressing, span-depth ratio, loading con- Numerous equations were developed to predict the stress
ditions, and loading patterns in continuous beams. The meth- fps at nominal bending resistance of beams prestressed with
odology and prediction equation proposed in this study attempt unbonded tendons. In an extensive study published in 1991,
to provide a rational solution to this problem and offer a 10 such equations were reviewed and evaluated (Naaman
recommendation for code implementation. and Alkhairi 1991a,b). They included those given in the ACI
Code (ACI Committee 318 2002), AASHTO (1996), Cana-
FUNDAMENTAL FORMULATION OF dian Code (CSA 1994), British Code (British Standard Insti-
PREDICTION EQUATION tution 1985), Swiss Code (1979), Dutch Code (1984) Dutch
To provide a rational procedure to predict the value of fps practice (1990), German Code (1980), French Code (1983),
at nominal bending resistance, a reference state of stress Mattock, Yamazaki, and Kattula (1971), Pannell (1969),
must be defined. Herein, the reference state is assumed to Tam and Pannell (1976), Harajli (1990), Du and Tao (1985),
represent the state of stress under the combined effect of Kordina, and Hegger (1987), Mojtahedi and Gamble (1978),
final or effective prestressing force F (after all prestress losses
have occurred) and dead load moment MD. ACI Structural Journal, V. 99, No. 4, July-August 2002.
Thus, under F + MD MS No. 01-376 received November 14, 2001, and reviewed under Institute publica-
tion policies. Copyright © 2002, American Concrete Institute. All rights reserved,
including the making of copies unless permission is obtained from the copyright pro-
fp = fpe (1) prietors. Pertinent discussion will be published in the May-June 2003 ACI Structural
Journal if received by January 1, 2003.

518 ACI Structural Journal/July-August 2002


Antoine E. Naaman, FACI, is a professor of civil engineering at the University of
Michigan, Ann Arbor, Mich. He is a member of ACI Committees 440, Fiber Reinforced
Polymer Reinforcement; 544, Fiber Reinforced Concrete; 549, Thin Reinforced
Cementitious Products and Ferrocement; and Joint ACI-ASCE Committee 423, Pre-
stressed Concrete. His research interests include high-performance fiber-reinforced
cement composites and prestressed concrete.

Ned Burns, FACI, is Zarrow Centennial Professor Emeritus at the University of Texas
at Austin, Austin, Tex., where he has been active in teaching and research since 1962.
He is a member of Joint ACI-ASCE Committee 423, Prestressed Concrete, and is also
a member of the National Academy of Engineering.

Catherine French, FACI, is a professor of civil engineering at the University of


Minnesota, Minneapolis, Minn. She is Chair of Joint ACI-ASCE Committee 445,
Shear and Torsion; and a member of ACI Committee 318, Structural Concrete Building
Code, and Joint ACI-ASCE Committee 352, Joints and Connections in Monolithic
Concrete Structures. Her research interests focus on the behavior of reinforced and
prestressed concrete structural systems, including the application of new materials,
retrofitting methods, and the development of new structural details.

William L. Gamble, FACI, is a professor of civil and environmental engineering at


the University of Illinois at Urbana-Champaign, where he has been a faculty member
since 1963. He is a member of Joint ACI-ASCE Committees 421, Design of Reinforced
Concrete Slabs; and 423, Prestressed Concrete; and ACI Committees 216, Fire Resis-
tance and Fire Protection of Structures; 543, Concrete Piles; and Subcommittee 318- Fig. 2—Correlation between predicted and experimental
F, Two-Way Slabs. His research interests include concrete bridges, reinforcing bar values of ∆fps used to determine Ωu from data (Naaman and
couplers, and concrete tunnel linings.
Alkhairi 1991b).
ACI Honorary Member Alan H. Mattock is Emeritus Professor of Civil Engineering
at the University of Washington, Seattle, Wash. He is a member of ACI Committee 318,
Structural Concrete Building Code; and Joint ACI-ASCE Committees 423, Pre-
stressed Concrete; 445, Shear and Torsion; and 550, Precast Concrete Structures. He
was the recipient of ACI’s Wason Medal in 1967, the ACI’s Alfred E. Lindau Award in
1970, and ACI’s Joe W. Kelly Award in 1991.

Fig. 3—Correlation between predicted and experimental


values of fps using reduced value of Ωu recommended for
code application in Eq. (14) (Naaman and Alkhairi 1991b).

Fig. 1—Correlation between predicted and experimental


values of fps used to determine Ωu from data (Naaman and
Alkhairi 1991b).

and Naaman and Alkhairi (1991b). The predictions given by


these equations were also compared with data from 143
beam tests carried out between 1962 and 1990 by 10 different
investigators around the world (see details in Naaman and
Alkhairi (1991a)). Particular attention was placed on com-
paring not only the predicted and observed values of fps but,
most importantly, the corresponding values of ∆ fps , where
the variability in the results is much more significant. The
equation developed by Naaman and Alkhairi (1991b) led to
the best correlation in all cases.
Figure 1 through 4, which are taken from Naaman and
Alkhairi (1991b), show the level of correlation for both fps
and ∆fps , where two values of the strain reduction coefficient
Ωu were used: one as obtained directly from the best correla- Fig. 4—Correlation between predicted and experimental
tion, and the other as recommended for code use to lead to values of fps using reduced value of Ωu recommended for
results on the safe side. code application in Eq. (14) (Naaman and Alkhairi 1991b).

ACI Structural Journal/July-August 2002 519


Since the aforementioned extensive study, three additional and Alkhairi (1991b) as the best candidate for a code rec-
equations known to the authors have been suggested: the ommendation. The following reasons are given for selecting
first is in a study by Lee, Moon, and Lim (1999); the second this equation. Most importantly, it offers a tradeoff among
is in a study by Allouche et al. (1998, 1999); and the third is accuracy, simplicity, and generality and, unlike any of the
in the 1998 edition of the AASHTO LRFD Specification for other equations reviewed, it applies to both steel and FRP
Bridge Design (1998) and MacGregor, Kreger, and Breen tendons. It is also very stable and remains valid over a wide
(1989). For the sake of brevity, these equations are not repro- range of parameters.
duced herein. The Subcommittee, however, gives as fol- For beams prestressed with bonded tendons, strains, and
lows the main reason these equations were evaluated but stress changes under load beyond the reference state Eq. (1)
not selected as its first recommendation. This does not imply can be determined in any typical section along the span using
that they cannot be used, but simply that they are less pre- equilibrium equations, stress-strain relations, and strain
ferred. Note that a prediction equation published the same compatibility. This is generally described as strain-compati-
year as Naaman and Alkhairi (1991b) by Harajli and Hijazi bility analysis. An underlying assumption of such analysis is
(1991) attempted to provide a unified formulation for both that perfect bond exists between steel and concrete, which
bonded and unbonded tendons; it was also reviewed but con- implies that the strain change under load in the steel is equal
sidered to be less desirable. to the strain change in the concrete at the level of the steel.
The equation proposed by Lee, Moon, and Lim (1999) Conventional strain compatibility does not apply as-is in
offered a combination of deformation compatibility based on computing the stress in unbonded tendons. This is because,
plastic hinge formation and regression analysis. It also repre- under loading, the strain change in the unbonded tendon is
sents an extension of Moon and Burns (1997). The authors averaged between end anchorages, and this is different from
carried out a correlation with essentially the same experi- the strain change in the concrete at the level of the tendon at
mental data used by Naaman and Alkhairi (1991a), and found any given section.
similar levels of correlation. Their equation, however, was con-
sidered less desirable by the committee because: 1) it did Strain reduction or bond reduction coefficient
not preserve the fundamental form of Eq. (3) (it was put in Baker (1949) was the first to suggest the use of a reduction
the form fps = 0.8fpe + ∆fps ); 2) it was significantly lengthier coefficient F to accommodate unbonded tendons. He defined
and used a coefficient f related to the length of plastic it as the: “(Average concrete stress adjacent to steel)/(Maxi-
hinge to be computed from another equation derived in mum concrete stress adjacent to steel),” and noted that “its
Harajli (1990), Harajli and Kanj (1991), and Harajli and value may be assumed to be 0.66 for beams supporting uni-
Hijazi (1991); and 3) it did not provide a significantly better formly distributed load, and not less than 0.5 for beams sup-
correlation than did the equation proposed by Naaman and porting central point load.” Baker also suggested that the
Alkhairi (1991b). coefficient F can be determined experimentally and, in a fol-
The equation suggested by Allouche et al. (1999) offered low-up paper (Baker 1951), he recommended a value of F =
a modification to the Canadian Code equation by introducing 0.1, based on an experimental test by Mattock (1949).
a multiplying factor (1 + cy /dp2 )that leads to a cubic equation Subsequent investigations by several researchers into the
in c, where c is the depth of the neutral axis; cy is the value ultimate resistance of beams prestressed with unbonded tendons
of c assuming yielding of the reinforcement; and dp is the led to a number of recommended values for F. From his tests,
distance from the extreme compression fiber to the centroid Revesz (1953) concluded that F = 0.0005/εcu. Basically, if εcu
of the prestressed reinforcement. No correlation with the ex- = 0.003, then F = 0.167. Gifford (1953, 1954) carried out tests
tensive body of experimental data was provided for simply where he directly measured the strain in unbonded steel wires
supported beams, and the use of the equation was suggested and recommended the use of F = n. Using the current ACI
as an improvement for continuous beams. The multiplier notation, the definition of n is: n = ρp(fpu /0.85fc′ ), where
could be reduced for all practical purposes to almost a con- 0.85fc′ is the average stress in the concrete stress block; thus,
stant, however. Thus, the equation can be shown to be really n is similar to the prestressing index and is expected to be less
insensitive to the level of continuity. Thus, it is not clear than approximately 0.30 for under-reinforced sections. Cowan
whether the improved predictions are due to the use of their (1955) and Tellerfield (1953) also published values of F versus
term le , which is related to the continuity of the member, or n ranging from 0.1 to 1. The factor F was also used by
simply to the use of the multiplier. Allouche et al. (1999) eval- Warwaruk, Sozen, and Siess (1962) in connection with the in-
uated only a handful of experimental data on continuous beams, crement of strain beyond that corresponding to fpe. The value
did not evaluate a large body of data on simply supported of F depended on the shape of the moment diagram as the depth
beams, and did not consider the span-depth ratio in their equa- of neutral axis c/dp. Pannell (1969) reviewed most of the above
tion. The Committee believed that any adopted equation should investigations and further clarified the definition of F as “the
also be particularly accurate for simply supported beams. ratio of steel strain to concrete strain at the level of the steel at
Finally, the 1998 AASHTO-LRFD equation leads to very ultimate load in unbonded prestressed concrete beams.”
conservative values of ∆fps that cannot be matched by the For all practical purposes, the coefficient F has the same
large body of experimental observations. The equation was basic definition as the coefficient Ωu , used in this study and
initially derived by MacGregor, Kreger, and Breen (1989) as expanded upon subsequently, except that Ωu uses increments
part of an investigation on a three-span, externally post-ten- of strains beyond a reference state defined as that corre-
sioned box girder bridge model. It is based on a prior equa- sponding to the effective prestress and dead load.
tion developed by Tam and Pannell (1976) and modified for Following the same general definition of the strain-reduction
continuous members by accounting for the number of poten- coefficient, Naaman (1990) was the first to develop analytical
tial plastic hinges forming at failure. expressions for the coefficient that apply to the elastic un-
The aforementioned, very brief review of more recent cracked, elastic cracked, and ultimate limit states in bending.
studies left the basic equation initially proposed by Naaman He integrated these coefficients in a modified strain-com-

520 ACI Structural Journal/July-August 2002


patibility analysis to predict the moment curvature of beams
prestressed with unbonded tendons. The solution proposed
essentially reduces the analysis of beams with unbonded
tendons to that of beams with bonded tendons through the use
of appropriate strain reduction (or bond reduction) coefficient
(Naaman 1990; Naaman and Alkhairi 1991b). Equivalently, the
coefficient reduces a deflection-compatibility analysis to a
strain-compatibility analysis (Naaman 1991). The strain-
reduction coefficient was defined as the ratio of the average
strain increment in the unbonded tendon to the strain increment
in the equivalent bonded tendon at the section of maximum
moment. Its derivation can be summarized as follows:
Simply supported beams with constant EIg throughout their
lengths, symmetrical loadings, and tendon profiles are consid-
ered. The critical design section is assumed to be the midspan
section, at which the maximum eccentricity of the tendons is
assigned. It can then be shown that the strain-reduction coeffi-
cient can be calculated in the most general case from (Naaman
1990; Naaman and Alkhairi 1991b)

2 L⁄2
Ω = --------------------------------------
∆ M max ( e o ) max L ∫0 ∆ M ( x ) e o ( x ) dx (5) Fig. 5—Strain distribution along section of maximum moment.

Naaman and Alkhairi attempted to extend the theoretical


Lc ⁄ 2
derivation of the strain-reduction coefficient to the ultimate
I cr 2  I cr ∆ M ( x ) e o ( x ) dx
Ω c = Ω -----
- + --- 1 – -----
Ig L 
-
Ig  ∫ ------------------------------------
∆ M max ( e o ) max
(6) limit state in bending; they defined the coefficient as Ωu and
tried to develop an equation for it that is of the same form as
0
Eq. (6) and (7). They worked in particular on the simplified
where Ω = for the case where the section is uncracked; Ωc = form of Eq. (7), where Lc becomes the plastic hinge length at
for the case where the section is cracked; eo (x) = eccentricity ultimate, and Icr becomes the moment of inertia of the
of the centroid of the tendons; (eo)max = eccentricity at the cracked section at ultimate; however, the task proved to be
midspan section; Ig = the moment of inertia of the uncracked extremely challenging and no satisfactory solution could be
section; found in terms of consistently fitting all available experi-
Icr = the moment of inertia of the cracked section; ∆M(x) = mental data. Hence, they analyzed 143 beam tests from ten
the moment in excess of dead-load moment; ∆Mmax = the different investigations spanning from 1962 to 1990 and
value of ∆M(x) at the midspan section; and x = the distance back-calculated a strain-reduction coefficient Ωu from the
from midspan to the section considered. data (Fig. 1 and 2). The coefficient accounted for two variables,
L is the span length, and Lc is the length of the segment of loading condition (midspan, third-point, or uniform loading),
the beam assumed to be cracked. Lc is generally very small and span-depth ratio. Naaman and Alkhairi (1991b) integrated
compared with L because, theoretically, a plastic hinge the value of Ωu in a strain compatibility analysis according to
forms at the cracked section, and the lack of bond does not Fig. 5, and showed that the relations obtained led to the
allow multiple cracking to develop beyond that region. Lc highest correlation (among existing prediction equations)
remains relatively small compared with L even when internal between predicted and experimental observations, not only for
bonded reinforcement is provided to produce well-distributed the values of fps, but also for the value of ∆fps. They then
cracks, as recommended in the ACI Building Code (1999, reduced the values of Ωu obtained by a safe margin (Fig. 3 and
2002) for slabs. Closed-form numerical solutions for Ω and 4), for the purpose of inclusion in a code recommendation and
Ωc for different cases of loading and tendon profiles are given developed several examples to illustrate the accuracy and
and discussed in Naaman (1990) and Naaman and Alkhairi validity of their method.
(1991b). They are very simple, especially for Ω. In particular,
they show that the strain-reduction coefficients depend only Prediction equation for fps
on the tendon profile and the applied load; moreover, their Based on their findings, Naaman and Alkhairi (1991b)
numerical value is quite sensitive to whether the eccentricity at proposed the following prediction equation for the stress in
the supports is negative, positive, or nil. the unbonded prestressing tendon at nominal bending resistance
It was also shown that, for the closed-form solution obtained
for Ωc , the following approximation holds in all cases L
fps = fpe + ΩuEpsεcu  ----p- – 1 -----1
d
(8)
(Naaman 1990, 1991; Naaman and Alkhairi 1991b) c  L2

Icr  I cr L c I cr
Ω c ≈ Ω ----- - ----- ≈ Ω -----
- + 1 – -----

- (7) where Eps = elastic modulus of prestressing steel; L = length
Ig Ig L Ig of span for which computation is carried out; L1 = the sum of
lengths of loaded spans containing tendon(s) considered; L2 =
Whether using the first or second approximation of the total length of tendon(s) between anchorages; εcu = assumed
right-hand term, Eq. (7) leads to a greatly simplified analysis failure strain of concrete in compression equal to 0.003; and
for the elastic cracked range of behavior of a member pre- Ωu = 3/(L/dp) for uniform or third-point loading, and 1.5/(L/dp)
stressed with unbonded tendons. for one-point midspan loading.

ACI Structural Journal/July-August 2002 521


The strain-reduction coefficient Ωu is defined as the ratio modulus of prestressing tendons (steel or FRP) as obtained
of strain increase in the unbonded tendon to that of the equiv- from tests or specified by the manufacturer. The value of Ep for
alent bonded tendon at the section of maximum moment, at prestressing steels may be taken as 28,000 ksi (196,000 MPa).
nominal bending resistance. When Eq. (8) is applied at inter- For simply supported beams or slabs prestressed with un-
mediate supports of continuous members, L can be taken as bonded steel tendons, Eq. (10b) leads to an increment of
the average of the two spans on either side of the support. stress of 15 ksi (105 MPa). This value has been extensively
Equation (8) can be derived directly from the strain diagram used in the past for its simplicity, and no adverse effects were
in Fig. 5, neglecting the effect of εce and assuming that the reported or are known to the Subcommittee.
stress in the prestressing steel is linear elastic; that is, Because of the ratio L1 /L2, the increment of stress ∆fps can
become very small in a multispan system. Experimental tests
fps = Epsεps (9) by Burns (1990) suggest that ∆fps does not vanish, due to
some friction. Thus, for one- or two-way slab systems with
Modifications and extension three spans or more, the subcommittee recommends the use
Because Eq. (8) is valid only when the stress in the pre- of Eq. (10) with an increment of stress limited as follows
stressing steel is in the elastic range of behavior, a limit of
0.80fpu is suggested. The limit represents 0.94fpy for stress- Ep
relieved strands, and 0.89fpy for low-relaxation strands. It is ∆ f ps ≥ -----------
- (11)
3732
consistent with Eq. (9).
Equation (8) is valid for simple spans where the tendon For prestressing steels, Eq. (11) implies an increment of stress
length between anchorages is the same as the span length. not less than 7.5 ksi (52.5 MPa); this limit is also recommended
When continuous tendons are used over several spans, the as a lower limit for all members with prestressing steel.
loading of one span creates deformations in that span leading Integrating Eq. (10) and (11) into a single equations leads to
to a strain increment in the tendons that is averaged between
the anchorages, that is, over several spans. To account for Ep L1 Ep
this effect, Eq. (8) was modified by adding the factor L1/L2, - ----- ≥ f pe + -----------
f ps = f pe + ----------- - (12)
1866 L 2 3732
where L1 is the length of the loaded span (or, more generally,
the sum of the lengths of loaded spans containing or crossed
Eq. (12) is recommended herein for code use. It applies to
by tendon[s] considered), and L2 is the length of the tendon
steel or FRP tendons and is independent of the system of
between anchorages. A discussion related to the use of L1/L2
units. If Eq. (12) is written for steel tendons and in U.S. units,
and the influence of the pattern and type of loading is given
it will read as follows
in Appendix B.
Because Eq. (8) is valid for the linear-elastic range of be-
L
havior of the reinforcement, it also applies to FRP tendons, f ps = f pe + 15 ----1- ≥ f pe + 7.5 ksi (13)
which are known to have a linear-elastic stress-strain response L2
up to failure. Thus, for FRP tendons, Eq. (8) will be valid
provided that the modulus of elasticity of the steel is replaced Refined approach
by that of the FRP tendon. The upper limit of 0.80fpu seems Steel tendons—
reasonable for safety reasons, even for FRP tendons, but
could be modified pending further calibration with FRP L
f ps = f pe + Ω u E ps ε cu  ----p- – 1 ----1-
d
(14)
materials data. c  L2

RECOMMENDED EQUATIONS TO REPLACE FRP tendons—


CODE SECTION 18.7.2(b)
The ACI 423 Subcommittee recognizes that the stress in-
L
f frp = f pe + Ω u E frp ε cu  ----p- – 1 -----1
crement in unbonded tendons is generally small and, often, d
(15)
the need for an accurate prediction equation may not be c  L2
warranted. Because a member-strain compatibility or a
deflection compatibility analysis can be time-consuming, Steel or FRP or any linear elastic reinforcement—If the
two equations are recommended subsequently as alterna- stress in the prestressing tendon (made of steel or FRP or any
tives: 1) a conservative approximation sufficient for slabs other material with linear elastic behavior) where nominal
and numerous other cases; and 2) a more refined prediction bending resistance is defined as fpn, and if the modulus of
equation applicable to all cases. For unbonded steel tendons, elasticity of the tendon is defined as Epr, then Eq. (14) and
a minimum stress increment limit of 7.5 ksi (52.5 MPa) is (15) can be represented by the same equation
also recommended for any span or loading case.
d L
f pn = f pe + Ω u E pr ε cu  ----p- – 1 -----1 (16)
Conservative approach c  L2
fps = fpe + ∆fps (10a)
where the strain-reduction coefficient Ωu is as given in Eq. (8).
Ep L1
∆fps = -----------
- ----- (10b) Use of recommended equation
1866 L 2 To obtain the value of fpn from Eq. (16), the equation of
force equilibrium in the section at ultimate and the prediction
where ∆fps is the increment of stress from the effective equation of fpn (Eq. (16)) must be solved simultaneously for
prestress level to nominal resistance; and Ep is the elastic c and fpn. A quadratic equation in c is obtained and solved

522 ACI Structural Journal/July-August 2002


first; then fpn is computed. A closed-form solution and some Proposed change to ACI Code Eq. (18-4) and (18-5)
detailed examples are given in Naaman and Alkhairi (1991b) The current ACI Code equation to predict the stress fps in
and Naaman (1995). It should be noted that Eq. (14) was ini- unbonded prestressing tendons is given in Eq. (3) and is still
tially adopted in the first draft of the AASHTO-LRFD Spec- useful in many cases. There are, however, several key rea-
ification for Bridge Design (1995), and then was replaced by sons to replace the code equation by Eq. (16), namely:
a new equation. Equation (14) through (16) are nondimen- • To accommodate partial prestressing; that is, the com-
sional; thus, they apply to all systems of units. Should only bined presence of reinforcing and prestressing steels;
steel be considered, it would be easy to further simplify Eq. (14) • To accommodate both prestressing steel and FRP ten-
and (16) by replacing the product Epr εcu with its numerical dons, which are increasingly considered in prestressed
value. Several examples illustrating the computation procedure concrete applications;
using Eq. (16) for beams and slabs are given in Appendix A. • To account for the effect of the elastic modulus of the
prestressing tendons used;
Advantages of proposed equation • To provide a more accurate formulation based on
First and foremost, the proposed prediction equation for deflection compatibility;
fps (or fpn ) is rational and based on a fundamental analysis • To incorporate the span-depth ratio, the tendon profile,
where equilibrium equations, stress-strain relations, and and the loading conditions in a more rational way; and
compatibility of deformations are satisfied. It is stable over a • To account for the effect of loaded span or spans, versus
wide range of variables. Equation (8) and (14) gave the high- length of tendon between anchorages.
est levels of correlation with experimental data using steel The proposed change was put in a code format in Table 1
tendons and predicted the trends for most important vari- to replace Section 18.7 of the ACI Building Code. The notation
ables. The value of the depth of the neutral axis c at ultimate of the code was used; however, the definitions of some terms
accounts for the presence of nonprestressed reinforcement need to be clarified to include prestressing reinforcement
(tensile and compression) and satisfies equilibrium at ulti- other than steel reinforcement. Namely, Ep can be defined as
mate for rectangular and T-section behavior. Other features the modulus of elasticity of the prestressed reinforcement to
are listed as follows: include steel or FRP.
1. Because it is based on compatibility and equilibrium,
Eq. (16) incorporates the elastic modulus of the tendons and EXAMPLES
is applicable to both prestressing steels and FRP tendons. The examples described in Appendix A are by no means
Equation (16) is currently needed when FRP tendons are exhaustive and are simply offered to provide typical exam-
used; it is particularly suitable because FRP tendons have an ples of solutions for the case of a simply supported beam and
elastic modulus spanning a very wide range, from one-tenth a continuous slab. Six other examples are developed in detail
(such as that of PVA) to more than twice (in the case of some in Naaman and Alkhairi (1991b). The closed-form solutions
grades of carbon) that of steel; of the two equations (Eq. (16) and the equation for force
2. Although the strain reduction coefficient Ωu was cali- equilibrium) for the case of rectangular and T-section behavior
brated based on tests of 143 simply supported beams pre- are given in Naaman and Alkhairi (1991b) and Naaman
stressed with unbonded steel tendons (Naaman and Alkhairi (1995). These are, essentially, the solutions of a quadratic
1991b), it should be applicable in principle to FRP tendons equation. A simple iterative procedure, however, can be fol-
as well because, by definition, the strain-reduction coeffi- lowed, for instance: 1) assume fps = fpe + 15 ksi (105 MPa)
cient (Eq. (5) and (6)) theoretically depends only on the ten- (or assume fps = 0.8fpu), and then compute c from equilibri-
don profile and the external loading conditions. Future um; 2) using the c obtained, compute fps from Eq. (14); and
modifications, if necessary, can be accommodated through 3) repeat the procedure until there is little gain in the itera-
the strain-reduction coefficient without changing the funda- tion. The examples in Appendix A illustrate the sensitivity of
mental nature of Eq. (16). For example, Ng and Tan (2002) the proposed equation in cases where reinforcing bars are
have recommended some adjustment to Ωu to accommodate used in addition to prestressing tendons, in case of FRP ten-
the effects of eccentricity variations in external prestressing; dons, in cases where one or two versus several spans are
3. Through the strain reduction coefficient Ωu , Eq. (16) loaded, and the case of the span-depth ratio.
also accommodates the span-depth ratio; The possible interpretation of the ratio L1/L2 used in Eq. (14)
4. Equation (16) takes into consideration the influence of through (16) was the subject of a long discussion by the
loaded span versus nonloaded span or spans through the ratio Subcommittee. Appendix B clarifies the influence of pattern-
L1/L2; and loading and type of load on the value of that ratio. It is pro-
5. Equation (16) may lead to prediction values of fpn sig- vided for completeness of information but, for all practical
nificantly higher than those obtained from the conservative purposes, should not affect the definition of L1/L2 used in
approach (Eq. (10) through (13)), resulting in higher values Eq. (8), (12), (14), (15), and (16).
of nominal bending resistance and potential cost savings.
It should be noted that, within a given span of a multiple-span CONCLUDING REMARKS
member, it would be adequate to compute fpn from Eq. (16) at The Subcommittee recognizes that, in unbonded prestressing
the critical sections, then use an average value for the loading tendons, the increment of stress from the effective prestress
condition selected. Equation (10) through (16) depend on the to ultimate or nominal resistance is generally small when
effective prestress as the reference stress fpe. It should be not- compared with that of bonded tendons and also when com-
ed that fpe is itself section-dependent as the section moves pared with the effective prestress itself. In many cases, an
from jacking end to holding end; thus, fpe can be different at approximation fps to estimate the stress at ultimate is sufficient,
different critical sections, and this may have a bearing on the such as in Eq. (13), where a maximum stress increment of
value of nominal bending resistance. The effect, however, is 15 ksi (105 MPa) is used for prestressing steel. The recom-
small and is often neglected in practice. mended equation (Eq. (16)) to predict fps , however, permits

ACI Structural Journal/July-August 2002 523


Table 1—Summary of code recommendation*
Recommended equations to replace ACI Code, Section 18.7.2(b) Remarks/Commentary
For members prestressed with unbonded tendons, instead of a deflection compat- Eq. (18-4) and (18-5) hold for all systems of units.
ibility analysis, one of the following approaches can be used:
When not obtained from test or supplied by the manufacturer, the elastic
1. Conservative approximate prediction equation modulus of prestressing steel may be taken as 28,000 ksi (196,000 MPa).

Ep L1 Ep Applied to prestressing steel and U.S. units, Eq. (18-4) can be written as
- ----- ≥ f se + -----------
f ps = f se + ----------- - (18-4)
1866 L 2 3732

where fps is the stress in the prestressed reinforcement at nominal bending resis- L
f ps = f se + 15 -----1 ≥ f se + 7.5 ksi
tance; other notation is as per Eq. (18-5). L2
2. Accurate prediction equation
It leads to an increment of stress beyond effective prestress equal to a
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1 ≤ 0.80f pu
d L
(18-5) maximum of 15 ksi (105 MPa) and a minimum of 7.5 ksi (52.5 MPa).
c L2
Equation (18-5) is based on deflection compatibility analysis along the
where member that is reduced to a section strain compatibility through the
strain reduction coefficient Ωu.
fse = effective prestress in the prestressed reinforcement;
Ep = elastic modulus of prestressed reinforcement; Equation (18-5) is derived assuming the tendons are in the linear range of
εcu = failure strain of concrete in compression, assumed = 0.003; behavior. The limitation of 0.80fpu corresponds to 94% of yield for stress
dp = distance from extreme compression fiber to centroid of prestressed reinforcement; relieved steel strand, and 89% of yield for low relaxation strand. It is also a
c = depth of neutral axis at nominal bending resistance; safe limit for FRP tendons.
L = length of span for which computation is carried out;
L1 = sum of lengths of loaded spans containing tendon(s) considered; The value of depth of neutral axis c at ultimate accounts for the presence
of nonprestressed reinforcement and satisfies equilibrium at ultimate.
L2 = total length of tendon(s) between anchorages; The equation of force equilibrium in the section and the prediction equa-
Ωu = 3/(L/dp) for uniform or third-point loading; tion of fps must be solved simultaneously for c and fps.
Ωu = 1.5/(L/dp) for one-point midspan loading; and
fpu = specified tensile strength of prestressed tendons. At intermediate supports of continuous members, L can be taken as the
average of the two spans on either side of the support.
*
ACI Code notation is used when available.

taking advantage of stress increments larger than 15 ksi British Standard Institution, 1985, “Structural Use of Concrete,” BS
(105 MPa) when justifiable. The proposed equation accounts 8110, Section 4.3.7.3, BSI, London.
Burns, N. H., 1990, “Post-Tension Force Changes in Continuous Beams,”
for a number of parameters not currently covered in any ASCE Structures Congress Abstracts, ASCE, New York, pp. 455-456.
other code. It is rational; that is, it is based on theory and Cowan, H. J., 1955, “The Ultimate Strength of Prestressed Concrete
supported by experimental validation—it accommodates the Beams,” The Structural Engineer, V. 33, No. 7, July, pp. 197-212.
combined presence of reinforcing steel and prestressing, the CSA (Canadian Standards Association) A23.3-94, 1994, “Design of
span-depth ratio, nonmetallic reinforcements, and differences in Concrete Structures,” Rexdale, Ontario, Canada.
Du, G., and Tao, X., 1985, “Ultimate Stress in Unbonded Tendons of
load patterns. Moreover, it provides a theoretical reference Partially Prestressed Concrete Beams,” PCI Journal, V. 30, No. 6, Nov.-
measure in cases where unusual conditions are encountered or Dec., pp. 72-91.
trends are to be estimated. The recommendation summarized Dutch Code, 1984, “Voorschriften Beton VB,” NEN 3880, Part H,
in Table 1 represents the Subcommittee's consensus given Section 503.1.3.
Dutch practice, 1990, private communication with A. S. G. Bruggeling.
the current state of knowledge. “French Code for Limit State Design of Prestressed Concrete,” 1983,
Regles B.P.E.L 83, Fascicule No. 62, Titre 1, Section 2, Marche Publics de
REFERENCES Travaux, Secretariat General de la Commission Centrale des Marches,
AASHTO, 1996, “Standard Specifications for Highway Bridges,” 16th Paris, France, Oct. (in French)
Edition, American Association of Highways and Transportation Officials, German Code, 1980, “Spannbeton, Bauteile mit Vorspannung ohne
Transportation Research Board, Washington D.C. Verbund,” (Entwurf), DIN 4227, Teil 6. (in German)
AASHTO LRFD Specification for Highway Bridge Design, 1995, Gifford, F. W., 1953, “Test of End-Anchored, Unbonded Prestressed
American Association of State Highway and Transportation Officials, Concrete Beams Having Parabolic Steel Eccentricity, Subjected to Uni-
Washington, D.C. formly Distributed Loading,” Magazine of Concrete Research, V. 5, No.
ACI Committee 318, 1999, “Building Code Requirements for Structural 13, Aug., pp. 27-36.
Concrete (ACI 318-99) and Commentary (318R-99),” American Concrete Gifford, F. W., 1954, “The Design of Simply Supported Prestressed Con-
Institute, Farmington Hills, Mich., 391 pp. crete Beams for Ultimate Loads,” Proceedings of the Institution of Civil
ACI Committee 318, 2002, “Building Code Requirements for Structural Engineers, Part III, V. 3, No. 1, Apr., pp. 125-143.
Concrete (ACI 318-02) and Commentary (318R-02),” American Concrete Harajli, M. H., 1990, “Stress in Unbonded Prestressed Members,” ACI
Institute, Farmington Hills, 443 pp. Structural Journal, V. 87, No. 3, May-June, pp. 305-312.
Allouche, E. N., Campbell, I. T., Green, M. F., and Soudki, K. A., Harajli, M. H., and Kanj, M. Y., 1991, “Ultimate Flexural Strength of
1998, “Tendon Stress in Continuous Unbonded Prestressed Concrete Concrete Members Prestressed with Unbonded Tendons,” ACI Structural
Members—Part 1: Review of Literature,” PCI Journal, V. 43, No. 6, Journal, V. 88, No. 6, Nov.-Dec., pp. 638-673.
Nov.-Dec., pp. 86-93. Harajli, M. H., and Hijazi, S., 1991, “Evaluation of the Ultimate Steel
Allouche, E. N.; Campbell, I. T.; Green, M. F.; and Soudki, K. A., 1999, Stress in Unbonded Prestressed Members,” PCI Journal, V. 36, No. 1, Jan.-
“Tendon Stress in Continuous Unbonded Prestressed Concrete Members— Feb., pp. 62-82.
Part 2: Parametric Study,” PCI Journal, V. 44, No. 1, Jan.-Feb., pp. 60-73. Kordina, K., and Hegger, J., 1987, “Determination of the Ultimate
Baker, A. L. L., 1949, “A Plastic Theory of Design for Ordinary Strength in Bending in the Case of Prestress Without Bond,” Beton—Und
Reinforced and Prestressed Concrete Including Moment Redistribution in Stahlbetonbau, Wilhelm Ernst + Sohn, Berlin, V. 82, Apr., pp. 85-90.
Continuous Members,” Magazine of Concrete Research (London), V. 1, (in German)
No. 2, June, pp. 57-66. Lee, L.-H.; Moon, J.-H.; and Lim, J.-H., 1999, “Proposed Methodology
Baker, A. L. L., 1951, “Recent Research in Reinforced Concrete and its for Computing of Unbonded Tendon Stress at Flexural Failure,” ACI Struc-
Application to Design,” Journal, Institution of Civil Engineers (London), tural Journal, V. 96, No. 6, Nov.-Dec., pp. 1040-1048.
V. 35, No. 4, Feb., pp. 262-329. MacGregor, R. J. G.; Kreger, M. E.; and Breen, J. E., 1989, “Evaluation

524 ACI Structural Journal/July-August 2002


of Strength and Ductility of a Three-Span Externally Post-Tensioned Box
Girder Bridge Model,” Research Report 365-3F, Center for Transportation
Research, University of Texas at Austin, Austin, Tex., Jan.
Mattock, A. H., 1949, “A Study of Design Methods for ‘End Anchored’
Prestressed Concrete Beams,” thesis submitted for MSc in Engineering,
University of London, June, 165 pp.
Mattock, A. H.; Yamazaki, J.; and Kattula, B. T., 1971, “Comparative
Study of Prestressed Concrete Beams, with and without Bond,” ACI JOURNAL,
Proceedings V. 68, No. 3, Feb., pp. 116-125.
Mojtahedi, S., and Gamble, W. L., 1978, “Ultimate Steel Stresses in
Unbonded Prestressed Concrete,” ASCE Structural Journal, July, pp.
1159-1165.
Moon, J.-H., and Burns, N. H., 1997, “Flexural Behavior of Members Fig. A1—Beam section and properties considered.
with Unbonded Tendons. I: Theory, II: Applications,” Journal of Structural
Engineering, ASCE, V. 123, No. 8, Aug., pp. 1087-1094. Eq. (18-4) or (18-5) and the proposed equations of Table 1
Naaman, A. E., 1982, Prestressed Concrete Analysis and Design: Fun- for the various cases shown as follows.
damentals, McGraw Hill, New York.
Naaman, A. E., 1990, “A New Methodology for the Analysis of Beams
Prestressed with External or Unbonded Tendons,” External Prestressing in Steel tendons: 1999 ACI Code
Bridges, SP-120, A. E. Naaman and J. E. Breen, eds., American Concrete The following can be computed
Institute, Farmington Hills, Mich., pp. 339-354.
Naaman, A. E., 1991, “Stress at Ultimate in Unbonded Tendons by A ps 0.612
Strain Compatibility,” Progress in Structural Engineering, D. E. Grierson, ρ p = -------- = ------------------ = 0.001417
A. Franchi, and P. Riva, eds., Kluwer Academic Publishers, pp. 217-227. bd p 24 × 18
Naaman, A. E., 1992, “Unified Design Recommendations for Reinforced,
Prestressed, and Partially Prestressed Concrete Bending and Compression
L 40 × 12
Members,” ACI Structural Journal, V. 89, No. 2, Mar.-Apr., pp. 200-210. ----- = ------------------ = 26.7 ≤ 35
Naaman, A. E., 1995, “Bending Strength Design of Prestressed and Partially dp 18
Prestressed Concrete Members with the New AASHTO Code,” Journal of
Structural Engineering, ASCE, V. 121, No. 6, June, pp. 964-970.
Naaman, A. E., and Alkhairi, F. M., 1991a “Stress at Ultimate in f c′
fps = fse + 10 + --------------------
Unbonded Prestressing Tendons—Part I: Evaluation of the State-of-the- 100 × ρ p
Art,” ACI Structural Journal, V. 89, No. 5, Sept.-Oct., pp. 641-651.
Naaman, A. E., and Alkhairi, F. M., 1991b, “Stress at Ultimate in
Unbonded Prestressing Tendons—Part II: Proposed Methodology,” ACI 7
Structural Journal, V. 89, No. 6, Nov.-Dec., pp. 683-692. = 150 + 10 + ---------------------------------- = 209.4 ksi
Ng, C. K., and Tan, K. H., 2002, “Evaluation of Tendon Stress in Exter-
100 × 0.00142
nally Prestressed Beams at Ultimate Flexural Limit State.” (in print)
Pannell, F. N., 1969, “The Ultimate Moment of Resistance of Unbonded fps = 209.4 ksi ≤ fse + 60 = 210 ksi O.K.
Prestressed Concrete Beams,” Magazine of Concrete Research, V. 21, No. 66, ≤ fpy = 230 ksi O.K.
Mar., pp. 43-54.
It should be noted that the ACI equation is not sensitive to
Revesz, S., 1953, “Factors Governing the Ultimate Bending Moment of
Reinforced and Prestressed Concrete Beams with Reference to a Proposed the presence of As, be it zero or double the value given.
Plastic Theory,” Magazine of Concrete Research, V. 5, No. 13, Aug., pp. 11-26.
Swiss Code, 1979, “Ultimate Load Behavior of Slabs,” SIA 162, Work- Steel tendons: proposed equation of Table 1
ing Party 5. (draft) (Eq. (14))
Tam, A., and Parnell, F. N., 1976, “The Ultimate Moment of Resistance The equation for force equilibrium of the section at nominal
of Unbonded Partially Prestressed Reinforced Concrete Beams,” Magazine
of Concrete Research, V. 28, No. 97, Dec., pp. 203-208. bending resistance assuming rectangular section behavior is
Tellerfield, D. W., 1953, “The Ultimate Flexural Strength of Prestressed written as
Concrete Beams,” thesis presented to the University of Sheffield, 127 pp.
Warwaruk, J., Sozen, M. A., Siess, C. P., 1962, “Investigation of Pre- Aps fps + As fy = 0.85fc′ β1c (A-1)
stressed Reinforced Concrete for Highway Bridges, Part III: Strength and
Behavior in Flexure of Prestressed Concrete Beams,” Bulletin No. 464, from which
University of Illinois Engineering Experiment Station, Urbana, Ill, Aug.,
105 pp.
A ps f ps + A s f y
c = ----------------------------- 0.612fps + 2 × 60-
- = ------------------------------------------- (A-2)
APPENDIX A: EXAMPLES 0.85f c′ b β 1 0.85 × 7 × 24 × β 1
It should be noted that, in the following computations, a
high number of decimals is used only to allow the reader to 0.612f ps + 120
verify the calculations; generally, this is not necessary. In the = ----------------------------------
-
following examples, the notation is given in Table 1 and is 99.96
consistent with the ACI Code used.
c = 1.2005 + 0.00612f ps (A-3)
EXAMPLE 1—SIMPLY SUPPORTED BEAM
The beam under consideration is the simply supported For simply supported beams with uniform loading
beam with the cross-sectional dimensions and material prop-
erties shown in Fig. A1. The section was reinforced with two 3 3
L1 = L2 and Ωu = ------------ = ---------------------------- = 0.1125
No. 9 reinforcing bars and prestressed with four 1/2 in.- L ⁄ dp 40 × 12 ⁄ 18
diameter unbonded prestressing strands. The span length L
was 40 ft.
L
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1
d
The stress in the unbonded tendons at nominal bending
c  L2
resistance should be computed using the 1999 ACI Code

ACI Structural Journal/July-August 2002 525


Fig. A2—Equivalent frame for slab considered.

= 150 + 0.1125 × 27,000 × 0.003  ------ – 1


18
c 
Fig. A3—Slab cross section and properties.
f ps = 150 + 9.1125  -----
18- – 1 (A-4)
c 
f ps = 150 + 2.7  ------ – 1
18
(A-5)
c 
Simultaneously solving Eq. (A-3) and (A-4) leads to a
quadratic equation in c
The solution of Eq. (A-2) and (A-5) leads to: c = 2.235 in.
2 and fps = 169.04 ksi.
c – 2.0627c – 1.0038 = 0

Its solution leads to Carbon FRP tendons


Assume that instead of having steel tendons, carbon FRP
c = 2.47 in. and fps = 207.3 ksi ≤ 0.8fpu = 216 ksi O.K. tendons of the same cross-sectional area Aps = 0.612 in2 are
used. Their strength and elastic modulus, respectively, are:
The solution of Eq. (A-3) and (A-4) could have also been fpu = 315 ksi and Ep = 21,000 ksi.
obtained by iteration such as that in the following steps:
L
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1
1. Starting with fps = 0.80fpu = 216 ksi, plug in Eq. (A-3) d
and compute c = 2.52 in.; c  L2
2. Plug c = 2.52 in. in Eq. (A-4) and compute fps = 206 ksi;
3. Plug fps = 206 ksi in Eq. (A-3) and compute c = 2.46 in.;
= 150 + 0.1125 × 21,000 × 0.003  ------ – 1
18
4. Plug c = 2.46 in. in Eq. (A-4) and compute fps = 207.5 ksi; c 
5. Plug fps = 207.5 ksi in Eq. (A-3) and compute c = 2.47;
and
f ps = 150 + 7.0875  ------ – 1
6. Plug c = 2.47 into Eq. (A-4) and compute fps = 207.3 ksi. 18
(A-6)
For the iteration, the starting value of fps could also be that c
recommended in Eq. (8); that is, fps = 150 + 15 = 165 ksi. The
iterations lead in order to the following successive results: c The solution of Eq. (A-2) and (A-6) leads to: c = 2.40 in.
= 2.21 in. → fps = 215.1 ksi → c = 2.517 in. → fps = 206.05 ksi and fps = 196.07 ksi.
→ c = 2.461 in. → fps = 207.54 ksi → c = 2.47 in. → fps =
207.3 ksi; that is, same answer given in the previous steps. EXAMPLE 2: THREE SPANS CONTINUOUS SLAB
It should be noted that, if the value of As is doubled, the pro- WITH STEEL TENDONS
posed procedure leads to c = 3.5463 in. and fps = 187.14 ksi. This An example of the design of a three-span continuous slab
shows that when the section tends toward over-reinforcement, using the equivalent frame method (Fig. A2) is covered in
the stress in the prestressing tendons at ultimate decreases. detail in Naaman (1982). An interior strip representing the
slab in one direction is considered herein. It is 30 ft wide and 9
Glass FRP tendons in. deep, with three equal spans of 30 ft each. The design
Assuming that instead of having steel tendons, glass FRP arrived at includes 35 strands 0.6 in. in diameter, all of which
tendons of the same cross-sectional area Aps = 0.612 in2 are were anchored at the two free ends of the slab. In addition,
used. Their strength and elastic modulus, respectively, are six No. 8 reinforcing bars are used at the supports along the
fpu = 300 ksi and Ep = 8000 ksi. column strip, and no reinforcing bars are used in span (Fig. A3).
The 1999 ACI Code Eq. (18-4) and (18-5) do not apply to The value of fps for the unbonded strands can be computed
FRP tendons. If used, they would lead to the same results as using various prediction equations:
those for steel tendons.
Using the proposed procedure leads to the same equation Using 1999 ACI Code Eq. (18-5)
of equilibrium as Eq. (A-2), and L = 30 ft dp = 7.75 in. ds = 7.75 in.

d L 30 × 12
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1
L
----- = ------------------ = 46 > 35
c  L2 dp 7.75

A ps
= 150 + 0.1125 × 8000 × 0.003  ------ – 1
18 7.56
ρ p = -------- = ----------------------------------- = 0.00271
c  bd p 30 × 12 × 7.75

526 ACI Structural Journal/July-August 2002


Fig. A4—Typical loading for maximum moment at interior Fig. A5—Typical loading for maximum moment in span.
supports.
f c′ 5 Critical section in span with two spans loaded (Fig. A5)—
f ps = f se + 10 + -------------------- = 160 + 10 + ---------------------------------- The main difference between this case and Case A is that
300 × ρ p 300 × 0.00271
there are no additional reinforcing bars in the section (Fig. A.3).
The force equilibrium equation leads to
= 176.15 ksi
A ps f ps 7.56f ps
This value of fps holds in span and at the supports. It remains c = ------------------------- = ---------------------------------------------------------
- (A-9)
the same whether the section has no reinforcing bars or 0.85f c′ b β 1 0.85 × 5 × 30 × 12 × 0.8
many reinforcing bars. It also remains the same for any loading
pattern; that is, whether one, two, or three spans are loaded. 7.56f ps
= ---------------- = 0.00618f ps
1224
Using the proposed equation for fps (Table 1)
Support sections with two spans loaded (Fig. A4)—The and the equation for fps is the same as Eq. (A-8) in Case A. The
equation of force equilibrium of the section at nominal bend- solution leads to: c = 1.1168 in. and fps = 180.71 ksi. The value
ing resistance is written as of fps herein is larger than that in Case A because the section
contains less reinforcement (that is, it has no reinforcing bars).
Aps fps + As fy = 0.85fc′ β1c (A-1) Critical section in span with only one span loaded (Fig. A6)—
Herein, because all spans are equal in length and the rein-
from which forcement is the same throughout, the same solution applies
to the two possible loading cases shown in Fig. A6.
A ps f ps + A s f y 7.56f ps + 4.74 × 60 The force equilibrium equation leads to
c = -----------------------------
- = ---------------------------------------------------------
-
0.85f c′ b β1 0.85 × 5 × 30 × 12 × 0.8
A ps f ps 7.56f ps
c = ------------------------- = ---------------------------------------------------------
- (A-9)
7.56f ps + 284.4 0.85f c′ b β 1 0.85 × 5 × 36 × 12 × 0.8
= ------------------------------------
1224
7.56f ps
= ---------------- = 0.00618f ps
c = 0.23235 + 0.00618fps (A-7) 1224

For two spans loaded out of three, L1/L2 = 2/3 For one span loaded out of three, L1/L2 = 1/3

3 - = --------------------------------
Ω u = ----------- 3 3 3
= 0.0646 Ω u = ------------ = -------------------------------- = 0.0646
L ⁄ dp 30 × 12 ⁄ 7.75 L ⁄ dp 30 × 12 ⁄ 7.75

d L
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1
L
f ps = f se + Ω u E p ε cu  ----p- – 1 -----1
d
c L2 c  L2

= 160 + 0.0646 × 27,000 × 0.003  7.75


---------- – 1 --- = 160 + 0.0646 × 27,000 × 0.003  7.75
---------- – 1 ---
2 1
 c 3  c 3

f ps = 160 + 5.2312  ---------- – 1 --- f ps = 160 + 5.2312  ---------


7.75- – 1 1
7.75 2
(A-8) --- (A-10)
 c 3  c 3

= 160 + 3.4884  7.75


---------- – 1 = 160 + 1.7442  ---------- – 1
7.75
 c   c 

Simultaneously solving Eq. (A-7) and (A-8) leads to a Simultaneously solving Eq. (A-9) and (A-10) for fps and c
quadratic equation in c. Its solution leads to leads to a quadratic equation in c. Its solution leads to

c = 1.3256 in. and fps = 176.9 ksi ≤ 0.8fpu = 216 ksi O.K. c = 1.057 in. and fps = 171.04 ksi ≤ 0.8fpu = 216 ksi O.K.

ACI Structural Journal/July-August 2002 527


In a three-span continuous beam, loading one span only,
such as the middle span, to obtain the maximum positive mo-
ment leads to a ratio L1/L2 = 1/3, whereas loading two adjacent
spans to obtain the maximum negative moment leads to a ratio
of 2/3. Using various loading patterns, it can be shown that, for
equal spans, continuous members (of any numbers of spans)
the ratio L1/L2 varies from 1/3 to 1. This is illustrated in Fig.
B1 in the pattern of Loading A to Loading I.
It can be argued, however, that pattern loading applies
only to the live load, whereas the dead load always applies
throughout. Moreover, at ultimate, each load is modified by
a different factor. This can have an influence on the value of
the stress increment induced in unbonded tendons, and an
adjustment to the multiplier factor L1/L2 may be warranted.
Fig. A6—Typical loading for one span. To evaluate this effect, an adjustment factor was derived and
evaluated. It is explained as follows.
It should be noted herein that the loading of one span versus
two spans results in a decrease of approximately 5.4% in fps Adjustment factor
and the corresponding value of nominal bending resistance. Assuming that the most common combination of dead and
The increment of stress here is 11.04 ksi; it is higher than that live load currently used in the ACI Code controls at ultimate,
using the conservative value of 7.5 ksi recommended in the section is subjected to the following loading
Eq. (18-4) in Table 1.
Critical section in span with three spans loaded—If the Fn = 1.4MD + 1.7ML (B-1)
critical section in span is analyzed assuming that all three
spans are loaded, the answer would be: c = 1.1706 in. and fps where Fn = the value of the prestressing force at nominal re-
= 189.41 ksi. This illustrates that the stress increment is high- sistance; MD = the dead load; and ML = the live load moment.
er than that for the cases where one or two spans are loaded. The nominal value of the prestressing force is due to the
In comparing this case to Example 1, which has a higher re- following stress in the prestressing tendons at ultimate
inforcement ratio, the influence of the span-depth ratio can
be observed. fps = fpe + ∆fps (B-2)
Four-span continuous slab with one span loaded—If the
slab considered previously had four continuous spans instead where fpe = the effective prestress; and ∆fps = the stress in-
of three, and if one span only is loaded, such as in Loading J crement achieved at nominal resistance.
of Fig. B.1 in Appendix B, the results obtained would be: c In principle, the effective prestress in prestressed concrete
= 1.041 in. and fps = 168.4 ksi. The stress increment herein accounts for the combined effects of prestress and dead load.
(in excess of fpe) is 8.4 ksi. This compares favorably with the This is how prestress losses are calculated. Thus, to be exact,
recommendation given in the approximate procedure (Eq. (18- 1 × D used throughout a continuous beam is already included
4) of Table 1) where stress increment is limited on the lower end in the value of fpe.
to 7.5 ksi. Hence, the stress increment ∆fps is due to any load in ex-
cess of the dead load. Assuming that the ultimate load is
Remarks (1.4D + 1.7L), then the increment of load affecting the value
The previous examples illustrate several points regarding of ∆fps is (0.4D + 1.7 L).
the proposed prediction equation for fps, namely that: In computing ∆fps for a pattern loading, one can further argue
• It accounts for the presence of the entire reinforcement that the increment of dead load up to ultimate should be
in the section (reinforcing bars and prestressing applied uniformly over the beam length (all spans), while the
strands); live load is applied selectively over some spans to create the
• It accommodates partial loading (such as one span out maximum effects.
of three) in continuous structures; The example of a three-span continuous beam should be con-
• It applies to steel or FRP tendons; and sidered, assuming interest in the loading that leads to maximum
• It is sensitive to the value of the span-depth ratio. More- moment at the first interior support. In a first assumption, the
over, it is stable over a wide range of variables and pattern loading shown in Fig. B2 (Assumption 1) is used as
parameters. assumed in Eq. (14). A second assumption of pattern loading
can be used, however, which is also shown in Fig. B.2
APPENDIX B: INFLUENCE OF (Assumption 2). The effect of using Assumption 2 versus
PATTERN LOADING ON ∆fps Assumption 1 on the increment of moment (and thus, the incre-
In the following examples, continuous beams of equal ment of stress) can be evaluated from the following factor
spans and a prestressing tendon running throughout the total
length of each beam are assumed; that is, L2 is equal to the L L
sum of the span lengths, while L1 represents the sum of the 0.4D ----1- + 1.7L ----1-
Assumption 1 L2 L2
lengths of the loaded spans. Factor = --------------------------------- = -----------------------------------------
- (B-3)
Because the stress in unbonded tendons depends on the Assumption 2 L1
0.4D + 1.7L -----
level of loading as modeled in Eq. (8) and (16), the pattern of L2
loading applied will influence the value obtained; this effect
is discussed subsequently. This factor can be computed for a real slab.

528 ACI Structural Journal/July-August 2002


Fig. B2—Two possible assumptions for incremental loading
to ultimate.

L L
0.2D -----1 + 1.6L ----1-
Assumption 1 L2 L2
Factor = --------------------------------- = -----------------------------------------
- (B-4)
Assumption 2 L1
0.2D + 1.6L -----
L2

Applied to the example of the slab described previously, it


will lead to a numerical value equal to 0.922 instead of
0.874; that is, the adjustment factor moves closer to unity
and the difference between the two assumptions of Fig. B2
becomes increasingly negligible.
Fig. B1—Examples of loading patterns for continuous
beams with different number of spans. In summary, it can be concluded that the influence of the
adjustment factor is generally not significant and could, for
all practical purposes, be ignored. Understanding its impli-
The slab under consideration is a three-equal-span continu- cations, however, may be useful in research studies or par-
ous slab described in detail in Naaman (1982). It is 9 in. deep ticular situations.
with a span length of 30 ft, and is subjected to a 25 lb/ft 2 The Subcommittee also believes that the main intent of using
superimposed dead load and an 80 lb/ft2 live load. Assuming load factors is to achieve a certain level of reliability or
normal-weight concrete leads to safety factor. For a target safety level, the increment of load
leading to failure can be achieved in many ways, whereas the
D = 112.5 + 25 = 137.5 lb/ft2 required nominal resistance can be the same.

L = 80 lb/ft2 Prestressed slabs and pattern loading


A final note regarding pattern loading and beams with un-
To design for the support section, a pattern loading of two bonded tendons is in order. Pattern loading is usually the
norm in conventional reinforced and prestressed concrete.
spans out of three is used (Fig. B1, Loading D); that is, L1/L2
When unbonded tendons are used, however, good engineering
= 2/3. The corresponding adjustment factor will be
judgment leads one to also consider a one-span loading such
as shown in Loading J of Fig. B1. This is because in satis-
2 2
0.4 × 1.375 × --- + 1.7 × 80 × --- fying the relationship
Assumption 1 3 3-
Factor = --------------------------------- = -----------------------------------------------------------------------
Assumption 2 2 Mu ≤ φMn (B-5)
0.4 × 137.5 + 1.7 × 80 × ---
3
both Mu and Mn are affected by the loading. Compared with
= 0.874 loading a single span, (such as in Loading J of Fig. B1), pat-
tern loading will increase Mu, but also will increase Mn
For a zero value of superimposed dead load, the factor through ∆fps. Thus, Loading J of Fig. B1 may be more criti-
would be 0.89. If the superimposed dead load is kept as is cal than Loadings H or I. The committee recommends that
and the live load is changed to 50 lb/ft2, the corresponding one-span loading always be considered in slab systems
because its likelihood is real, such as in the cases of fire or
value of the adjustment factor would be 0.84.
impact. The value of L1/L2 in such a case may be quite small,
In similar computations, it can be shown that, if L1/L2 = 1/2 which also leads to a small ∆fps. The designer is thus advised
(such as for Loadings F and G of Fig. B1) or L1/L2 = 3/5 (such to use the simple approximate formula for slabs suggested in
as for Loadings H and I of Fig. B1), the corresponding adjust- Eq. (12) and (13).
ment factor for the previous example would, respectively, be In summary, the Subcommittee believes that using the ratio
0.78 and 0.84. L1/L2 is a very good way to illustrate that the strain incre-
Returning to the derivation of the adjustment factor and using ment due to loading in one span is distributed throughout the
the newly recommended load combination of the ACI 2002 length of the tendon (as predicted by theory) and that one
Code; that is, (1.2D + 1.6L) instead of (1.4D + 1.7L), leads to cannot always count on the most favorable conditions.

ACI Structural Journal/July-August 2002 529

You might also like