DYNA

http://dyna.medellin.unal.edu.co/

Stiffness matrix and loading vector of a prestressed concrete beam including long term effects: i)
theory
Abstract
The stiffness matrix and load vector of a prismatic prestressed-concrete beam with symmetrical cross section about its major axis and
subject to transverse loads along its span including long term effects and prestress losses are presented. The proposed formulation is based
on equilibrium conditions, strain compatibility, and constitutive laws of all materials involved. It also includes thermal strains and
deformations of both steel and concrete, as well as the long-term effects caused by creep and shrinkage of the concrete and relaxation of
the prestressed steel. In the development of the proposed formulation, the equivalent transversal load method is shown as a corollary. Three
comparative and verification examples are presented in a companion paper that shows the accuracy and simplicity of the proposed method
and corresponding equations.
Keywords: Beams; Creep; Losses; Prestressed concrete; Shrinkage; Stiffness matrix; Structural Analysis.

Resumen
Se presentan la matriz de rigidez y el vector de carga de una viga prismática de concreto pretensado con una sección simétrica alrededor
de su eje mayor y sujetas a cargas transversales a lo largo de su luz, incluidos los efectos de largo plazo y las pérdidas de pretensado. La
formulación propuesta se basa en condiciones de equilibrio, compatibilidad de deformación y leyes constitutivas de todos los materiales
involucrados. También incluye deformaciones térmicas y deformaciones tanto del acero como del concretodo. En el desarrollo de la
formulación propuesta, el método de carga transversal equivalente se muestra como un corolario. Tres ejemplos comparativos y de
verificación se presentan en un documento complementario que muestra la precisión y simplicidad del método propuesto y las ecuaciones
correspondientes.
Palabras clave: Vigas; Repteo; Pérdidas; Concreto pretensado; retraccion; Matrix de rigidez; Análisis estructural.

© The authors; licensee Universidad Nacional de Colombia.

DYNA, pp. 2019. Medellín. ISSN 0012-7353 Printed, ISSN 2346-2183 Online
1. Introduction
Prestressed concrete beam structures are currently analyzed using
either the basic method, the pressure-line method, or the
equivalent transverse load method. The benefits and difficulties
of each of these three methods are well known in the technical
literature [1]-[3]. For statically determinate structures, the
application of these three methods presents the same level of
difficulty. On the other hand, for statically indeterminate
structures, the rotations and displacements of the ends of each
member are evaluated first using the virtual work method in order
to determine the bending moment, shear and axial forces
diagrams along the span each element using the flexibility method
[4]. On the other hand, in the structural analysis of indeterminate
structures the redundant degrees of freedom are first released so
that the structure as a whole becomes statically determinate, then
compatibility conditions are applied at the locations of the
released degrees of freedom to determine the magnitude of the
redundant forces and moments. Superposition of all moments
including those caused by the tensioning process (primary
moments) along with the moments caused by the redundant forces
(secondary moments) results in the moments diagrams. Finally,
once the moment diagram for each beam has been found, the
method of the pressure line can be used to find the stresses across
in the beam section along its span.
However, it is possible to include the combined effects of losses
by friction, shrinkage of the concrete, and thermal expansion of
all materials in a rational and efficient manner as a routine in the
matrix software. The load vector presented herein includes a set
of equivalent loads of the equivalent transverse load method, that
is, uniform or concentrated transverse loads along the beam span
and the loads and moments at both ends of the beam.
2. Proposed model and equilibrium equations
The variations of the curvature and the axial deformation of the
centroidal axis of a prestressed beam with the imposed loads can
be obtained by applying equilibrium, compatibility of
deformations and constitutive laws of the materials along the
geometric characteristics of the beam cross section [1].
2.1 Strain Compatibility between concrete and reinforcing steels
As shown in Fig. 1, it is assumed that the beam cross sections
remain plane, therefore the strains of the concrete, steel bars, and
prestressed steel (i.e. 𝜀𝑐 , 𝜀𝑠 and 𝜀𝑝 , respectively) can be expressed
in terms of the beam curvature 𝜙 and the concrete strain at the
centroid of the cross section as follow:
𝜀𝑐 = 𝜀𝑐𝑒𝑛 − 𝜙𝑦 (1)

Another method used in the analysis of statically 𝜀𝑠 = 𝜀𝑐𝑒𝑛 − 𝜙𝑦 (2)

indeterminate prestressed structures is the equivalent load method
[4]-[8]. In this method the tensioning forces are replaced by 𝜀𝑝 = 𝜀𝑐𝑒𝑛 − 𝜙𝑦 + ∆𝑒𝑝 (3)
equivalent transverse loads along each member and it is
commonly used along with the moment distribution method (i.e, Where: ∆𝑒𝑝 = initial strain or basic “signature” of the prestressing
the Cross method) to determine the moment diagrams of each operation imposed to the prestressed steel.
member.
2.2 Constitutive laws of each material
Losses in the tensioning forces of the prestressed cables along The concrete, steel rebars and prestressed cables are assumed to
each member must be taken into account in the analysis and be linear elastic as indicated by Eqs. (4)-(6) and Fig. 1.
design of prestressed concrete members due to the fact that they
can become significant in their actual behavior. Generally, the
losses caused by friction between the cables and the duct and
those caused by curvature of the cable affect the behavior and load
capacity of prestressed members. In the technical literature, it is
common to assume that the total loss is the sum of these two
components. The losses which are time-dependent, such as those
due to creep and shrinkage of concrete and relaxation of
prestressing steel, can be considered in the proposed matrix
method simply by modifying the modulus of elasticity of the
materials involved. The combined effects of losses caused by
friction, shrinkage and temperature in statically indeterminate
structures such as continuous beams and frames have been
generally treated in the technical literature using a compact
formulation. This is the main reason why the work presented
herein deals with the stiffness matrix and load vector (i.e., fixed
end moments and forces) of a prestressed beam element.
FIG. 1 Symmetric cross-section: Strain and Stress
It is assumed that the profile of the prestressed steel cable can distributions, and linear elastic materials
be defined by second-order polynomials or parabolas along each
member span. The proposed stiffness matrix and load vector can Concrete: 𝒇𝒄 = 𝑬𝒄 𝜺𝒄𝒇 ; 𝜺𝒄𝒇 = 𝜺𝒄 − 𝜺𝒄𝒐 (4)
be used together in a classical manner in the analysis of two- Steel rebars: 𝒇𝒔 = 𝑬𝒔 𝜺𝒔𝒇 ; 𝜺𝒔𝒇 = 𝜺𝒔 − 𝜺𝒔𝒐 (5)
dimensional indeterminate prestressed concrete structures. The
Prestressed steel: 𝒇𝒑 = 𝑬𝒑 𝜺𝒑𝒇 ; 𝜺𝒑𝒇 = 𝜺𝒑 − 𝜺𝒑𝒐 (5)
proposed method is based on the basic laws of compatibility of
deformations, constitutive laws of materials and static where: 𝜀𝑐𝑒𝑛 = strain along the beam centroidal axis;
equilibrium to find the global stiffness matrix and load vector of
complex framed structures. Unlike the flexibility method 𝜙= Curvature along the beam centroidal axis;
mentioned above, when the proposed method is used, it is not
necessary to find the primary- and secondary-moment diagrams 𝑐, 𝑠, 𝑝: Sub-indexes referring to concrete, steel bars and
caused by the prestressed steel cables. Similar to the method of prestressed steel, respectively;
the equivalent transversal load, the moment diagrams obtained
are directly those produced by the tensioning of the cables.

© The authors; licensee Universidad Nacional de Colombia.

 𝐸𝑐 , 𝐸𝑠 , 𝐸𝑝 : Elastic moduli of concrete, steel rebars and Likewise, using Eqs. (1)-(6), the bending moment M and the
prestressed steel, respectively; curvature 𝝓 of the beam are:
𝑀 = − ∫𝐴 𝑓𝑐 𝑦𝑑𝐴𝑐 − ∫𝐴 𝑓𝑠 𝑦𝑑𝐴𝑠 − ∫𝐴 𝑓𝑝 𝑦𝑑𝐴𝑝 = 𝐸𝑐 𝜙𝐼𝑡𝑟 + 𝑀𝑜
𝜀𝑐 , 𝜀𝑠 , 𝜀𝑝 : Net strains of concrete, steel rebars and 𝑐 𝑠 𝑝

prestressed steel, respectively;

(9)
𝜀𝑐𝑓 , 𝜀𝑠𝑓 , 𝜀𝑝𝑓 : Strains caused by the applied forces,
𝑑2𝑦 𝑀−𝑀𝑜
respectively; 𝜙= =
𝑑𝑥 2 𝐸𝑐 𝐼𝑡𝑟

𝜀𝑐𝑜 = 𝜀𝑐𝑠ℎ + 𝜀𝑐𝑡ℎ : Concrete Strains caused by shrinkage (10)

and temperature, respectively; 𝐸𝑠 𝐸𝑝
where: 𝐼𝑡𝑟 = ∫𝐴 𝑦 2 𝑑𝐴𝑐 + ∫𝐴 𝑦 2 𝑑𝐴𝑠 + ∫𝐴 𝑦 2 𝑑𝐴𝑝 ; and
𝑐 𝑠 𝐸𝑐 𝑝 𝐸𝑐
𝜀𝑠𝑜 = 𝜀𝑠𝑡ℎ : Strains in the steel bars caused by
temperature;
𝑀𝑜 = − ∫ 𝐸𝑝 ∆𝑒𝑝 𝑦𝑑𝐴𝑝 + ∫ 𝐸𝑐 𝜀𝑐𝑜 𝑦𝑑𝐴𝑐 + ∫ 𝐸𝑠 𝜀𝑠𝑜 𝑦𝑑𝐴𝑠
𝐴𝑝 𝐴𝑐 𝐴𝑠
𝜀𝑝𝑜 = 𝜀𝑝𝑡ℎ : Strains in the prestressed steel caused by
temperature. + ∫ 𝐸𝑝 𝜀𝑝𝑜 𝑦𝑑𝐴𝑝
𝐴𝑝
The effects of creep on the concrete elastic modulus are based
on experimental data [1] according to the following expression: Itr and Mo are generally referred in the technical literature [1] as
𝐸𝑐𝑖 the second moment of inertia of the beam transformed cross-
𝐸𝑐,𝑒𝑓𝑓 = ) 1+𝜙(𝑡,𝑡𝑖 section and moment of decompression, respectively. Eqs. (8) and
(10) relate the axial strain and the curvature of the beam in terms
𝐻 (𝑡−𝑡𝑖 )0.6
where: 𝜙(𝑡, 𝑡𝑖 ) = 3.5𝑘𝑐 𝑘𝑓 (1.58 − ) 𝑡𝑖−0.118 ; of the applied axial load N and bending moment M and along the
120 10+(𝑡−𝑡𝑖 )0.6
1 beam longitudinal centroidal axis. The solution of these two
𝑘𝑓 = (with 𝑓𝑐 in psi) equations allows to establish the behavior of the beam at the
0.67+(𝑓𝑐` /9000)
sectional level. To determine the behavior of the beam, both at the
1 sectional level and for any section along its span, the following
𝑘𝑓 = (with 𝑓𝑐 in Mpa); and 𝑘𝑐 = volume-to-
0.67+( 𝑓𝑐` /62) equations of equilibrium of the differential element must be
surface ratio factor and H is the relative humidity in percent, [1]. satisfied:
The effect of relaxation of the prestressed steel on its elastic 𝑑𝑁
𝑓𝑝 =0 (11)
𝑑𝑥
modulus 𝐸𝑝 is assumed to be as follows, [1]: 𝐸𝑝,𝑒𝑓𝑓 = 𝐸𝑝 ,
𝑓𝑝𝑖
𝑓𝑝 log(𝑡) 𝑓𝑝𝑖 𝑑𝑉
where: =1− ( − 0.55) = −𝑞 (12)
𝑓𝑝𝑖 45 𝑓𝑝𝑦 𝑑𝑥

𝑑𝑀
2.3 Equilibrium at the beam section level =𝑉 (13)
𝑑𝑥
The equilibrium of horizontal forces and bending moments at the
beam section level is presented. The axial force 𝑁𝑜 and the 3. Profile of the prestressed steel and losses in the initial
moment of decompression 𝑀𝑜 are defined below. Note that the tension forces
subscript s used in this formulation refers to the conventional steel
rebars, but it could also be used for any type of linear elastic It is assumed that the eccentricity of the prestressed steel tendon
reinforcement such as bars or sheets made of fiber reinforced (i.e., its profile) measured with respect to the neutral axis of the
polymers (GFRP, CFRP). transformed section is assumed to be parabolic as follows:

2.3.1 Longitudinal equilibrium 𝑒(𝑥) = 𝛼1 + 𝛼2 𝑥 + 𝛼3 𝑥 2 (14)

Using Eqs. (1)-(6), the net axial force N and the corresponding
axial strain 𝜀𝑐𝑒𝑛 at the centroid of the beam can be found as
follows:

(0) m L
𝑁 = ∫𝐴 𝑓𝑐 𝑑𝐴𝑐 + ∫𝐴 𝑓𝑠 𝑑𝐴𝑠 + ∫𝐴 𝑓𝑝 𝑑𝐴𝑝 = 𝐸𝑐 𝜀𝑐𝑒𝑛 𝐴𝑡𝑟 + 𝑁𝑜 o
𝑐 𝑠 𝑝
(6)
e'(x) X

 e'(x) _ e'(0)
𝑑𝑢 𝑁−𝑁𝑜
𝜀𝑐𝑒𝑛 = =
𝑑𝑥 𝐸𝑐 𝐴𝑡𝑟
(8)
x
𝐸𝑠 𝐸𝑝
where: 𝐴𝑡𝑟 = ∫𝐴 𝑑𝐴𝑐 + ∫𝐴 𝑑𝐴𝑠 + ∫𝐴 𝑑𝐴𝑝 ; and 𝑁𝑜 =
𝑐 𝑠 𝐸𝑐 𝑝 𝐸𝑐
∫𝐴 𝐸𝑝 ∆𝑒𝑝 𝑑𝐴𝑝 − ∫𝐴 𝐸𝑐 𝜀𝑐𝑜 𝑑𝐴𝑐 − ∫𝐴 𝐸𝑠 𝜀𝑠𝑜 𝑑𝐴𝑠 −
𝑝 𝑐 𝑠

∫𝐴 𝐸𝑝 𝜀𝑝𝑜 𝑑𝐴𝑝
𝑝
FIG. 2 Assumed parabolic profile of the prestressed steel tendon
Atr and No are generally referred in the technical literature [1] as The coefficients 𝛼1 , 𝛼2 and 𝛼3 in Eq. (14) can be determined
the transformed area of the beam cross-section and the by the eccentricities at both ends and at x= 𝜆L (0 < 𝜆 < 1) of the
decompression axial force, respectively. beam as shown in Fig. 2. Appendix I presents useful relationships

2.3.2 Rotational equilibrium

3
for the determination of cable eccentricities used in multi-span
beams. According to Fig. 2:
𝑒(0) = 𝛼1 = 𝑒0 ; 𝑒(𝜆𝐿) = 𝛼1 + 𝛼2 (𝜆𝐿) + 𝛼3 (𝜆𝐿)2 = 𝑒𝑚 ; and
𝑒(𝐿) = 𝛼1 + 𝛼2 𝐿 + 𝛼3 𝐿2 = 𝑒𝐿
Eq. (17) is a second-order differential equation that governs
the elastic axial behavior of a beam subject to an axial force from
a prestressed cable along its span including the cable force losses
caused by friction and wobble effects. The solution to Eq. (17) is
as follows:

𝛼1 𝐿3 𝜆 − 𝐿3 𝜆2 0 0 𝑒0 𝑒𝑥𝑝𝑓𝑥𝑃0
1 𝑢 = 𝑑1 + 𝑑2 𝑥 − (18)
Therefore: [𝛼2 ] = 3 3 2 [ 𝐿2 𝜆2 − 𝐿2 𝐿2 −𝐿2 𝜆2 ] [𝑒𝑚 ] 𝐸𝐴𝑓
𝐿 𝜆−𝐿 𝜆
𝛼3 𝐿 − 𝐿𝜆 −𝐿 𝐿𝜆 𝑒𝐿
Substituting (18) into Eq. (16):
The losses in the initial tension forces along the prestressed
tendon caused by the combined effects of friction and curvature 𝑁 = 𝐸𝐴𝑑2 (19)
[1] along the beam are generally calculated using the following
expression: Using Eqs. (18) and (19), the boundary conditions at x = 0
and x = L become:
𝑃(𝑥) = 𝑃0 𝑒𝑥𝑝−(𝜇𝛼+𝐾𝑥)
𝑁(0) = 𝑁𝑎 = −𝐸𝐴𝑑2 ; 𝑁(𝐿) = 𝑁𝑏 = 𝐸𝐴𝑑2 ; and
where: 𝑃0 = Axial force in the steel tendon at x = 0; 𝑃0
𝑢(0) = 𝑢𝑎 = 𝑑1 − ; 𝑢(𝐿) = 𝑢𝑏 = 𝑑1 + 𝑑2 𝐿 −
𝐸𝐴𝑓
𝜇 = Friction coefficient (0.05 ≤ 𝜇 ≤ 0.5). See Table 2- 𝑒𝑥𝑝𝑓𝐿𝑃0
2 of Ref. [1]; 𝐸𝐴𝑓

𝛼(𝑥) = 𝑒 ′ (𝑥) − 𝑒 ′ (0) = 2𝛼3 𝑥, angle made by the These four end conditions can be expressed in matrix form as
tangent of the steel cable profile at x with respect to that the left shown by Eqs. (20) and (21):
support of the beam (see Fig. 2); and
𝑁 0 −𝐸𝐴 𝑑1
[ 𝑎] = [ ][ ] (20)
𝐾= Wobble coefficient per unit length of steel tendon. 𝑁𝑏 0 −𝐸𝐴 𝑑2
See Table 2-2 of Ref. [1].
𝑢𝑎 1 0 𝑑1 𝑃 1
[𝑢 ] = [ ][ ] − 0 [ ] (21)
Therefore: 𝑃(𝑥) = 𝑃0 𝑒𝑥𝑝 𝑓𝑥 (15) 𝑏 1 𝐿 𝑑2 𝐸𝐴𝑓 𝑒𝑥𝑝 𝑓𝑙

where: 𝑓 = −(2𝜇𝛼3 + 𝐾) Combining Eqs. (20) and (21), Eq. (22) can be obtained:

4. Governing diferential equations and their solutions 𝑁𝑎 𝐸𝐴 1 −1 𝑢𝑎 𝑃 1

[ ]= [ ] [ ] + 0 (1 − 𝑒𝑥𝑝 𝑓𝐿 ) [ ] (22)
Fig. 3 shows the 6-DOF of the prestressed beam and the local 𝑁𝑏 𝐿 −1 1 𝑢𝑏 𝑓𝐿 −1
coordinates (𝑥̅ , 𝑦̅, 𝑧̅). The governing differential equations for the
axial degrees of freedom (1 and 4), transverse and rotational 4.2 Solution of governing differential equation of flexural
degrees of freedom (2, 3, 5 and 6) at the ends of the beam. Their deformations
respective solutions and corresponding forces and bending Assuming that 𝜀𝑐𝑜 , 𝜀𝑠𝑜 and 𝜀𝑝𝑜 are zero, the moment of
moments are presented and discussed in the next two sections. decompression Mo becomes:

𝑀𝑜 = − ∫ 𝐸𝑝 ∆𝑒𝑝 𝑦𝑑𝐴𝑝 = −𝐸𝑝 ∆𝑒𝑝 𝑒 ∫ 𝑑𝐴𝑝 = −𝐸𝑝 ∆𝑒𝑝 𝐴𝑝 𝑒

𝐴𝑝 𝐴𝑝
= −𝑃(𝑥)𝑒(𝑥)

Using Eq. (10):

FIG. 3 The 6-DOF and corresponding local coordinates of the
prestressed beam 𝑑2𝑦
𝐸𝐼 = 𝑀 − 𝑀𝑜 ≡ 𝑀𝑛𝑒𝑡 (23)
𝑑𝑥 2
4.1 Solution of the differential equation of axial deformations
Differentiating Eq. (23) with respect to x once and twice and
Assuming that 𝜀𝑐𝑜 , 𝜀𝑠𝑜 and 𝜀𝑝𝑜 are zero, the decompression axial taking into account the static equilibrium of the differential
force No becomes: 𝑁𝑜 = ∫𝐴 𝐸𝑝 ∆𝑒𝑝 𝑑𝐴𝑝 = 𝐸𝑝 ∆𝑒𝑝 ∫𝐴 𝑑𝐴𝑝 = element according to Eqs. (12) and (13), the following
𝑝 𝑝 expressions for 𝑉𝑛𝑒𝑡 and 𝑞𝑛𝑒𝑡 can be obtained:
𝑃(𝑥) = 𝑃0 𝑒𝑥𝑝 𝑓𝑥
𝑑3𝑦 𝑑𝑀𝑜
𝐸𝐼 =𝑉− ≡ 𝑉𝑛𝑒𝑡 (24)
and substituting this expression into Eq. (8), the axial strain along 𝑑𝑥 3 𝑑𝑥
the beam span becomes:
𝑑4𝑦 𝑑 2 𝑀𝑜
𝐸𝐼 = −𝑞 − ≡ 𝑞𝑛𝑒𝑡 (25)
𝑑𝑢 𝑁−𝑃0 𝑒𝑥𝑝𝑥𝑓 𝑁𝑛𝑒𝑡 𝑑𝑥 4 𝑑𝑥 2
= ≡ (16)
𝑑𝑥 𝐸𝑐 𝐴𝑡𝑟 𝐸𝑐 𝐴𝑡𝑟
where: 𝑀𝑜 = −𝑃0 𝑒𝑥𝑝 𝑓𝑥 (𝛼1 + 𝛼2 𝑥 + 𝛼3 𝑥 2 )
Where the beam net axial force 𝑁𝑛𝑒𝑡 = 𝑁 − 𝑃0 𝑒𝑥𝑝 𝑥𝑓
Eqs. (23) and (24) define the net bending moment and the net
Deriving Eq. (16) with respect to x, using Eq. (11) and making transverse shear force along the beam span. Whereas Eq. (25)
𝐴𝑡𝑟 = 𝐴, 𝐼𝑡𝑟 = 𝐼 and 𝐸𝑐 = 𝐸: shows that the net load on the structure is the sum of two
components: the applied transverse load q and that related to the
𝑑2𝑢 𝑒𝑥𝑝𝑓𝑥𝑓𝑃0 tensioning load of the cables. It is interesting to note that Eqs.
=− (17) (16), (23)-(25) when used in the analysis of prestressed concrete
𝑑𝑥 2 𝐸𝐴

4
beams gives identical results to those obtained by the equivalent 𝑦𝑎
transverse load method proposed by T.Y. Lin [5]. 𝜃𝑎
[𝑦 ]
𝑏
Finally, Eq. (25) can be expressed as follows: 𝜃𝑏
1 0 0 0 𝑐1
𝑑4𝑦 𝑒𝑥𝑝𝑓𝑥 2 1 0 0 0 𝑐2
𝐸𝐼 = −𝑞 + 𝑃0 ((𝑓 𝛼1 + 2𝑓𝛼2 + 2𝛼3 ) + (𝑓 2 𝛼2 + =[ ][ ]
𝑑𝑥 4 𝐸𝐼 1 𝐿 𝐿2𝐿3 𝑐3
4𝑓𝛼3 )𝑥 + 𝑓 2 𝛼3 𝑥 2 ) (26) 0 1 3𝐿 𝑐4
2𝐿 2
𝑃0
(𝑓 2 𝛼1 − 2𝑓𝛼2 + 6𝛼3 )
Eq. (26) is the governing differential equation of the 𝐸𝐼𝑓 4
transverse deflection y of a R/C beam of symmetrical cross 𝑃0
section (Fig. 1) subject to uniform transversal load q as well as to (𝑓 2 𝛼1 − 𝑓𝛼2 + 2𝛼3 )
𝐸𝐼𝑓 3
a prestressed force applied by bonded steel cables with a parabolic +
profile including the prestressed force losses along its span. Note 𝐿4 𝑞 𝑒𝑥𝑝 𝑓𝐿 𝑃0 2
− + (𝑓 𝛼1 + 𝑓(𝑓𝐿 − 2)𝛼2 + (6 − 4𝑓𝐿 + 𝑓 2 𝐿2 )𝛼3 )
that the assumed cable profile can be used to model horizontal 24𝐸𝐼 𝐸𝐼𝑓 4
straight cables, inclined straight cables and parabolic cables. Eq. 𝐿3 𝑞 𝑒𝑥𝑝 𝑓𝐿 𝑃0 2
(26) is a linear non-homogeneous 4th-order differential equation − + (𝑓 𝛼1 + 𝑓(𝑓𝐿 − 1)𝛼2 + (2 − 2𝑓𝐿 + 𝑓 2 𝐿2 )𝛼3 )
[ 6𝐸𝐼 𝐸𝐼𝑓 3 ]
with constant coefficients whose full solution is as follows:
or {𝛿} = [𝑅]{𝑐} + {𝐹} (29)
𝑞𝑥 4 𝑒𝑥𝑝𝑓𝑥 𝑃0
𝑦 = 𝑐1 + 𝑐2 𝑥 + 𝑐3 𝑥 2 + 𝑐4 𝑥 3 − + ((𝑓 2 𝛼1 −
24𝐸𝐼 𝐸𝐼𝑓4
Where: {𝑐} = {𝑐1 𝑐2 𝑐3 𝑐4 }𝑇
2𝑓𝛼2 + 6𝛼3 ) + (𝑓 2 𝛼2 − 4𝑓𝛼3 )𝑥 + 𝑓 2 𝛼3 𝑥 2 ) (27)
Substituting (29) into Eq. (28), the following matrix equation
Using Eqs. (24), (23) and (27) the following expressions for is obtained:
the shear V and the bending moment M can be obtained: 𝑉 =
𝑞𝑥 2
6𝐸𝐼𝑐4 − 𝑞𝑥 and 𝑀 = 2𝐸𝐼𝑐3 + 6𝐸𝐼𝑐4 𝑥 − . Therefore, the {𝑀} = [𝐾𝑚 ]{𝛿} + {𝑀𝑚 } (30)
2
shear force and bending moment at x = 0 and x = L become: 𝑉𝑎 =
6𝐸𝐼𝑐4 ; 𝑀𝑎 = −2𝐸𝐼𝑐3 ; 𝑉𝑏 = −6𝐸𝐼𝑐4 + 𝑞𝐿; and 𝑀𝑏 = where: [𝐾𝑚 ] = [𝑆𝑅−1 ], {𝑀𝑚 } = {𝐽} − [𝐾𝑚 ]{𝐹}
𝑞
2𝐸𝐼𝑐3 + 6𝐸𝐼𝑐4 𝐿 − 𝐿2 These values can be expressed in matrix
2 5. Stiffness matrix and loading vector in local coordinates
form as follow:
Combining Eqs. (22) and (30) into a single matrix expression (see
𝑉𝑎 0 0 0 6 𝑐1 0 Appendix I) the following matrix equation can be obtained:
𝑀𝑎 0 0 0 −2 𝑐2 0
[ ] = 𝐸𝐼 [ ] [𝑐 ] + 𝑞𝐿
𝑉𝑏 0 0 0 −6 3 𝑁𝑎 𝑘𝑝 0 0 −𝑘𝑝 0 0 𝑥𝑎 𝑚𝑃1
𝑞 2
𝑀𝑏 0 0 2 6𝐿 𝑐4 [− 2 𝐿 ] 𝑉𝑎 0 𝑘11 𝑘12 0 𝑘13 𝑘14 𝑦𝑎 𝑚𝑃2
𝑀𝑎 0 𝑘21 𝑘22 0 𝑘23 𝑘24 𝜃𝑎 𝑚𝑃3
= + 𝑚
or {𝑀} = [𝑆]{𝑐} + {𝐽} (28) 𝑁𝑏 −𝑘𝑝 0 0 𝑘𝑝 0 0 𝑥𝑏 𝑃4
𝑉𝑏 0 𝑘31 𝑘32 0 𝑘33 𝑘34 𝑦𝑏 𝑚𝑃5
Using Eq, (27) the displacements and rotations at x= 0 and x= L [𝑀𝑏 ] [ 0 𝑘41 𝑘42 0 𝑘43 𝑘44 ] [𝜃𝑏 ] [𝑚𝑃6 ]
become:
Using the local coordinate system of the beam, this last equation
𝑃0 can be expressed in condensed form as follows:
𝑦𝑎 = 𝑐1 + (𝑓 2 𝛼1 − 2𝑓𝛼2 + 6𝛼3 )
𝐸𝐼𝑓 4
{𝐹̅ } = [𝐾
̅]{𝑋̅} + {𝐹̅0 } (31)
𝑃0
𝜃𝑎 = 𝑐2 + (𝑓 2 𝛼1 − 𝑓𝛼2 + 2𝛼3 )
𝐸𝐼𝑓 3 Where the matrix [𝐾 ̅] and the vector {𝐹̅0 } represent the stiffness
matrix and the load vector (fixed end actions) respectively in local
𝐿4 𝑞 coordinates.
𝑦𝑏 = 𝑐1 + 𝑐2 𝐿 + 𝑐3 𝐿2 + 𝑐4 𝐿3 −
24𝐸𝐼
𝑒𝑥𝑝 𝑓𝐿 𝑃0 2 5.1 Stiffness matrix and loading vector in global coordinates-.
+ (𝑓 𝛼1 + 𝑓(𝑓𝐿 − 2)𝛼2 Eq. (31) can be transformed from the local to the global
𝐸𝐼𝑓 4
+ (6 − 4𝑓𝐿 + 𝑓 2 𝐿2 )𝛼3 ) coordinate system by means of the appropriate transformation
matrix as follows:
𝐿3 𝑞
𝜃𝑏 = 𝑐2 + 2𝑐3 𝐿 + 3𝑐4 𝐿2 − {𝐹} = [𝐾]{𝑋} + {𝐹0 } (32)
6𝐸𝐼
𝑒𝑥𝑝 𝑓𝐿 𝑃0 2
+
𝐸𝐼𝑓 3
(𝑓 𝛼1 + 𝑓(𝑓𝐿 − 1)𝛼2 where: ̅][𝑇]−1 ;
[𝐾] = [𝑇][𝐾 {𝐹0 } = [𝑇]{𝐹̅0 }; [𝑇] = [𝜆 0
];
0 𝜆
+ (2 − 2𝑓𝐿 + 𝑓 2 𝐿2 )𝛼3 ) 𝑐 −𝑠 0
𝜆 = [𝑠 𝑐 0]; 𝑐 = 𝑐𝑜𝑠𝜃; and 𝑠 = 𝑠𝑖𝑛𝜃
These displacements and rotations can be expressed in matrix 0 0 1
form as follows:
5.2 Net axial force, shear and bending moment diagrams-. Once
the integration constants of each element are known, the axial
force, shear and net bending moment can be obtained from Eqs.
(16), (24) and (23) as follow:

𝑁𝑛𝑒𝑡 = 𝐸𝐴𝑑2 − 𝑒𝑥𝑝 𝑓𝑥 𝑃0 (33)

5
 𝑉𝑛𝑒𝑡 = 6𝐸𝐼𝑐4 − 𝑞𝑥 + 𝑒𝑥𝑝 𝑓𝑥 𝑃0 (𝑓𝛼1 + (1 + 𝑓𝑥)𝛼2 + 𝑥(2 + Loading Vector including force losses in a prismatic
𝑓𝑥)𝛼3 ) (34) prestressed beam
𝑞
𝑀𝑛𝑒𝑡 = 2𝑐3 𝐸𝐼 + 6𝑐4 𝐸𝐼𝑥 − 𝑥 2 + 𝑒𝑥𝑝 𝑓𝑥 𝑃0 (𝛼1 + 𝑥𝛼2 + 𝑚𝑃1 0
2 6𝐿
2
𝑥 𝛼3 ) (35) 𝑚𝑃2
𝑚 𝑞 𝐿2
{𝐹̅0 } = 𝑚𝑃3 = 0
5.3 Stress Diagrams-. The stress diagrams of the top and bottom 𝑃4 12
𝑚𝑃5 6𝐿
fibers of the beam cross-section can be determined from Eqs. (36)
{𝑚𝑃6 } −𝐿2
and (37): { }
𝑁𝑛𝑒𝑡 𝑀𝑛𝑒𝑡
𝐴1
𝜎𝑏 = + (36) −6𝐿𝐴2 − 12𝐴3 + 12𝐴5 − 6𝐿𝐴6
𝐴 𝑍𝑏
2 2
𝑁𝑛𝑒𝑡 𝑀𝑛𝑒𝑡
𝑃0 −4𝐿 𝐴2 − 6𝐿𝐴3 + 6𝐿𝐴5 − 2𝐿 𝐴6
𝜎𝑡 = − (37) + 3 𝐴4
𝐴 𝑍𝑡 𝐿
6𝐿𝐴2 + 12𝐴3 − 12𝐴5 + 6𝐿𝐴6
Where: 𝑍𝑡 , 𝑍𝑏 are the top fiber and bottom fiber modules of the −2𝐿2 𝐴2 − 6𝐿𝐴3 + 6𝐿𝐴5 − 4𝐿2 𝐴6
beam cross section, respectively. { }

𝐿2 (𝑓2 𝛼1 −𝑓𝛼2 +2𝛼3 )

Summary and conclusions Where: 𝐴1 = (1 − 𝑒𝑥𝑝 𝑓𝐿 ); 𝐴2 = ; 𝐴3 =
𝑓 𝑓3
The stiffness matrix, the load vector and the transfer functions of (𝑓2 𝛼1 −2𝑓𝛼2 +6𝛼3 ) 𝐿2
a prestressed concrete beam with a tendon that follows a parabola ; 𝐴4 = − (1 − 𝑒𝑥𝑝 𝑓𝐿 ); 𝐴5 =
𝑓4 𝑓
of second order subject to a uniform transverse load q are 𝑒𝑥𝑝𝑓𝐿 (𝑓2 𝛼 2 2
1 +𝑓(−2+𝑓𝐿)𝛼2 +(6−4𝑓𝐿+𝑓 𝐿 )𝛼3 )
; and 𝐴6 =
presented in detail. The proposed method and corresponding 𝑓4
𝑒𝑥𝑝𝑓𝐿 (𝑓2 𝛼1 +𝑓(−1+𝑓𝐿)𝛼2 +(2−2𝑓𝐿+𝑓2 𝐿2 )𝛼3 )
equations can be used for the analysis indeterminate structures
𝑓3
made of prestressed continuous beams with prestressed or post-
tensioned tendons with horizontal, inclined straight lines and
Loading Vector excluding prestressed force losses of a
parabolic profiles. The proposed method is based on the
prismatic prestressed beam
constitutive laws of the materials, compatibility of deformations
and static equilibrium. The effects caused by force losses in the
The fixed end moments and forces of a prestressed beam
tendons, thermal deformation and shrinkage have been included
neglecting the losses of the applied prestressed force (i.e., f= 0)
in the proposed method. It is shown that as f tends to zero (i.e.,
are as follow:
when the effects of the force losses in the tendons are negligible)
the stiffness matrix and the load vector correspond to that of a −𝑃0
regular reinforced concrete beam while effects of concrete creep 𝐿𝑞
and relaxation of the tensioning steel are taken into account − 𝑃0 (𝛼2 + 𝐿𝛼3 )
rationally by affecting their modulus of elasticity. 2
2
Equations are presented to calculate the diagrams of the axial 𝐿 𝑞 𝐿2 𝛼3
+ 𝑃0 (𝛼1 − )
force, shear, net bending moment and the stresses in the extreme 12 6
fibers of the beam cross section along its span. Three {𝐹̅0 } = 𝑃0
comprehensive examples are presented in detail in a companion 𝐿𝑞
paper that show the effectiveness and accuracy the proposed + 𝑃0 (𝛼2 + 𝐿𝛼3 )
2
method and corresponding equations. 𝐿2 𝑞 5𝐿2 𝛼3
It is concluded that the effects of force losses caused by − + 𝑃0 (−𝛼1 − 𝐿𝛼2 − )
12 6
friction and curvature along the prestressed tendons are of great
[ ]
importance in the analysis of prestressed structures and should be
taken into account in their analysis and design. The proposed
stiffness matrix and load vector can be incorporated into
conventional matrix structural analysis software. The equivalent
e 2m
e3

load method for prestressed and post-tensioned beam analysis

proposed by T. Y Lin [5] has been demonstrated to be valid.
e2

Therefore, the effects of post-tensioning can be replaced by a set

of external equivalent loads.
e1m

Appendix I
e1

Stiffness Matrix of a prismatic prestressed beam

𝑘𝑝 0 0 𝑘𝑝 0 0
0 𝑘11 𝑘12 0 −𝑘11 𝑘12 L1 L1 L2 L2
0 𝑘12 𝑘22 0 −𝑘12 𝑘24 2 2 2 2
̅ = 𝑘𝑝
𝐾 0 0 𝑘𝑝 0 0
L1 L2
0 −𝑘11 −𝑘12 0 𝑘11 −𝑘12
0 𝑘12 𝑘24 0 −𝑘12 𝑘22
[ ]
FIG. 4 Variables used in the prestressed cable parabolic profile
12𝐸𝐼 6𝐸𝐼 4𝐸𝐼 2𝐸𝐼
where: 𝑘11 = ; 𝑘12 = ; 𝑘22 = ; 𝑘24 = ; and 𝑘𝑝 =
𝐿3 𝐿2 𝐿 𝐿 Variables used in the parabolic profile of a prestressed cable
𝐸𝐴
𝐿
Input data: 𝑒1 , 𝑒3 , 𝐿1 and 𝐿2 (see Fig. 4)

6
 (3𝑒1 +𝑒3 )𝑙1 +4𝑒1 𝑙2 (𝑒1 +3𝑒3 )𝑙2 +4𝑒3 𝑙1
Results: 𝑒1𝑚 = ; and 𝑒2𝑚 =
4(𝑙1 +𝑙2 ) 4(𝑙1 +𝑙2 )

𝑒2 = eccentricity at the inflection point of the cable profile=

𝑒3 𝑙1 +𝑒1 𝑙2
(𝑙1 +𝑙2 )

Acknowledgements
The writers wish to thank the Structural Stability Group (GES)
of the l Universidad Nacional de Colombia at Medellin for its
support.

List of references

[1] Michael. P. Collins and Denis Mitchell, “Prestressed Concrete

Structures”, Taylor & Francis Group, 784 pages, 1991.

[2] T. Y. Lin and N. H. Burns. “Design Of Prestressed Concrete

Structures”, 3rd Edition, John Willey and Sons, New York, 656
pages, 1981.

[3] E. G. Nawy. “Prestressed Concrete A Fundamental

Approach”, Prentice Hall; 5th edition, 984 pages, 2005.

[4] T.Y. Lin and K. Thornton. "Secondary Moment and Moment

redistribution in a Continuous Prestressed Beam," PCI Journal,
Vol. 17, 1, pp. 8-20, 1972.

[5] T. Y. Lin. "Load-balancing Method for Design and Analysis

of Prestressed Concrete," Journal Proceedings, ACI, International
Concrete Abstracts Portal, Vol. 60, pp. 719–742, 1963.

[6] T. Y. Lin. "A New Concept for Prestressed Concrete"

Australian Constructional Review, Vol. 34, p. 21–31, Sept. 1961.

[7] B. O. Aalami. "Load Balancing: A comprehensive Solution to

Post-tensioning" ACI Struct. J., Vol. 87, 6, p. 662-670, 1990.

[8] B. O. Aalami. "Structural Modeling of Post-tensioned

Members," J. Struct. Engineering, ASCE, Vol. 126, 2, pp. 157-
162, 2000.

[9] A. E. Naaman. “Prestressed Concrete Analysis and Design:

Fundamentals”, 3rd Edition, Techno Press 3000, Michigan, 1176
pages, 2012.