Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
95
96 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
A fundamental difference exists between analysis and design. In dealing with civil
engineering structures, design implies an unknown product, at least in part, while
analysis implies investigating or reviewing a finished or proposed product. The
analysis process, also called investigation or review process, is concerned with
assessing the response of the structure to the application of loadings. It deals in part
with the determination of stresses and stress or force resultants and in part with
checking if the structure satisfies acceptable design criteria. This is often done by
comparing actual findings with corresponding limits and ranges recommended in
prevailing codes. To analyze is also to compare with what engineering judgment or
the state of the art considers acceptable. Since it generally involves no unknown, the
analysis process is easier than the design process.
In civil engineering structures, design involves the selection of many particulars
among a large array of possibilities, such as the structural layout, the shape of a
member, the structural material, and even the construction process. Within each step
the design deals with the actual versus the ideal and at different levels of details.
Although design does not necessarily imply finding the optimum solution, it
certainly aims at being within an appropriate range of the optimum. Because of its
inherent nature of dealing with unknowns and because infinite combinations of
possibilities exist, design is mostly an iterative process. An efficient design process
is one in which the number of iterations is reduced to a minimum. This often
depends on the experience and skill of the designer. A distinction is made between
preliminary design or designs in which many alternatives can be explored quickly
(using engineering judgment, rule of thumb, etc.), and final design which is a more
refined solution, ready for implementation.
More on sorne detailed steps of what "analysis versus design" implies in dealing
with prestressed concrete beams can be found in Chapter 4. Similar comparisons
could apply to other structural elements and systems.
In arder to fulfill its purpose, a structure must satisfy a number of design objectives.
Assuming that the need for a structure has been identified and justified, most basic
design objectives are [Ref. 3.26]:
• feasibility
• constructability
• safety
• serviceability
• economy
• functionality
• aesthetics.
Chapter 3 - THE PHILOSOPHY OF DESIGN 97
Although implied, the following criteria are also being increasingly considered in
modem design:
• environmental impact,
• sustainability and maintainability
• ease of future demolition
• replacement.
Given available materials and technologies, a structure must be first of all
feasible. At present, for instance, it is not possible to build a simply supported
prestressed concrete bridge with a main span of more than about 500 meters.
Feasibility and economy often go together. To be built, a proposed design must be
reasonably economical in comparison to other potential altematives. Assuming the
above objectives are within range, a structure must be safe, that is, it should not
collapse under loads foreseeable during its service life. It should also be serviceable,
that is, it should perform properly under load and render the service for which it was
built. A prestressed beam may be completely safe, yet unserviceable if it undergoes
very large cambers. Similarly, a water tank may be safe but unserviceable if it
cracks and leaks substantially.
The safety and serviceability of structures are generally achieved by satisfying a
number of code limitations or criteria. To ensure safety, severa! design approaches
are available and accepted by various codes. They are treated in more detail in the
next sections. Serviceability generally includes aspects of cambers and deflections,
fatigue, corrosion, cracking, and fire resistance. Serviceability criteria and how to
satisfy them in the design of prestressed concrete structures are addressed in Chapter
7 and Ref. 1.48.
considered a serviceability limit state; however, it can be also argued that since
fatigue leads to failure, it should be considered an ultimate limit state. Either way it
is accommodated by the limit state design philosophy.
To accommodate a limit state, a design criterion is set. For instance, a crack
width limit state implies that the crack width is less than or equal to acode specified
value. It is then said in design that the structure must satisfy the maximum crack
width limit state when subjected to service loads. Typically a design criterion takes
on the following logical form:
Whether the inequality is more than or equal or less than or equal depends on
the particular case. Deflection should be less than or equal to a code defined limit,
while resistance should be more than or equal to a code set limit.
4. Reliability Criteria:
[where any ofthe above criteria represents a limit state, anda given probability of
exceeding a limit state is the objective]
5. Special Criteria:
Related to extreme earthquakes; nuclear explosions; accidental impact; volcanoes.
following other design approaches, used at different times, and/or for different
structural materials.
Currently encountered design approaches utilize the following terminology and
are listed primarily in the order they were historically used by the design profession:
The working stress design (WSD) or allowable stress design (ASD), sometimes
also called permissible stress design (PSD), are different terminologies for the same
approach in which maximum stress criteria are specified; that is, sorne allowable
stresses are not to be exceeded under service loads [Ref. 3 .1, 3. 7]. The ultima te
strength design (USD), and the load and resistance factor design (LRFD) imply
essentially the same philosophy wherein strength limit states and possible collapse
are considered under factored loads [Refs. 3.2, 3.4, 3.8]. The limit state design
(LSD), described in the previous section, is a general approach where ali limit states
such as stresses as well as collapse can be considered simultaneously. LSD is the
general modem approach to follow (Fig. 3.1) and encompasses all the other
approaches. The reasons for which a subset approach, such as WSD, is used, are: 1)
either because WSD was used historically first, or 2) because, WSD was found from
experience to be generally most critical ( controlling) for a design with a typical
material or structure.
Other terminology related to structural design approaches includes: plastic
design, nonlinear design, and probabilistic design; these refer to a specific method of
design but essentially should follow the same philosophy as the limit state design.
Note that WSD and USD may result in the same member design, but USD
generally allows for a more rational distribution of safety reserve. Although a single
approach may generally be sufficient, current practice in prestressed concrete
(Section 3.10.1) involves the combination of working stress design (WSD) and
ultimate strength design (USD or SD). LSD philosophy allows combining the two
WSD and USD, while safety factors, load factors and resistance factors are obtained
from a rational and complex calibration of data to achieve a consistent level of safety
and reliability ( or probability of survival) for every limit state considered.
Except for probabilistic design, design approaches begin with the choice of the
specified loads to design for. The nature and magnitude of these loads depend on the
type of structure. Sorne are described in Section 3. 7.
Chapter 3 - THE PHILOSOPHY OF DESIGN 101
In this approach the stresses under working loads are limited to permissible values or
allowable stresses and the structure is analyzed assuming linear elastic materials and
thus linear elastic behavior. Safety is ensured by selecting allowable stresses as
relatively small fractions of the characteristic strengths of the component materials.
Allowable stresses are specified in various codes and may vary from one code to
another. Typical values for prestressed concrete are described in Section 3.7. For
instance, the maximum permissible compressive stress on concrete flexural members
may be taken as 0.45 f'c· This implies a safety factor of (1/0.45) = 2.22 against
concrete compressive failure under short term loading. Note that in the working
stress design ali types of loads are treated the same, no matter how different their
variability is. This is less realistic than in the USD approach where each load is
treated according to its degree of uncertainty. The design of prestressed concrete
beams using the working stress design approach is covered in detail in Chapter 4.
When the WSD (or ASD) approach is used, the following general formulation
(criterion) applies:
where absolute values of stresses are considered. The allowable stress can be
obtained from:
Assuming linear elastic behavior under bending, axial tensile load, or shear, the
actual stress induced by loading in an uncracked section is directly proportional to
the applied load:
N
Actual Stress = for axial tensile load (3.4)
Ac
M
Actual Stress = z
for a bending moment (3.5)
Thus the allowable stress design criterion (Eq. 3.2) can be generally written as
follows:
(3.7)
where <Y¡ is the stress due to loading i, <Yn is the equivalent elastic stress at nominal
resistance, and SF is a safety factor larger than one. The summation of stresses is
done over the applied load combination of interest. Assuming for instance a
combination of dead load and live load (often termed service load, that is, (D + L))
leads to:
where OJJ is the stress due to dead load and O"L is the stress due to live load.
Since, under elastic bending, stresses and moments are directly proportional, Eq.
(3.8) can be written:
where Mn is the nominal bending resistance. The safety factor is often taken
between 2 and 2.5 in the WSD ( or ASD) approach.
In this approach, the design working loads are multiplied by load factors and the
structure is designed to resist the factored loads, at its ultimate capacity. The load
factors are different for each type of load and are adjusted to reflect the degree of
variability and uncertainty of that load. This is more realistic than in the WSD
approach, where all loads are treated the same.
When either the USD, LSD, or LRFD approach is applied, the general format is
given by the formula [Refs. 3.2, 3.8, 3.22]:
(3.10)
where:
i type of load, i.e., dead load, live load, wind, etc.
Q¡ nominal load effect
y¡ load factor corresponding to Q¡
Chapter 3 - THE PHILOSOPHY OF DESIGN 103
When the USD approach is applied to bending, direct axial tension, or shear in
reinforced or prestressed concrete, the following formulation is obtained:
(3 .11)
(3.12)
(3.13)
where M stands for bending moment, N for tensile axial load, V for shear force, and
other notation is as above.
Note that the load and resistance factors are obtained from calibration using
reliability analysis. Load factors are higher for loads with higher degree of
uncertainty. This is why the live load factor for reinforced concrete is higher than
the dead load factor. On the other hand, resistance factors are lower for types of
members or materials with higher degree of uncertainty. This is why the capacity
reduction factor, <P, for shear is smaller than that for bending.
In comparing Eq. (3.9) with Eq. (3.11), it can be observed that in the USD-LRFD
approach, the effects of loads are multiplied by a factor ( different for each load) and
compared to the nominal resistance, while in the WSD-ASD approach, the load
effects are used without modification and the resistance is divided by a factor.
However, the difference lies in the fact that in USD-LRFD each load has a different
load factor, while in WSD-ASD the same safety factor is applied to the effects of all
loads. Figure 3.2 illustrates the two approaches assuming a tensile member of
known nominal tensile resistance, Ni;
U)
U)
w YoNo +yLNL s ,Nn
o::
1-
U) ªn Nn
w ,Nn
...J ::,
cñ z
z e
w .J
z.J
1-
ªn
<
o
w 0allowable =
:,...
...J
1- +
Safety Factor
cñ e w
o
a.
.J ze ...J
b ªn cñ
:i1i: cro + crl ~ - :,...
zw
o +8 SF
u 1-
o ELONGATION
•
Figure 3.2 Illustration of the WSD (left axis) and USD (right axis) approach for a tensile
member.
Chapter 3 - THE PHILOSOPHY OF DESIGN 105
Although basic engineering concepts and judgment can be used to design a structure,
most of the guesswork can be reduced and better efficiency achieved if structural
requirements set by design codes are satisfied. Design codes ( such as Refs. 3 .1 to
3.11) are written to protect the user and society as a whole. They provide
information on methods of analysis and design, minimum design requirements and
minimum expected performance. They represent a summary of the collective
opinion or agreed upon state of knowledge of the profession.
In the United States most reinforced and prestressed concrete structures ( except
bridges) are designed in accordance with the Building Code Requirements for
Reinforced Concrete (ACI 318) published by the American Concrete Institute [Refs.
3.3 and 3.4]. lt is regarded as an authoritative statement of current good practice in
the field of concrete structures. The ACI code is incorporated entirely or in part in
106 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
many municipal and regional codes in the United States and is used as a reference in
many foreign countries.
Most prestressed concrete bridges for highways or railways in the United States
are designed in accordance with three major codes: the AASHTO (American
Association of State Highway and Transportation Officials) Standard Specifications
for Highway Bridges [Ref. 3.1], the AASHTO LRFD Specification for Highway
Bridge Design [Ref. 3.2] and the AREA (American Railway Engineering
Association) Manual for Railway Engineering [Ref. 3 .1 O, 3 .11]. The AASHTO
Standard specifications are essentially based on the working stress design, while the
AASHTO LRFD specifications are essentially based on limit state design. Except
for the LRFD specification and for sorne subtle differences, the sections of AASHTO
Standard and the those of the AREA manual related to prestressed concrete are
essentially identical in scope to the corresponding sections of the ACI code.
However, the AASHTO LRFD specifications are significantly different. Sorne of
these differences are discussed in Chapter 5 in relation to various limitations on
reinforcement.
In this text, reference to the above documents ( or simply codes) will frequently be
made and preferably their latest editions should be used. These codes contain
specific information on analysis and design methods, as well as service loads, load
factors, and allowable stresses. These last three items will be discussed next.
3.7 LOADS
For the purpose of design, loads are classified into two main categories: dead loads
(also identified as permanent loads) and live loads (also identified as transient loads).
Dead loads include primarily the self-weight of the structure and any permanent
component such as tiles, false ceiling, and partitions. They are assumed to remain
constant during the life of the structure. Self-weight is estimated from the
dimensions of the element and the unit weight of the material. Dead loads due to
partitions and the like are generally approximated by an equivalent uniform load
applied to the surface area of the structure. Typical values of unit weights of sorne
common materials are given in Table 3.1. In general, one can assume a unit weight
of 150 lb/ft-' (23.6 kN/m3) for normal weight concrete (including the steel) and
between 100 and 120 lb/ft3 (15.7 to 18.9 kN/m3) for structural lightweight concrete.
Unlike dead loads, live loads are variable in nature and fluctuate with time. They
include (1) occupancy loads caused by people, furnishings, or movable objects, (2)
vehicle loads such as trucks or trains, (3) snow, rain, water, ice, wind, earth pressure,
temperature loads, effects of earthquakes, and the like. Occupancy loads are
generally prescribed in various codes as uniformly distributed loads. The ACI code
does not prescribe occupancy loads. However, the American National Standards
Institute (ANSI) and the American Society of Civil Engineers (ASCE), as well as
local and regional codes, prescribe such values. Typical selected values of
occupancy loads recommended by the ASCE/ANSI code [Refs. 3.7 and 3.12] are
Chapter 3 - THE PHILOSOPHY OF DESIGN 107
Live load
Occupancy or use lb/ft2 kPa or
kN/m2
Assembly halls
Fixed seats 60 2.9
Movable seats 100 4.8
Balcony (exterior) 100 4.8
Dining rooms, restaurants, dance halls 100 4.8
Garages (passenger cars) 100 4.8
Floors shall be designed to carry
150% ofthe maximum wheel load
anywhere on the floor
Hospitals
Operating rooms 60 2.9
Private rooms 40 1.9
Hotels
Guest rooms 40 1.9
Public rooms, lobbies, first floor 100 4.8
corridors
Housing
Private houses and apartments 40 1.9
Public rooms (in multifamily units) 100 4.8
Libraries
Reading rooms 60 2.9
Stack rooms 150 7.2
Office buildings
Offices 80 3.8
Lobbies 100 4.8
Schools
Classrooms 40 1.9
Corridors 100 4.8
Sidewalks, driveways subject to trucking 250 12
Stairs and exitways 100 4.8
Storage warehouses
Light 125 6
Heavy 250 12
Yards and terraces, pedestrians 100 4.8
Chapter 3 - THE PHILOSOPHY OF DESIGN 109
reproduced in Table 3.2. A more extensive list can be found in Ref. [3.12] as well
as particular details for other live loads. Wind loads of 15 to 30 lb/ft2 (~O. 7 to 1.4
kN/m2) and snow loads of 1 O to 40 lb/ft2 ( ~ 0.5 to 2 kN/m2) are common ranges for
design in the US.
Vehicle loadings are prescribed by the AASHTO and AREA specifications for
highway and railway bridges, respectively [Refs. 3.2, 3.10]. The live loads
recommended by AASHTO comprise two groups: moving vehicle loads and
equivalent lane loading in the form of a uniform load. Most critical load
combinations and/or arrangement have to be considered in design. Many factors,
such as impact, the number of lanes, the width of the lanes, and the stiffness of the
structure are associated with bridge loadings; their description and treatment is
discussed in Chapter 14.
In the working stress design approach a number of stress limits, called allowable
stresses or permissible stresses, are needed. These allowable stresses are not to be
exceeded by actual stresses under application of service loads. They are in general
prescribed by the code adopted for the design. Typical values of allowable stresses
imposed by the ACI code are summarized in Tables 3.3 to 3.8 for concrete,
prestressing steel, and reinforcing steel, respectively.
3.8.1 Concrete
Allowable stresses for concrete (Table 3.3) are separated into two groups: the first
one corresponds to initial stresses at time of transfer of prestress (before prestress
losses), and the second one corresponds to final or service load stresses (after
allowance for prestress losses). The subscript i such as inf'ci is associated with the
first group. Tension in concrete has been given a negative sign in accordance with
the sign convention adopted in this text. Also given in Table 3.3 are the stresses in
the equivalent SI system. Note that a tension of -6,[iZ is generally allowed during
service on the concrete precompressed fiber. A fictitious value of -12,[iZ or more
is allowed to accommodate partial prestressing.
Two allowable values of extreme fiber compressive stress in service due to
prestress and total load are given to reflect the nature of the service load and its
effect on long term properties, particularly creep, deflections, and fatigue. Thus,
when the sustained dead load and live load are a substantial portion of total service
load, the 0.45f'c allowable stress limit may control. On the other hand when a large
portion of the total service load consists of a transient temporary service load, the
stress limit of 0.6f'c may control. Most frequently the stress in compression under
service load does not control the design of prestressed concrete flexura! members.
110 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
Thus, the above distinction between the two values of allowable stress may not be of
significance. In most of the examples given in this text, the lower limit of 0.45f'c is
considered to be on the safe side. Engineering judgment should be used to decide
whether the higher allowable stress limit applies.
Numerical values of allowable stresses are given in Tables 3.4 and 3.5 for typical
concrete strengths, in U.S. and SI units, respectively.
Table 3.3 Allowable stresses in concrete in prestressed flexura! members. (Adapted from ACI 318 Code,
Ref 3.3)
psi MPa
(a) Extreme fiber stress in compression dueto prestress plus 0.45!; 0.45.fc'.
sustained load
(b) Extreme fiber stress in compression due to prestress and 0.60 .f; 0.60.fc'.
total load
(e) Extreme fiber stress in tension in precompressed tensile
zone -6Ji -o.sofi
(d) Extreme fiber stress in tension in precompressed tensile
zone ofmembers (except two-way slab systems) where
-12¡¡; .i:
analysis based on transformed cracked sections and on
bilinear moment-deflection relationships shows that
immediate and long-time deflections comply with
requirements stated elsewhere in the code
The allowable stresses recommended by AASHTO for bridges [Refs. 3.1, 3.2]
are different from the corresponding ACI code values. Mostly, the AASHTO LRFD
code recommends allowable tensile stresses smaller than those given by ACI
because of concems about corrosion. Different stress limits are given for segmental
and non-segmental construction. Allowable stresses by AASHTO should be used in
bridge design and are given in Chapter 14.
Chapter 3 - THE PHILOSOPHY OF DESIGN 111
-o.2s.¡¡;;· Cl
-1.22 -1.41 -1.58 -l.73 -1.87
0.60.t; 18 24 30 36 42
Allowable stresses in prestressing steels specified by the ACI code are summarized
in Table 3.6. Different values are given depending on whether the stress applied
occurs just before or after transfer of the force from the steel to the concrete.
Previous editions of the code also used to specify a limiting final stress of 0.60Jpu
after all losses. This <loes not seem to be necessary because, in practice, the actual
final stress or effective prestress is seldom larger than 0.60Jpu; it is commonly of the
order of 0.5Jpu to 0.55Jpu, The second colurnn of Table 3.6 contains the code
provisions in terms of the yield strength of the prestressing steel. The third and
112 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
fourth columns show the corresponding values in terms of the ultimate strength,/p,,,
assuming either a stress relieved strand or a low relaxation strand, and keeping in
mind the maximum limit of 0.8fp11 given in the code.
Table 3.6 Allowable stresses in prestressing steel. (Adaptedfrom ACI 318 Building Code, Ref. 3.3)
l. Due to prestressing steel jacking force, but not 0.94fpy 0.80.fpu 0.80./~u
greater than the lesser of 0.80li,u and the
maximum value recommended by the
manufacturer of the prestressing steel or
anchorage devices
Allowable stress 145 1000 160 1104 235 1622 250 1723 270 1860
J, ksi MPa ksi MPa ksi MPa ksi MPa ksi MPa
At jacking 0.80fi,u 116 800 128 882 188 1296 200 1378 216 1488
After transfer 107 740 118 817 174 1200 185 1275 200 1376
0.74[¡,!I
After transfer 101.5 700 112 773 164.5 1134 175 1206 189 1302
O. 7fi,u
Note, from the third column of Table 3.6, that if fpy is taken equal to 0.85/p,, such
as for a stress-relieved strand (see Section 2.3), then the limits of 0.82/py and 0.94{py
become 0.7/p,, and 0.80/p11, respectively. On the other hand for a low relaxation
strand for which/py = 0.90fp11, the above limits of 0.82/py and 0.94{¡,y become 0.74{¡,u
and 0.85/p,,, respectively. This last term is then limited by 0.80/p,,. Often in this text,
for simplicity, a limit of O. 7/p,, is used as the transfer stress.
Chapter 3 - THE PHILOSOPHY OF DESIGN 113
Typical numerical values of allowable stresses as per the ACI code for various
types of prestressing steels are given in Table 3.7 in both the US and the SI systems.
Stress limits for prestressing tendons given by AASHTO LRFD specifications are
very similar to those given by the ACI code.
Allowable stresses in the reinforcing steel are given in Table 3.8. As the current ACI
code deals mostly with ultimate strength design, these stresses are not given directly
in the body of the most current code but in Appendix B of the 1983 edition where
working stress design (termed altemate design method) provisions are explained. In
principie, they should apply to reinforcing steels used in prestressed concrete. Note
that for cases where temporary tension in the concrete is to be resisted (such as in the
note of case l(c) of Table 3.3) a permissible stress of 0.60Jy or 30 ksi (207 MPa),
whichever is smaller, can be used throughout.
Table 3.8 Allowable stresses in reinforcing steels. (Adaptedfrom ACJ 318-83 Code, Ref. 3.3)
Allowable stress, fs
In ultimate strength design (USD) or strength design (SD), service loads are
multiplied by load factors and the member is designed to resist the factored loads.
Load factors adopted by the ACI codeare summarized in Table 3.9 and are based on
severa! studies dealing with reliability analysis and code calibration [Refs. 3.17,
3.18, 3.20, 3.22]. Different factors are given for different loadings. Their magnitude
is in effect adjusted to reflect the degree of uncertainty associated with the type of
loading. Thus, load factors for dead loads are smaller than those for live loads.
Severa! load combinations are to be considered in design. Reduced load factors are
used on sorne load combinations because of the low probability of their simultaneous
occurrences.
114 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
The load factor on L in Required strength U shall be at least equal to the effects of
Eqs. (9-3) to (9-5) shall be factored loads in Eqs. (9-1) through (9- 7). The effect of
permitted to be reduced to one or more loads not acting simultaneously shall be
0.5 except for garages, investígated.
areas occupied as places of
public assembly, and ali • U= I.4(D + F) (9-1)
areas where the live load L
is greater than 100 psf. • U= l.2(D + F + 1) + 1.6(L + H) + 0.5(Lror Sor R) (9-2)
When wind load Whas not • U= f.2D + 1.6(Lr or Sor R) + (l.OL or 0.8 W) (9-3)
been reduced by a
directionality factor, it shall • U= l.2D + 1.6 W + l.OL + 0.5(Lr or Sor R) (9-4)
be permitted to use 1.3 W in
place of l .6W in Eqs. (9-4) (9-5)
• U= l.2D + 1.0E + l.OL + 0.2S
and (9-6).
• U= 0.9D + 1.6W + l.6H (9-6)
Where earthquake load E is
based on service-level (9-7)
seismic forces, l.4E shall • U= 0.9D + 1.0E + 1.6H
be used in place of 1.0E in where:
Eqs. (9-5) and (9- 7). U= required strength to resist factored loads or related
moments and forces
The load factor on H shall D= deadloads
be set equal to zero in Eqs. E= load effects of earthquake
(9-6) and (9- 7) if the F= loads due to fluids wíth well defined pressures and
structural action due to H maximum heights
counteracts that due to W H= loads dueto weight and lateral pressure of soil and
or E. water in soil
L= live loads
Where lateral earth L r = roof live loads
pressure provides R= rain load
resistance to structural S= snow load
actions from other forces, it T= self-straining forces and effects arising from
shall not be included in H contraction or expansion resulting from temperature
but shall be included in the changes, shrinkage, moisture changes, creep in
design resistance. component materials, movements dueto differential
settlements, or combinations thereof
W= wind loads
One of the combinations that often controls the design is the combination of dead
and live loads (1.2D + I.6L); it is often used in this text as typical in many examples.
Chapter 3 - THE PHILOSOPHY OF DESIGN 115
lt replaces the combination ( 1.4D + l. 7L) which was used in prior versions of the
ACI code.
The reader is referred to the ACI code for further details about special conditions
of applications and exceptions of the load combinations given in Table 3.9. In
particular the ACI code leaves sorne leeway to engineering judgment as to when
such load factors can be increased. Indeed it states: "If special circumstances
require greater reliance on the strength of particular members than encountered in
usual practice, some reduction in the stipulated strength reduction factors rjJ or
increase in the load factors y may be appropriate for such members."
A large number of load types, load factors and their combinations is given in the
AASHTO LRFD specifications for bridge design (Tables 14.1 and 14.2). These
factors may be substantially different in magnitude from those recommended by the
ACI building code. For flexural design, the simplest combination for strength is
(l.25D + l.75(L + IM + PL)), in which L stands for vehicular live load, IM for
impact and PL for pedestrian load. An approach similar to that of AASHTO is taken
by ACI Committee 343 in its report on the analysis and design of reinforced concrete
bridge structures [Ref. 3.5].
Table 3.10 Strength reduction factors ~- (Adaptedfrom ACI 318-02 Code, Ref 3.3)
Resisting effect ~
Flexure without axial load 0.90
or tension controlled sections
Compression-controlled sections:
(a) members with spiral reinforcement 0.70
(b) other reinforced members 0.65
Shear and torsion 0.75
Reinforced In earthquake resistant structures:
and • Shear 0.60
prestressed • Shear in joints and coupling beams 0.85
concrete Bearing on concrete ( except for anchorages) 0.65
Bearing in post-tensioned anchorage zone 0.85
Bearing for strut-and-tie models and nodal zones 0.75
Development length, bond 1.00
Any effect using working stress design 1.00
Plain Bending, shear, compression and bearing in 0.55
concrete structural plain concrete
The principies of limit state design (LSD) described in Section 3.4 apply to both
reinforced and prestressed concrete. However, practica! experience leads to
selecting a sequence of steps in prestressed concrete design that is different from that
used for reinforced concrete (Fig. 3.3). The design of prestressed concrete members
is primarily controlled by allowable stresses (serviceability criteria), while the design
of reinforced concrete members is controlled primarily by ultimate strength criteria.
Typically the design of prestressed concrete starts by working stress design
procedures (WSD) to help dimension the member; then ultimate strength design
(USD) criteria are checked, and adjustments are made should a design criterion be
violated. In comparison, the design of reinforced concrete members starts directly
by USD procedures; generally allowable stresses are not of concem and
serviceability criteria such as crack width and deflections can be accommodated by
proper dimensioning and detailing. Thus, for all practica! purposes, the design of
reinforced concrete is primarily controlled by USO, while that of prestressed
Chapter 3- THE PHILOSOPHY OF DESIGN 117
2 Serviceability l
limit states Í'1LL ~-
360
3 Serviceability l Chapter 7
limit states Í'1LL ~-
360
Figure 3.3 Typical design steps sequence comparing reinforced and prestressed concrete.
Integrating all the stresses overa prestressed concrete section leads to a compressive
force resultant defined as the C force. In the absence of any externa! moment, the C
force is equal and opposite to the prestressing force. Similarly to the prestressing
118 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
force F, which has an eccentricity ea with respect to the neutral axis of bending, the
location of the C force can be defined by an eccentricity, e¿ (Fig. 3.4a).
The application of any extemal moment (Fig. 3.4b) adds stresses to the section;
however, since a moment is a couple and the two forces of a couple balance each
other (i.e., the sum of the forces of a couple is equal zero ), the resulting compressive
force in the concrete section remains equal to C. Adding a moment changes the
stress diagram, but does not change the value of C. However, the line of action of C,
that is its eccentricity, ec, shifts from the line of action of F, which remains as placed
at ea. The difference between them is equal in magnitude to M/Fas shown in Fig.
3.4c. This is exactly the same as to what happens in a column subjected to a
concentric axial load, when a moment is added. The effect of the moment is to shift
the action of the compression force to a line parallel and eccentric to the column's
axis. In beams, the extemal moment varíes along the span; thus the eccentricity of
the C force also varíes. The combined action of an external moment, such as the
moment due to self-weight of the member, and the prestressing force, can be
replaced by the action of the C force acting along its eccentricity, e.: The use of the
C force or the C-line (that is, the geometric location of the C force along the span)
can be advantageous in sorne calculations such as for deflections, load balancing,
and continuous members.
1~----
C=F
1 •
F
3.10.3 Characteristic Response of RC, PC, and PPC in Bending in the Elastic
Range of Behavior
Typically a reinforced concrete beam is cracked under the effect of dead load, a fully
prestressed concrete beam is uncracked under the effect of full service load (dead
load plus live load), and a partially prestressed beam can have any behavior in
between.
Usually a partially prestressed beam is designed to be uncracked under dead load
and cracked under full service load. In all three cases linear elastic behavior is
assumed up to at least foil service load. Figure 3.5 provides a schematic
Chapter 3- THE PHILOSOPHY OF DESIGN 119
representation of typical stress or strain diagrams (assumed linear) along the section
of either a reinforced, or a prestressed, or a partially prestressed concrete beam,
under service loads. The behavior is significantly different in the three cases. lt can
be observed, for instance, that the neutral axis of bending of the cracked section in a
reinforced concrete beam also corresponds to the zero stress point, and is
independent of the applied moment. Thus an applied moment causes an increase in
the resulting forces of the couple acting on the section ( C = T) but the lever arm of
the couple remains constant (Fig. 3.5a).
:r
Tension
(b)
Tension-¡- Compression
Ect
(c)
Figure 3.5 Comparison of bending response of RC, PC, and PPC sections in service.
120 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
In fully prestressed concrete (Fig. 3 .5b), the section is uncracked and the neutral
axis of bending is the centroidal axis of the uncracked section. The force resultant,
C, of compressive stresses in the section is equal to F. An externa] moment causes
no change in the force resultant in the concrete, that is, C = F; however, unlike in
reinforced concrete, it causes a shift in the line of action of C, which deviates from
the line of action of F, leading to an increase in the lever arm between them. In a
partially prestressed beam (Fig. 3.5c), the zero stress point moves up or down
depending on the magnitude of applied moment (below or above the cracking or
decompression moment) and the neutral axis of bending is different for the cracked
or uncracked section. Both the forces and the lever arm between them may vary
simultaneously when an externa! moment is applied.
Tension Compression
--:r-
( - ) .......
1----,1----, +)
..~ (
1
cctop ¡..-¡ccbod ...... ¡
----------------------·7
/
/
/
/
/
/
/ /
/ /
/ /
/ /
/ /
/ /
/ /
/
/ /
/
Figure 3.6 Definition of curvature. (a) Negative curvature of uncracked section. (b) Positive
curvature of uncracked section. (e) Positive curvature of cracked section.
éctop - écbot
(f)=-~--- (3.14)
h
Chapter 3- THE PHILOSOPHY OF DESIGN 121
or
&ctop
(f)=-- (3.15)
e
where &ctop is the strain in the top fiber, &cbot is the strain in the bottom fiber, h is the
depth of the section, and e is the depth of the neutral axis in a cracked section.
Compressive strain is considered positive and tensile strain is considered negative. A
positive value of curvature leads to a downward deflection in a beam while a
negative value leads to an upward camber. Note that Eq.(3.14) may be more
convenient for a prestressed concrete section, since stresses can be readily computed
and strains are related to stresses by the Hooke's law, that is: a- = E&. Equation
(3 .15) is more convenient for a cracked section such as reinforced concrete under
either service or ultimate loads, or for prestressed concrete at ultimate. Note that the
two expressions are essentially the same. Indeed if a line parallel to the strain
diagram is plotted in Fig. 3.6c starting at the bottom fiber, it will lead to a strain on
the top equal in magnitude to &ctop + l&cbot ¡ . In such a diagram the value of h is the
same as e in Fig. 3.6b and Eq. (3.15) applies.
The above expressions of curvatures are derived from the following fundamental
relationship of mechanics:
(3.16)
Me
0'=- (3.17)
1
la- JE& e
(f)=-=-=- (3.18)
Elc Ele e
which is essentially the same equation as Eq. (3.15). Note that M should include the
moment due to prestressing, if any.
dead load, generates an externa) moment with a parabolic moment diagram; the
effect of such a moment can be balanced by a prestressing force with a parabolic
tendon profile that generates a prestressing moment equal and opposite to the
externa! moment.
w/2
Parabolic tendon profile: F =
85
Uniform load. w
(a)
F
T~
+
{b)
~-
Moment diagram is plotted positive
2Fv = 2F sinti ~ 2F tantl = 2F l ~ 2 = P
downward to mimic the shape of the
prestressing force profile
i eT - ~ - - - - - - - - - - - - - - - - (e)
: F F
:
:
!
Stress diagrams at F
midspan for all the =
above cases: 4: (d)
The 2002 edition of the ACI code attempts to provide a smooth transition from fully
prestressed concrete to fully reinforced concrete, as well as a seamless transition
from bending members to compression members. The two main items of concem in
this text are discussed next.
Prestressed concrete flexural members are classified as Class U, Class T and Class C
based on the computed extreme fiber stress, CJis, at service loads in the
precompressed tensile zone, as follows:
• Class U: lats 1::::; 7 .sfJ! in psi or lats 1::::; O. 7 fJ! in MPa
• Class T: 7 .sfJ!::::; lats 1::::; 12fJ! in psi or O. 7 fJ!::::; lats 1 ::::; fJ! in MPa
• Class C: lats 1 > 12fJ! in psi or lats > fJ! 1 in MPa
124 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
in which U stands for uncracked, C for cracked, and T for transition between
uncracked and cracked.
Table 3. 11 ACI 318-2002 serviceability design requirements and related code sections.
Prestressed
Nonprestressed
Class U Class T Class C
Transition between
Assumed behavior Uncracked uncracked and cracked Cracked Cracked
Allowable stress at
18.4. 1 18.4.1 18.4.1 No requirement
transfer
Tensile stress at
service load 18.3.3
~7.5.fj; psi 7.5.fj; < .t; ~ 12.fj; No requirement No requirement
(Absolute value)
~0.7.fj; MPa O. 7 .¡¡: < .t; ~ frc
Deflection 9.5.4.1 9.5.4.2 9.5.4.2 9.5.2, 9.5.3
calculation Cracked section, Cracked section Effective moment
Gross section
basis bilinear of inertia
10.6.4
Crack control No requirement No requirement 10.6.4
Modified far
strand
or 0.6fy
Side skin
No requirement No requirement 10.6.7 10.6.7
reinfarcement
Table 3.11 summarizes the intent of the code. lt can be observed that linear
elastic behavior is assumed whether the section is cracked or not and that the main
difference between Class U and Class T is that in Class T the stresses are computed
as if the section is uncracked, while in reality the section is likely to be cracked.
Indeed the tensile stress limi t for Class T is higher than the modulus of rupture of
concrete and thus is truly fictitious. In Class C a cracked section analysis is to be
carried out. While the recornmendation of the code implies that a cracked section
Chapter 3 - THE PHILOSOPHY OF DESIGN 125
analysis may not be necessary for Class T sections, the author believes that a cracked
section analysis may be necessary, depending on the conditions of the structure. For
instance, an inverted T beam subjected to loads leading to a Class T member is likely
to violate sorne serviceability limit states related to crack width and fatigue, and
should therefore be analyzed as a cracked section.
In this book, mostly fully prestressed concrete (Class U) is considered for which
uncracked sections are the norm. However, partially prestressed concrete beams are
treated in Ref. [ 1.48] where details for a cracked section analysis are covered. Table
3.11 contains sorne other serviceability recommendations given in the code, such as
whether to use the gross section or the cracked section to compute deflections; they
are self-explanatory and will not be expanded upon here. The reader is referred to
the code for additional details.
The ACI code specifies a capacity reduction factor, </J = 0.9, for beams (in pure
bending) larger than that for columns (</J = 0.70 or 0.65) in which primarily axial
compression exists (see Section 3.9.2). However most columns are also subjected to
sorne bending, and the code suggests a transition from compression to bending when
the axial compression is smaller than ten percent of the nominal resistance of the
column in compression, i.e., 0.1 Of'cAg. This transition is achieved by evaluating the
net tensile strain in the extreme layer of reinforcement closest to the tensile fiber.
--~J
Spiral: Other:
(a)
q:l= 0.70+0.2[-1
el d1 3
q:l = 0.65 + 0.25[-1-
el d1
-~J
3
(b)
Figure 3.8 Definition of net tensile strain for a prestressed concrete beam and corresponding
;, factor according to the 2002 ACI code.
126 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
The net tensile strain is defined as the tensile strain in the extreme tension steel at
nominal strength, exclusive of strains due to prestress, creep, shrinkage and
temperature (Fig. 3.8a). When the net tensile strain in the extreme tension steel is
equal to or greater than 0.005, the section is defined as tension-controlled. In that
case the ¡p factor is taken as 0.9, according to Table 3.1 O.
For Grade 60 reinforcement and for all prestressed concrete reinforcement, a
section is assumed to be compression-controlled when the net tensile strain in the
extreme tension steel is equal to or less than 0.002. In that case the ¡p factor is taken
as 0.70 or 0.65 depending on whether the compression member is spirally reinforced
or not. For sections in which the net tensile strain in the extreme tension steel at
nominal strength is between the limits for compression-controlled and tension-
controlled sections, ¡p shall be permitted to be linearly increased from that of
compression controlled sections to 0.90, since the net tensile strain in the extreme
tension steel at nominal strength increases from the compression-controlled strain
limit to 0.005 (Fig. 3.8b). Equivalently, a similar approach is used in compression
members when the axial compression decreases from </iPn = O. IQf',Ag to zero where
pure bending occurs.
Assuming the ¡p factors defined by ACI in Table 3 .1 O are u sed, linear equations
can be derived to determine the ¡p factors in the transition region. They are given in
Fig. 3.7b in terms of the net tensile strain; however, they can also be expressed, as
also shown, in terms of the ratio c/d, (which is proportional to the net tensile strain)
where e represents the depth of neutral axis at ultimate and d, represents the distance
from the extreme compression fiber of concrete to the extreme tension layer of
reinforcement; so, if there are two layers of prestressing tendons, d, is taken to the
]ayer closest to the surface.
To allow moment redistribution of negative moment (see Sections 5.21 and
10.14.3) a net tensile strain limit of at least 0.0075 is recommended by the code.
A discussion on the interpretation of the above provisions and their use as
amended in this text is given in Section 5.5.3.
The designer should be able to see the whole and the detail. Although this text is not
intended to cover detailing procedures, a mínimum number of details is necessary in
the early stages of design. Once the required amount of reinforcement is determined,
it is the designer's responsibility to make certain that such reinforcement can be
properly placed inside the concrete section. Information related to spacing of the
tendons and the mínimum protection or clear concrete cover of the reinforcement is
needed.
Section 7.6 of the ACI code prescribes a clear distance between pretensioning
tendons at each end of a member not less than four times the diameter for wires and
three times the diameter for strands. Closer vertical spacing and bundling of strands
Chapter 3 - THE PHILOSOPHY OF DESIGN 127
posttensioned tendons at the ends of a member where the anchors are placed. In
general, additional stirrups are recommended near the ends of the member (with or
without end blocks) and confining reinforcements in the form of spirals or grids are
placed in the concrete directly behind the anchors.
A conservative rule of thumb ( or heuristic rule) is to place the end anchors in
such a way that the average stress in the concrete directly behind the anchor plate is
less than 0.80f'ci (up to f'ci is often acceptable in confined concrete). A more
representative approach can be found in Section 4.17.
Prestressing tendons must be placed inside a concrete section so as to have
sufficient protection. A summary of the ACI (2002) code provisions for the
mínimum clear concrete cover of the reinforcement in prestressed concrete is given
in Table 3.12. Note that the values are significantly smaller than those prescribed in
previous editions of the code. Engineering judgment should always be exercised for
any particular conditions.
It has been this author's experience that in dealing with prestressed pretensioned
beams where tendons of not more than 0.6 in (15 mm) diameter are used, a cover of
2 inches (50 mm) to the centroid of the tendon and a center-to-center spacing of 2
inches (50 mm) between tendons commonly lead to a feasible and reasonable design
(Fig. 3.9). Note that a center-to-center spacing of two inches (50 mm) leaves a clear
spacing of about 1 Yi in (38 mm), often just sufficient to allow the intemal vibration
of the concrete during casting.
ACI CODE:
~ 4db for wires
~ 3db for strands
~ 4/3 maximum aggregate size
.-------.Rule ofthumb:
J = 2 in for db ;S; 0.5 in
l. = 50 mm for db ;S; 13 mm
J = 2 in (50 mm) for beams
1,
1-..........., .. l. = 1 ~ in (38 mm) for slabs
depend on the capacity of the cable. Figure 3.9 summarizes information related to
spacing that would satisfy the ACI code as well as a number of other codes, while
very practica! to remember; it forms a grid of 2 in (50 mm) spacing in each direction.
As pointed out in Chapter 2, substantial losses of stress occur in the prestressing steel
due to relaxation of the steel and creep and shrinkage of the concrete. Other losses
also take place, due to elastic shortening of the concrete at load transfer and due to
friction between the tendons and the concrete. Several methods exist to estímate the
total loss of prestress and may require different levels of difficulty. A basic and
detailed treatment of prestress losses is given in Chapter 8. However, it is important
to estímate sorne prestress losses, in advance, in routine situations, in arder to
proceed with the design. Their values may later be revised if necessary.
Lump sum estimates of individual losses are given in many technical documents
(see References of Chap. 8). Lump sum values of total losses were suggested by the
ACI-ASCE Joint Committee on Prestressed Concrete [Ref. 3.6]. They were
recommended in previous editions of the ACI code and its commentary; however,
they were later found to be unconservative.
The AASHTO Standard specifications [Ref. 3 .1] suggest lump sum loss values
for bridges where average standard conditions prevail, while the AASHTO LRFD
specification provides more detailed estimates depending on the compressive
strength of concrete, the partial prestressing ratio and the shape of the section.
In preliminary design, there is need to estímate sorne of the losses that occur after
transfer of the stress from the prestressing steel to the concrete. Let us define by Ípi
the initial stress in the prestressing steel at time of load transfer and by Ípe the
effective stress remaining after all losses have taken place. The corresponding
prestressing forces are defined as F¡ and F. F¡ is the force that the concrete
experiences under initial loading. Thus Ípi may mean the stress just after transfer for
a pretensioned member and just befare transfer for a posttensioned member (see Eqs.
8.5 and 8.8). However, in preliminary design,Jp¡ is considered here to be the stress
just after transfer for all cases. The difference between Ípi and Ípe essentially
represents the time-dependent prestress losses after transfer. Thus, it <loes not
include elastic shortening or initial relaxation for a pretensioned member, or
frictional loss and anchorage set loss for a posttensioned member. In Table 3 .13 the
author provides for preliminary design a rough estímate of time-dependent losses
after transfer, that is, the difference between jj., andfpe. Additional information can
be obtained from reviewing Chapter 8. Note that in the preliminary design and
dimensioning of the structure the above difference betweenjj., andfpe is notas useful
as their ratio. Let us define:
130 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
Ípe F
r¡=-=- (3.19)
/» F¡
where F¡ is the initial and F = r¡F¡ the final or effective prestressing force. Equation
(3.19) can also be written as:
Since the stress in the steel generally varies along the length of the member, the
coefficient r¡ will also vary. However, r¡ is needed mostly at the critica} section of a
simply supported beam or at a limited number of critical sections of a continuous
beam. For all practica] purposes it could be assumed to be the same for all critica}
sections. As a first approximation in design the value of r¡ can be taken between
0.75 and 0.85. Indeed, assuming a 270 ksi (1860 MPa) stress-relieved strand with an
initial stress Ípi = 0.70Jpu, and using the estímate given in Table 3.13 for normal
weight concrete, leads to r¡ = 0.79. For a low relaxation strand with.,fp¡ = 0.74.,fpu, r¡
will be 0.825. Assuming a value of /p¡ = 0.70/pu, the corresponding value of the
effective prestress, Ípe. will be somewhere between 0.55/pu and 0.58/pu for normal
weight concrete. Such approximations will frequently be used in preliminary
designs. However, a more exact assessment of prestress losses is recommended for
final designs.
Table 3.13 Suggested lump sum estimates of time-dependent prestress losses after transfer of
prestress for routine design (not including friction and anchorage set effects).
([p¡ -Jpe)*
Stress relieved Low relaxation
strand strand
ksi MPa ksi MPa
Pretensioned members:
• Normal weight concrete 40 276 35 242
• Structural lightweizht concrete 45 310 40 276
Posttensioned members:
• Normal weight concrete 35 242 30 207
• Structural lightweizht concrete 40 276 35 242
* Assuming Ípi :'.':'. 0.82!'¡,y
Chapter 3 - THE PHILOSOPHY OF DESIGN 131
Note that for a given section design, under-estimating prestress losses affects
serviceability and allowable stress criteria, but has little effect on the ultimate
resistance of the member.
The limit state design philosophy covered in this chapter is not likely to change in
the foreseeable future. Similarly, the key features of prestressed concrete and its
characteristics in comparison to reinforced concrete will not change. However,
additional research, experience, and knowledge will likely lead to changes in load
and resistance factors, allowable stresses, code specifications, and construction
details. In design, engineering judgment should always be exercised even in dealing
with simple problems. Moreover, having a clear understanding of the fundamentals
will help in handling any difficult situation, especially if not addressed by available
codes of practice.
REFERENCES
3.15. British Standards Institution, British Standard Code of Practice for the Use of Structural
Concrete, CP-11 O, London, 1972.
3.16. Comité Euro-Intemational du Béton, CEB-FIP Model Codefor Concrete Structures, 3'd ed.,
Paris, 1990.
3.17. Ellingwood, B., T. V. Galambos, J. G. MacGregor, and C. A. Come 11, " Probability Based
Load Criteria: Load Factors and Load Combinations," Journal of Structural Division, ASCE,
108(ST5): 978-97, 1982.
3.18. Ellingwood, B., T. V. Galambos, J. G. MacGregor, and C. A. Comell, "Development of a
Probability Based Load Criterion for American National Standard A58," NBS Special
Publication 577, Washington, OC: National Bureau ofStandards, 1980.
3.19. Hill, P. H., The Science of Engineering Design. New York: Holt, Reinhart and Winston, lnc.,
1975.
3.20. MacGregor, J. G., "Safety and Limit States Design for Reinforced Concrete," Canadian
Journal of Civil Engineering, 3(4), 1976.
3.21. MacGregor, J. G., Design, Reinforced Concrete: Mechanics and Design, 3rd ed. Upper Saddle
River, NJ: Prentice Hall, 1997, 939 pp.
3.22. Nowak, A. S., and K. R. Collins, "Reliability of Structures," Boston: McGraw Hill, 2000, 338
pp.
3.23. Naaman, A. E., and A. Siriaksom, "Reliability of Partially Prestressed Beams at Serviceability
Limit States," Journal of the Prestressed Concrete lnstitute, 27(6): 66-85. 1982. Also, Report
No. 80-1, University of Illinois, Chicago, Department of Materials Engineering, June 1980,
126 pp.
3.24. Naaman, A. E., and A. M. Hamza, "Prestress Losses in Partially Prestressed High Strength
Concrete Beams," Journal of the Prestressed Concrete Jnstitute PCI Joumal, 38(3): 98-114,
1993.
3.25. Winter, G., "Safety and Serviceability Provisions in the ACI Building Code," ACI Special
Publication, SP-59, 1979.
3.26. White, R. N., P. Gergely, and R. G. Sexsmith, Structural Engineering. Vol. 1: lntroduction to
Design Concepts and Analysis. New York: John Wiley & Sons, 1972.
Chapter 3 - THE PHILOSOPHY OF DESIGN 133
Fox Hollow Pedestrian Bridge, Calgary, Canada. This two-span, 90 m (295 ft), symmetrical
cable stayed bridge used precast prestressed components for both the deck aod the tower A-
frame. (Courtesy Precast/Prestressed Concrete lnstitute.)
134 Naaman - PRESTRESSED CONCRETE ANAL YSIS ANO DESIGN
Arcadia parking structure illustrates the tlexibility and aesthetics acbievable using precast
prestressed concrete elements. (Courtesy Car/ Walker Jne.).
Erection of precast prestressed concrete elements for a parking structure. (Courtesy Car/ Walker
/ne.).