You are on page 1of 74

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329235631

Thermal loading effects in beams

Thesis · June 2014

CITATIONS READS

0 1,478

1 author:

Michele De Filippo
The Hong Kong University of Science and Technology
9 PUBLICATIONS   3 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A Pseudo-lower Bound Method for the assessment of RC Slabs View project

All content following this page was uploaded by Michele De Filippo on 28 November 2018.

The user has requested enhancement of the downloaded file.


Università degli Studi di Napoli
”Federico II”

Facoltà Di Ingegneria
Corso di Laurea in Ingegneria Edile

Tesi di Laurea in Scienza delle Costruzioni


Thermal loading effects in beams

Relatore: Candidato:
Chiar.mo Prof. Ing. Michele De Filippo
Giovanni Romano N41/561

Correlatrice:
Prof. Ing. Marina Diaco

Anno Accademico 2013/2014


To my parents
For their endless love,
support and encouragement

1
Contents

Abstract 1

1 Fundamental principles 3
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Thermal stresses . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Statically determined structures . . . . . . . . . . . . . . . . . 7
1.4 Statically undetermined structures . . . . . . . . . . . . . . . 7
1.5 Thermal expansion . . . . . . . . . . . . . . . . . . . . . . . . 8
1.6 Thermal bowing . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.7 Combinations of thermal expansion and bowing . . . . . . . . 13
1.8 The temporal variable . . . . . . . . . . . . . . . . . . . . . . 15

2 The beam model 17


2.1 Introduction to Saint Venant’s problem . . . . . . . . . . . . . 17
2.2 Saint Venant’s principle . . . . . . . . . . . . . . . . . . . . . 18
2.3 Assumptions on stress state . . . . . . . . . . . . . . . . . . . 19
2.4 Principal directions . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 Equilibrium conditions . . . . . . . . . . . . . . . . . . . . . . 26
2.6 Characteristics of interactions . . . . . . . . . . . . . . . . . . 28
2.7 Constitutive relations . . . . . . . . . . . . . . . . . . . . . . . 30

3 Mechanical behavior 32
3.1 Elastic equilibrium . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2 Simple flexure . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2.1 Straight flexure . . . . . . . . . . . . . . . . . . . . . . 42
3.2.2 Skew flexure . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 Centered normal force . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Eccentric normal force . . . . . . . . . . . . . . . . . . . . . . 45
3.5 Homogenous sections . . . . . . . . . . . . . . . . . . . . . . . 47

2
4 Thermal behavior 50
4.1 Analysis of thermal loading effects . . . . . . . . . . . . . . . . 50
4.2 Navier’s theory including thermal loads . . . . . . . . . . . . . 54
4.3 Measurements of temperature variations . . . . . . . . . . . . 54
4.4 Elastic equilibrium . . . . . . . . . . . . . . . . . . . . . . . . 55

5 An application of the theory 60


5.1 Bimetallic strip subjected to temperature rise . . . . . . . . . 60
5.2 Application of Navier’s theory including thermal loads . . . . 62

Bibliography 67

3
List of Figures

1.1 Uniform heating of a simply supported beam . . . . . . . . . . 8


1.2 Axially restrained beam subjected to uniform temperature rise 8
1.3 Buckling of an axially restrained beam subjected to uniform
temperature rise . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Simply supported beam subjected to a trasversal uniform
temperature gradient . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 Laterally restrained beam subjected to uniform thermal gradient 11
1.6 Fixed end beam subjected to uniform thermal gradient . . . . 12
1.7 Beam with finite rotational restraint subjected to uniform
thermal gradient . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.8 Combined thermal expansion and bowing in a fixed end beam 13

2.1 Beam model and cross-section . . . . . . . . . . . . . . . . . . 19


2.2 Illustration of a plane stress state . . . . . . . . . . . . . . . . 23
2.3 Mohr Circle illustrating a generic stress state . . . . . . . . . 24
2.4 Trend of the shear stress on the boundary . . . . . . . . . . . 27

3.1 Beam model loaded by normal force and bendig moment . . . 33


3.2 Generic cross-section and its relative stress diagram . . . . . . 35
3.3 Bending moment applied on beam model and its relative
cross-section . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.4 Stress and strain diagrams of a generic cross-section composed
by two different elastic Young’s moduli . . . . . . . . . . . . . 40
3.5 Straight flexure on a generic cross-section . . . . . . . . . . . . 42
3.6 Skew flexure on a generic cross-section . . . . . . . . . . . . . 43
3.7 Eccentric normal force on a generic beam model and its
relative cross-section . . . . . . . . . . . . . . . . . . . . . . . 45

4.1 Thermal crack on a concrete bridge . . . . . . . . . . . . . . . 52


4.2 Closely spread thermal cracks on asphalt . . . . . . . . . . . . 52
4.3 Vertical or near vertical cracks in poured concrete . . . . . . . 53
4.4 Diagonal cracks in poured concrete foundations . . . . . . . . 53

4
5.1 Bimetallic strip subjected to temperature increase . . . . . . . 61
5.2 Cross-section of bimetallic strip . . . . . . . . . . . . . . . . . 62
5.3 Bending of a shelf . . . . . . . . . . . . . . . . . . . . . . . . 64
5.4 Dilatation and stress diagrams at the cross-section of the
bimetallic strip . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5
Abstract

A temperature variation applied on a beam tends to make the material


physically expand or contract.
When a structure is subjected to a temperature increase, it results a
reduction in terms of stiffness and strength, which significantly affects the
structural performance.

Consequently it would come to be very useful to describe the behavior of


beams under thermal loads in order to fully understand which stresses and
strains are occurring in it.
This thesis has the aim to illustrate a relatively simple method that
allows to evaluate the effects of arbitrary changes of thermal scenery
interesting the cross-section of a beam.

The theory is valid for small dilatations, for linear and non-linear
stress-strain relationships, and has been assumed the hypothesis of conser-
vation of planar sections that allows the geometrical dilatation diagram to
be linear.

The object of the analysis, in the most general case, is a beam constituted
by longitudinal fibers of different Young’s modulus and coefficient of thermal
expansion, therefore not homogeneous.
The materials constituting the longitudinal fibers of the beam are assumed
elastically and thermally isotropic.
The hypothesis of thermal isotropy allows to simplify the problem from a
computational point of view, so to consider a unique coefficient of thermal
expansion.

1
The problem can be divided into two parts: thermal conduction on the
beam and effects of thermal variation on the beam

Consider to have already solved the first of them, so the temperature


variations are assumed as problem data.

The solution of the elastic equilibrium problem is given by the


introduction of moments of order 0 and 1 of the stresses distribution, rather
fictitious normal force and bending moment induced by the temperature
variation.
These lasts are only computational entities with the dimension of forces
elicited by fictitious thermal stresses equivalent to the thermal expansion.

Evaluated the stresses as difference of the geometrical dilatation and


the elastic dilatations equivalent to the thermal dilatation, both multiplied
by the elastic modulus, the normal force and bending moment induced by
mechanical loads have to be summed to fictitious normal force and bending
moment in order to derive characteristics of interactions on the beam.

Through this kind of procedure the geometrical dilatations, rather total,


on the cross-section of the beam, which vary with a linear law according to
the hypothesis of conservation of planar section, can be calculated.
By having derived the thermal dilatations since proportional to the co-
efficient of thermal expansion and the temperature change, the elastic
dilatations can be calculated with a simple subtraction.

Evaluated the elastic dilatations, which vary with a non-linear law,


such as also the thermal ones, the stresses can be calculated simply by
multiplying the elastic dilatations by the different Young’s modulus of each
longitudinal fiber constituting the whole beam.

Stresses and strains on the cross-section, in the case in which also


mechanical interactions are acting on the beam, can be defined by the
Navier ’s formula including thermal loads.

2
Chapter 1

Fundamental principles

1.1 Introduction
When a beam is subjected to temperature gradient or when a composite
material consisting of two or more materials of different coefficients of
thermal expansion is subjected to a temperature rise, either uniformly or
non-uniformly, the longitudinal fibers tend to expand different amounts in
accord with their individual temperatures and coefficients of thermal expan-
sion.
To enable the body to remain continuous rather than allowing each fiber to
expand individually, a system of thermal stresses and associated stresses may
be introduced depending upon the shape of the beam and the temperature
distribution.
If the material cannot withstand the stresses, rupture will occur.
Brittle and ductile materials react in considerably different manner to
stresses.
Brittle materials can endure only a very small amount of strain before
rupture, while ductile materials can undergo appreciable strains without
rupture.
Thermal stress behavior depends essentially on the ability of the material
to absorb the induced strains necessary to maintain a continuous body
upon the application of thermal gradient, Brittle materials cannot readily
withstand these superimposed strains without inducing enough stresses to
cause rupture.
Ductile materials, on the other hand, can usually withstand these additional
strains, but they may ultimately fail if subjected to a number of cycles of
imposed temperature.

3
The problem of thermal stress is of great importance in current
high-power engines.
The present trend toward increasing temperatures has necessitated the use
of refractory materials capable of withstanding much higher temperatures
than normal engineering materials.
One salient property of these materials is lack of ductility.
For this reason, thermal stress is one of the most important design criteria
in the application of these materials.
Thermal stress is also currently receiving considerable attention in
connection with ductile materials since there is significant evidence that
failure of many ductile engine components can be attributed to thermal
cycling.
The problem of high-speed flight, with attendant increases of temperature
and thermal gradients in aircraft bodies, has further generated concern over
the significance of thermal stress field in ductile materials.

Thermal stress and thermal shock may be distinguished by the fact that
in thermal shock thermal stresses are produced by transient temperature
gradients, usually sudden ones.
For example, if a body originally at one uniform temperature is suddenly
immersed in a medium of different temperature, a condition of thermal shock
is introduced.
At any instant the stresses are determined by the temperature
distribution and are not the same as they would be if the temperature
distribution is obtained in a steady-state condition.
The temperature gradients, that can be established in the transient state,
are generally much higher than those occurring in the steady state.
Hence thermal shocks are of relative concern in comparison with ordinary
thermal stresses because of higher stress field so induced.
Another peculiar aspect of thermal shocks is that the rate of variation
of stress is very high, and many materials are affected by the rate at which
stress increases.
Some materials are embrittled by rapid application of stress and therefore
may not be able to withstand thermal shock stresses while stress variation
of some amount could readily be absorbed.

4
1.2 Thermal stresses
Mechanical stresses are induced in a beam when some or all of its parts are
not free to expand or contract in response to changes in temperature.
In most continuous bodies, thermal expansion or contraction cannot occur
freely in all directions because of geometry, external constraints, existence
of non-uniform temperature gradients, and so stresses are produced.
Such stresses caused by temperature changes are known as thermal stresses.

Problems of thermal stress arise in many practical design problems, such


as those encountered in the design of steam and gas turbines, diesel engines,
jet engines, rocket motors, and nuclear reactors.
The high aerodynamic heating rates associated with high-speed flight present
even more severe thermal-stress problems for the design of spacecraft and
missiles.

Understanding the behavior of structures under thermal effects is


therefore essential, but this topic remains controversial in the field of
structural engineering.
Although advances in analytical modeling have enabled structural designers
to develop sophisticated models, but actual structural behavior under
thermal effects still requires further investigation.

Usually, when the temperature of a body is raised or lowered, the


material expands or contracts according to physical material expansion laws.
If this expansion or contraction is entirely or partially resisted, as already
stated, stresses are set up in the body.

The whole problem is analyzed with the assumption of elastic and


thermal isotropy of all the fibers constituting the beam model.
The hypothesis of thermal isotropy allows to simplify the problem from a
computational point of view, and to consider just one coefficient of thermal
expansion instead of six generated in the case of thermal anisotropy.

The temperature change creates thermal strain in the material.


This dilatation, in an isotropic beam, is given by the coefficient of thermal
expansion, α, times the temperature change of the beam ∆T at each point,
times the length of the beam.

εthermal = α ∆T

5
The most fundamental relationship that governs the behavior of
structures when subjected to thermal variations is the law of addition of
stretchings

εgeometrical = εthermal + εmechanical

The geometrical strains, given by the direct sum of the thermal and
mechanical contributes, govern the deformed shape of the structure through
kinematic or compatibility considerations.
In chapters 3 and 4 the mechanical and thermal contributions will be studied
in depth.

Neglecting the mechanical interactions, and by focusing the attention


principally on thermal effects, or rather on the εthermal previously defined,
it comes to be fundamental to clarify how the structural behavior changes,
depending on different constraints configurations.

The single most important factor that determines a real structure


response to a temperature rise is the manner in which it responds to the
unavoidable thermal strains induced in its members.
These strains induce in the beam an uniform dilatation (under an average
temperature rise) and a curvature (due to a temperature gradient through
the section).

ˆ If the structure has an insufficient translational restraint to thermal


expansion, the considerable strains are taken up in expansive displacements,
producing a displacement dominated response.
Thermal gradients induce curvature leading to bowing of a member whose
ends are free to rotate, again producing large displacements (deflections).

ˆ Straight members whose ends are restrained against translation


produce mechanical strains opposing to thermal expansion, and therefore
large compressive stresses.
Curvature strains induced by the thermal gradient in members whose ends
are rotationally restrained can lead to large bending moments throughout
the length of the member without deflection.
The effect of induced curvature in members whose ends are rotationally
unrestrained, but translationally restrained, is to produce tension.

6
These two behavior cases correspond to two precise constraints
configurations:

1.3 Statically determined structures


In statically determined structures no equilibrium configuration exists with
respect to thermal loads applied.
The frame does not have enough constraint reactions in order to create a
self-balanced force system due to thermal changes on it, so just strains arise,
with no stresses.

When thermal strains are free to develop in an unrestricted manner, the


structure suffers elongations or contractions as consequence of the physical
effect due to a thermal load applied on it.
So, ignoring other kinds of strains, the geometrical strain is

εgeometrical = εthermal 6= 0

1.4 Statically undetermined structures


In statically undetermined structures, one and only one equilibrated and
congruent equilibrium configuration exists with reference to thermal loads
applied.

In this case as effect of the thermal load applied, forces explicated by the
constraints (reactions) may occur, so the strain state does not vanish, but
the frame is not free to expand, and a self-equilibrated stress state will arise.

So in this case, in order to satisfy equilibrium and congruence conditions,


will be

εthermal + εmechanical = 0

7
1.5 Thermal expansion
If in each cross-section a uniform temperature rise, ∆T is applied to a
simply supported beam without axial restraint, the result will simply be an
increase in terms of length equals to α∆T l as shown in Fig. 1.1.
Therefore the total (geometrical) strain, εg , is equal to the thermal strain
(indicating εt = εthermal ) and there is no mechanical strain (indicating
εm = εmechanical ) which means that no stresses develop in the beam.

Figure 1.1: Uniform heating of a simply supported beam

Usually beams do not have the freedom to elongate in the manner


described above.
A more common case is to consider an axially restrained beam subjected to
a uniform temperature rise, ∆T (as shown in Fig. 1.2).
It is clear to see that in this case the geometrical strain (indicating
εg = εgeometrical ) is zero (no displacements).

Figure 1.2: Axially restrained beam subjected to uniform temperature rise

This is because the thermal expansion is canceled out by equal and


opposite elastic contractions caused by the restraining force P (i.e.
εg = εm + εt therefore εt = εm ).
Now it exists a uniform axial stress σ in the beam equal to E εm .
The magnitude of the restraining force P is,

P = E A εm = − E A εT = −E A α ∆T

8
If the temperature is allowed to rise increasingly, then there will be two
basic responses, depending upon the slenderness of the beam.

ˆ If the beam is sufficiently stocky, then the axial stress will sooner
or later reach the yield stress σy of the material, and if the material has
an elastic–plastic stress–strain relationship, the beam will continue to yield
without any further increase in stress, but it will also store an increasing
magnitude of plastic strains.

ˆ If the beam is slender it means that the Euler buckling load is


relatively low, so easily reachable, so it will buckle before the material
reaches its yield stress.
In this case, if the temperature is allowed to rise further, then the total
restraining force will stay constant (assuming an elastic material and no
thermal degradation of properties) and the thermal expansion strains will
continue to be accommodated by the outward deflection of the beam δ as
shown in Fig. 1.3.

Figure 1.3: Buckling of an axially restrained beam subjected to uniform


temperature rise

The above cases represent the two fundamental responses in beams


subjected to restrained thermal expansion.
Either of the two (yielding or buckling) may occur on its own (based upon
the slenderness of the beam).
In general more complex response consisting of a combination of yielding
and buckling may also occur.

9
1.6 Thermal bowing
In the previous sections the effects of a uniform temperature rise on axially
restrained beams have been discussed.
In real thermal irradiations the temperature distributions are however not
uniform.
In a moderate size compartment of regular shape one may assume that the
compartment temperature is roughly uniform at a given time.
The temperature of the structural members in the compartment depends on
the material they are made of and on other details of geometry, construction
and design.
Imagine that the outer surfaces of it on the roof are exposed to irradiations,
so they are at a much higher temperature than the ones on the inside of the
compartment.
This causes the outer surfaces to expand much more than the inner surfaces
inducing bending in the member, this effect is called thermal bowing.

Fig. 1.4 shows a beam subjected to a uniform temperature gradient


through its depth (d) along its whole length (l).

Figure 1.4: Simply supported beam subjected to a trasversal uniform


temperature gradient

10
Assuming that the beam is simply supported (as shown in Fig. 1.4), it
is possible to derive the following relationships

The gradient of thermal increase (T,y ) over the depth is

T2 − T1
T,y =
d
A uniform curvature (θ) is induced along the length as a result of the
thermal gradient T,y ,

θ = α T,y

Due to the curvature of the beam, this latter tends do deflect.


If this deflection is interpreted as a simultaneous contraction and expansion
strain εθ , then the value of it can be calculated by analyzing Fig. 1.4 as

 

1 − sin 2
εθ = lθ
2

Now consider the laterally restrained beam of Fig. 1.3.


If a uniform thermal gradient T,y is applied to this beam (as shown in
Fig. 1.5), then the result is a thermally induced tension in the beam
and corresponding reactions at the support (opposite to the pure thermal
expansion case discussed earlier).

Figure 1.5: Laterally restrained beam subjected to uniform thermal gradient

This effect is caused by the translational restraint against the contraction


strain (εθ ) induced by the thermal gradient.

11
Figure 1.6: Fixed end beam subjected to uniform thermal gradient

Figure 1.7: Beam with finite rotational restraint subjected to uniform


thermal gradient

Fig. 1.6 shows a fixed ended beam (by adding rotational end restraints to
the beam of Fig. 1.5) subjected to a uniform temperature gradient through
its depth.
Recall that a uniform curvature θ = αT,y exists in a simply supported beam
subjected to gradient T,y : If that beam is rotationally restrained by support
moments M (uniform along length) an equal and opposite curvature induced
by the support moments cancels out the thermal curvature and, therefore,
the fixed ended beam remains ”straight” with a constant moment M = EI θ
along its length.
From the above discussion it is clear that the effect of boundary restraints is
crucial in determining the response of structural members to thermal actions.

The key conclusion to be drawn from the discussion so far is that,


thermal strains will be manifested as displacements if they are unrestrained
or as stresses if they are restrained through counteracting elastic or plastic
strains generated by restraining forces.

Perfect rotational restraint (such as that one shown in Fig. 1.6) is


also not very easily achieved in real structures (other than for symmetric
loading on members over continuous supports, without hinges from strength
degradation).
Fig. 1.7 shows a beam rotationally restrained at the ends by rotational
elastic springs of stiffness kr :

12
In this case, the restraining moment in the springs as a result of a
uniform thermal gradient T,y can be found to be

EI α T,y
Mk =
1 + 2kEI
rl

1.7 Combinations of thermal expansion and


bowing
In the previous sections, the response of beams to either the thermal
expansion or thermal bowing has been considered in isolation.
To study the combined response, let first consider the case of a fixed ended
beam as shown in Fig. 1.8, which is both rotationally and translationally
restrained at both ends.

Figure 1.8: Combined thermal expansion and bowing in a fixed end beam

If the beam is subjected to a mean temperature rise and to a thermal


gradient through the depth, then it will experience a uniform compressive
stress because of restrained expansion and a uniform moment due to the
thermal gradient.
The typical stresses on a cross-section because of the combined effect of the
two thermal actions are shown in Fig. 1.8.
The bottom of the beam will experience notable compressive stresses, while
the top may be anywhere between significant compression to significant
tension.

13
The fundamental pattern of behavior of a beam whose ends are laterally
restrained (but rotationally unrestrained, see Fig. 1.5), subjected to thermal
expansion and thermal bowing separately was established in the previous
sections.
Restrained expansion resulted in compression and bowing resulted in
tension.
This helped to illustrate that two opposite stress regimes can occur
depending upon the thermal regime applied, however, the apparent response
of the beam is the same (i.e. downward deflection).

The main parameters that determine the response are an average


temperature equivalent rise (∆T ) and an average equivalent thermal
gradient (T,y ).

To study the effects of the applying combinations of thermal expansion


and thermal bowing, it is useful to define the effective strain as follows

εef f = ε1 ± ε2

Where ε1 is the strain field induced by the mean temperature rise, and
ε2 that one induced by the thermal gradient.
The variation of εef f (for various thermal regimes) can produce a large
variety of responses.
Positive values of εef f imply compression (if the effect of mean temperature
rise is dominant) and negative values imply tension (if the effect of thermal
gradients is dominant).

14
1.8 The temporal variable
Another very relevant factor, wrongly not considerated till now in the
applications viewed, is the temporal variable.

Normally, in structural design, as seen till this moment, an element


interested by a temperature change is simply modeled as a constraint defined
beam model under a specific thermal load, but in reality the problem comes
to be much more complex than how it looks.

A temperature gradient is not a simple increase or decrease of


temperature applied to the structural element, it is a phenomenon that
expresses the temperature difference between a surface and another one.
It is due to time, therefore it has not to be studied only the most critical
situation of it, but instead the whole temperature variation process during
a specific time.

A variety of responses can indeed exist in structures if one considers the


different types of thermal loads a structure may be subjected to.
A fast burning irradiation that reaches flashover and high temperatures
quickly and then dies off can produce high thermal gradients (i.e. in re-
reinforced concrete frame: hot steel and relatively cold concrete) but lower
mean temperatures.
By contrast, a slow irradiation that reaches only modest temperatures but
induces a temperature rise for a long time could produce considerably higher
mean temperature and lower thermal gradients.

Two important consideration cannot be avoided :

ˆ Firstly, it is absolutely not always true that the thermal behavior of


a common ideal beam model is linear and elastic, contrariwise by definition
thermal loads could even produce plastic effects that vary with a non-linear
law and produce non-completely reversible deformations.

15
ˆ Secondly, by introducing the temporal variable, the whole procedure
would have to be integrated between an initial time t0 and a final one t1 .
This would be the most correct and rigorous method of analyzing the
temperature variations and the consequent changes in terms of stresses and
strains in the structure.
This procedure would be computationally too expensive, therefore comes to
be useful to refer to some single temporal moments, considered to be the
most crucial.

16
Chapter 2

The beam model

2.1 Introduction to Saint Venant’s problem


The Saint Venant beam model is made of a homogenous, isotropic and
iso-resistant material with an elastic-linear behavior, having the shape of
a right prism of length l, loaded on the terminal bases by two systems of
forces that overall constitute a system statically equivalent to zero.
The side mantle of the beam is unloaded and inertial forces are assumed
equal to zero.
The elastic equilibrium problem is well known as Saint Venant’s problem,
from the name of the man who first gave a rigorous solution of it in 1855,
making a major contribution to the theory of elasticity and to Construction
Science.
In addition to historical, the importance of the problem is related to the
generalization of the solution made by the conjecture that bears his name,
which may be a representation of a fairly broad class of problems of the
theory of beams.

17
2.2 Saint Venant’s principle
The solution provided by Saint Venant relates to particular fields of stresses
acting on the basis of the beam.
It would seem, therefore, of limited interest, and by the way not useful for
applications.
The possibility of giving validity to such general solution, although in an
approximate format, but still quite satisfactory for applications, is all due
to the genius of Saint Venant.

He formulated the following principle:

Systems of forces statically equivalent applied on the bases of the


beam, produce the same effects on the whole length of the elem,
with exception of areas of limited extension in vicinity of the bases

More well known as Principle of Saint Venant.

This principle of indifference allows to evaluate the stress state of a beam


having as data the only static characteristics of systems of forces acting on
the bases.
The precise mathematical formulation and demonstration of this principle
have been one of the most fascinating problems of Mechanics and Structural
Engineering.

18
2.3 Assumptions on stress state
In the following will be denoted with σ̃ the tensor of stress state, and with ε̃
the strain state one.

Prior to the introduction of the technical theory of Saint Venant’s beam,


it is appropriate to make considerations with respect to the simplified beam
model proposed.

The assumptions made by the French savant are useful to simplify


the entire discussion that otherwise would occur too complex, but that
in anycase reflects a structural modeling adhering to the real mechanical
behavior.

The solution to the problem of the beam provided by Clebsch and Saint
Venant is based on the idea of considering the beam as a bundle of
longitudinal fibers that transmit only tangential interactions, and therefore
assumes the absence of any transversal interaction.
This is equivalent to assume zero normal stresses on lying postures parallel
to the beam axis, which is the geometric locus of the centers of the straight
sections.
To show the implications of this hypothesis on the stress state, consider an
orthonormal reference {O, ī, j̄, k̄} with the unit vector k̄ parallel to the axis
of the beam.
The intersections between the prism and the planes perpendicular to the
prism generators are called cross-sections of the beam.

Figure 2.1: Beam model and cross-section

Let x̄ ∈ V be a generic position vector with components x, y, z so that

x̄ = xī + y j̄ + z k̄

19
Let n̄ be the unit vector normal to the lying posture parallel to k̄, then
clearly, for the orthogonality of n̄ and k̄ their vectorial product is equal to
zero.

n̄ · k̄ = 0

Applying the tensional matrix staggered along n̄ are obtained its generic
components, or rather the direction cosines nx e ny .

The hypothesis of purely tangential interaction between the longitudinal


fibers of the beam requires that the vectorial product of σ̃n̄ and n̄ is zero.

σ̃n̄ · n̄ = 0 ∀n̄ ∈ V : n̄ · k̄ = 0

That could be written in terms of components as

σn = σx n2x + σy ny 2 + 2 τxy nx ny ∀ nx , n y

The resolution of the previous equation shows that this condition is


satisfied only if results

σx = σy = τxy Components of the stress state equal to zero

The matrix is representative of the stress state in each point of the beam,
thus takes the form as following

 
0 0 τxy
σ̃ =  0 0 τyz 


τzx τzy σz

To simplify the notation the normal stress on the cross-section will be


denoted with σ = σz ,and the shear stress vector with τ̄ of components τx
and τy .

Therefore

τ̄ = τx ī + τy j̄ + 0 k̄ = τx ī + τy j̄

20
With τ instead will be indicated the magnitude of the shear stress

q
τ = |τ | = τx2 + τy2

The matrix of the whole stress state in the orthonormal reference


{O, ī, j̄, k̄} will be written therefore as

 
0 0 τxy
σ̃ = 0 0 τyz 


τzx τzy σ

and in tensorial notation

σ̃ = σ k̄ ⊗ k̄ + 2 sym (k̄ ⊗ τ̄ )

At the current point of the discussion the stress matrix σ̃ is already


considerably reduced in comparison with the initial conditions, in which
would have been considered with each of its components.
The only assumption of purely tangential interactions greatly reduces the
difficulty of the problem, since applying the stress matrix along any position
vector the transversal components of tensions will always be zero.

It remains anyway to define the stress state, which is easily derived from an
elementary assessment concerning the invariants of the tensor σ̃.

If algebraically were applying the determinant of the strain matrix ε̃


along any position vector n̄, representative of a longitudinal fiber, this latter
generally suffers elongations or contractions along the principal directions of
strain and some angular slidings.

Specifically in this case the invariants of the tensor σ̃ are the unknown
factors of the equation in terms of the eigenvalues, more commonly known
as the secular equation

λ3 − I1 λ2 − I2 λ − I3 = 0

21
The equation is of 3rd degree, and due to simmetry, it admits three
solutions that are the eigenvalues associated with the eigenvectors, dependent
on the I, scalar invariants, representative of the true identity of the tensor.
The stress state is defined by the invariants, solutions of the secular equation.
The invariants of the tensor σ̃ are

Linear Invariant I1 = trσ̃ = σ


Quadratic Invariant I2 = 21 [(trσ̃)2 − tr(σ̃)2 ] = −τ 2
Cubic Invariant I3 = detσ̃ = 0

Recalling that the determinant is equal to the product of the eigenvalues


associated with a principal orthonormal triad, the vanishing of the
determinant ensures that at least one of these eigenvalues is equal to zero.

Therefore is possible to adfirm that

The stress state of Saint Venant is a plane state of stresses.


This latter π has as orthogonal the principal direction with null eigenvalue.
The expression of the stress acting on the generic lying posture with
orthogonal n is

σ̃n̄ = [σ̃ k̄ ⊗ k̄ + 2sym(k̄ ⊗ τ̄ )]n̄ = σ̃(k̄ · n̄)k̄ + (k̄ · n̄)τ̄ + (τ̄ · n̄)k̄

and shows that it is a linear combination of τ̄ and k̄.

So the plan π is at every point identified by the unit vector k̄ parallel to


the beam axis and the shear stress τ̄ acting on the cross-section.
If at one point the shear stress is equal to zero, there is an uniaxial stress
state on the k̄ axis.
For the hypothesis of linearity of the material behavior a linear relationship
between the strain tensor ε̃ and the stress one σ̃ will exist, and they are
linked by a proportionality constant E, modulus of longitudinal elasticity,
more commonly known as Young’s modulus.

22
Figure 2.2: Illustration of a plane stress state

Here is a summary of what has been shown so far

Assumptions on stress state


Geometrics:
The solid has the shape of a right prism
On the material
The material is homogeneous, isotropic
and iso-resistant and exhibits an elastic behavior
On stresses:
σx = σy = τxy = 0
The stress state is plane and identified in each point of the beam
axis by the unit vector k̄ and the shear stress vector τ̄
On loads:
p̄ = σ̃n̄ = 0 Side mantle unloaded
p̄ = σ̃ k̄ On the bases
b̄ = −divσ̃ = 0 Mass forces equal to zero

23
2.4 Principal directions
The principal directions are no more than those preferred directions
according to which the respective vectors parallel to these directions suffer
the maximum strain, dilatation or contraction, without angular slidings.
The principal stresses are trivially those related to the principal directions,
and are of particular interest because they are the most critical in terms
of strain and can be easily calculated since, being the stress state plane, it
means that at least one of them is surely null and therefore the determination
of the other requires to find the zeros of a polynomial of 2nd degree.

Assuming that the y-axis is parallel to the τ in the point in question, the
secular equation is

 
−σp 0 0
det(σ − σp ) = det  0 −σp τ  = −σp [−σp (σ − σp ) − τ 2 ]
 

0 τ σ − σp

and provides the principal stresses

s
σ σ2
σp1 = 0 σp2 , σp3 = + ± τ2
2 4

Figure 2.3: Mohr Circle illustrating a generic stress state

24
It is possible to get to the same result by plotting the Mohr circle relative
to the bundle of lying postures orthogonal to the stress plane (Ref. Fig. 2.3).

ˆ The stress on the lying posture othogonal to k̄ has components


according to Mohr equal to {σ, τ }

ˆ The stress on the orthogonal lying posture, whose normal is parallel


to τ̄ has components according to Mohr equal to {0, −τ }.
Then the center of the circle in the Mohr plane has abscissa equal to σ2 , the
radius and the intersections with the axis of σ are respectively equal to

s s
σ2 σ σ2
r= + τ2 σmax , σmin = ± + τ2
4 2 4

25
2.5 Equilibrium conditions
The discussion on the equilibrium of Saint Venant is nothing more than
a specialization of the equilibrium equations of Cauchy in relation to the
assumptions made on the beam model, therefore, the significant reduction
of the matrix σ̃ previously seen.
Since the hypothesis of null mass forces, the differential condition of
equilibrium b̄ = −divσ̃ = 0 dictates that

dτx dτy dτx dτy dσ


=0 =0 + + =0
dz dz dz dz dz
therefore

τ¯0 = 0 div τ̄ = −σ 0

In order to simplify the dissertation have been defined σ 0 = and
dz
dτ̄
τ¯0 =
dz
ˆ The condition τ¯0 = 0, is obtained very trivially from the first two
equations, and results in the property: The field of shear stresses is repeated
identically on each cross-section of the beam.

ˆ The condition div τ̄ = − σ, that comes from a more complex


treatment, allows to evaluate the divergence of the field of τ̄ as a function
of the gradient of the normal stress σ.

Differentiating the expression for div τ̄ = − σ with respect to z can be


deducted

σ 00 = (div τ̄ )0 = σ 00 + div τ¯0 = 0

Therefore, since τ¯0 = 0, it must be σ 00 = 0.


So it follows that the normal stresses vary with affine law along the axis of
the beam.

The conditions of equilibrium on the side surface of the beam p̄ = σ̃n̄ = 0,


which by hypothesisis is unloaded, then impose that

26
τ̄ · n̄ = 0 ⇐⇒ τx nx + τy ny = 0

In the points of the contour of the cross-sections the shear stresses τ are
therefore tangent to the contour line, as shown in Fig 2.4.

Figure 2.4: Trend of the shear stress on the boundary

The condition of boundary tangency may also be expressed by saying that

ˆ The cross-section is a flow tube for the shear stresses.

¯ are written
The equilibrium conditions on the basis of normal k̄ and −k
respectively

p̄ = σ̃ k̄ p̄ = −σ̃ k̄

The vector τ̄ on the bases is composed of two components, one in the x


direction and the other in the y one.
In vicinity of the mantle there are not tangential loads, therefore on the
bases by getting closer to the edge of the same, the τ tend to increase or
decrease progressively in such a way as to act in a tangential way on the
mantle.

27
2.6 Characteristics of interactions
Occurs to define, at this point of the dissertation, which are the possible
characteristics of interactions that may arise within the solid due to an
applied load, with obvious reference to what has been said so far, and for all
the assumptions made.

Let O be the origin of a reference and r̄ its relative radius vector which
identifies the points in the plane of a cross-section of the beam of area A.
The resultant of the tensions, acting on a cross-section with the outgoing
normal vector k̄, is the vector

Z
R= (σ̃(r̄) k̄) dA = N k̄ + T
A

ˆ The scalar component of N along the axis of the beam is the normal
force

Z Z
N= (σ̃(r) k̄ · k̄) dA = σ̃(r̄) dA
A A

If N is positive, the normal force will be assumed of traction, otherwise


compression.

ˆ The T component of the resulting vector on the plane of the


cross-section is the shear force
Z Z
T = [σ̃(r̄) k̄ − σ(r̄) k̄] dA = τ̄ (r̄) dA
A A

and in components

Z Z
Tx = τx (r̄) dA Ty = τy (r̄) dA
A A

The axial vector of the resultant moment with respect to a pole O is worth

Z
M0 = r̄ × σ̃ k̄ dA = Mt k̄ + Mf
A

28
where the symbol × denotes the vectorial product

ˆ The scalar component Mt along the axis of the beam is the twisting
moment

Z Z
Mt = (r̄ × τ (r̄) · k̄) dA = (k̄ × r̄) · τ (r̄) dA
A A

Or rather

Z
Mt = [τy (r̄) x − τx (r̄) y] dA
A

ˆ The vector applied in the plane of the cross-section is the bending


moment

Z
Mf = r̄ × k̄ σ dA
A

and in terms of components

Z Z
Mx = σ(r̄) y dA My = σ(r̄) x dA
A A

29
2.7 Constitutive relations
Consider the isotropic linear elastic relationship σ − ε, for the hypothesis
of linearity, as already said, therefore will exist a factor of proportionality
between stress and strain defined as the longitudinal modulus of elasticity,
and for the isotropy hypothesis, this factor will be independent from the
position of the object considered.
But it is although important to consider that in addition to the deformation
concordant with the applied efforts, the material will receive a further
transversal strain, for the hypothesis of isotropy to a longitudinal expansion
along the direction 1 correspond two contractions in the directions 2 and 3
of equal intensity ν, defined as Poisson’s ratio.
So what has been said means that to a state of uni-axial stresses corresponds
a state of tri-axial strain.
It is now possible do define the Navier’s elastic relation.

1 ν 1+ν ν
ε1 = v σ̃1 − tr(σ̃) I = σ̃1 − tr(σ̃) I
2G E E E
substituting the expression of the stress state of Saint Venant

σ̃ = σ k̄ ⊗ k̄ + 2 sym (k̄ ⊗ τ̄ )

and noting that trσ̃ = σ, it is possible to deduce the relation

1+ν ν 1 1
ε= [σ k̄ ⊗ k̄ + 2 sym (k̄ ⊗ τ̄ )] − σ I = σ (k̄ ⊗ k̄ − ν Π) + sym (k̄ ⊗ τ̄ )
E E E G
where Π̄ = I¯ − k̄ ⊗ k̄ is the projector orthogonal to the plane of the
cross-section.
The strain tensor ε̃ is then the sum of an axial component and a transversal
one

1
εa = σ k̄ ⊗ k̄ = εa k̄ ⊗ k̄
E
1 ν
εt = sym(k̄ ⊗ τ̄ ) − σ Π̄ = sym (k̄ ⊗ γ) + εt Π̄
G E

30
where the following relations have been imposed.

1 1
εa = σ εt = −ν εa ν= τ
E G
Decomposing now the displacement field in the sum of the axial
component w̄ and the transversal one v̄

ū = w k̄ + v̄ with v̄ · k̄ = 0

if the strain field is congruent with the displacement field ū = w k̄ + v̄,


the consequence is

σ = E εa = E w 0 εt = −ν εa = −ν w0

and being σ 00 = 0, it is immediate that

ε00a = w000 = 0
ε00t = −ν w000 = 0 =⇒ ε00 = 0
γ 0 = G1 τ 0 = 0

it may therefore be said that

The elastic strain field ε̃ of the longitudinal fibers of Saint Venant


if congruent varies with an affine law along the axis of the beam.

31
Chapter 3

Mechanical behavior

Starting from the assumptions made so far, as following will be threated the
mechanical behavior of the beam model introduced in the previous chapter.
It comes to be fundamental to introduce the elastic theory of Navier,
formulated by the French scientist and engineer at the beginning of the 19th
century.
In structural mechanics, the theory of beams, or technical theory of beams, is
a simplification of the theory of elasticity in linear field for the analysis of
beams’ mechanical behavior.
In particular, it is a simplification of Saint Venant’s problem.
It was formulated around 1638 and developed in the 17th and 18th centuries.
After the success demonstrated in the 19th century with the construction of
metal bridges in France and England, the Eiffel Tower and Ferris wheels, the
theory of beams had a big success and was considered one of the cornerstones
of engineering.

Consider the case in which the stresses on beams’ cross-section are


equivalent to a normal force N and a bending moment Mf .
The solution to the problem of elastic equilibrium is suggested by the
following assumptions on stress and strain state on the beam.

32
ˆ The stresses on the cross-sections are purely normal, it is assumed
that the stress state tensor is uniaxial with axis parallel to the geometric
axis k of the beam

σ̃ = σ k̄ ⊗ k̄

ˆ The cross-sections remain plane when deformations occurred

Figure 3.1: Beam model loaded by normal force and bendig moment

This last conjecture is called principle of conservation of planar sections.


This assumption is equivalent to impose that the dilatation diagram ε of the
longitudinal fibers is linear, in each cross-section.
The strain state is tri-axial with a principal dilatation associated to the
beam axis direction and a double principal dilatation associated with the
transversal lying posture.
The assumptions allow to determine uniquely the stress and strain state in
a beam subjected to normal force and bending moment, even in absence of
homogeneity of elastic Young’s moduli of the longitudinal fibers.
In the following will be therefore considered the general case in which the
individual fibers of the beam are made of different elastic materials.
This case has a great importance in applications.

ˆ To each fiber competes therefore a different value of Young’s modulus.


ˆ The Poisson ratio is assumed instead equal for all fibers.

33
3.1 Elastic equilibrium
As consequence of the absence of shear stresses, the differential condition of
equilibrium in the direction of k̄ requires that the diagram of normal stresses
σ is repeated identically in each cross-section

τ̄ = 0, div τ̄ = − σ =⇒ σ = 0

The linearity of the dilatation diagram ε is expressed by writing

ε(r̄) = ḡ · r̄ + ε0

or in terms of components

ε(x, y) = gx x + gy y + ε0

being gx and gy the components of the gradient ḡ of the dilatation

The constant ε0 is the dilatation of the fiber passing through the


origin of reference, and therefore also of the fibers passing through the
points of the axis passing through the origin and orthogonal to the gradient ḡ.

The magnitude of the gradient ḡ measures the slope of the dilatation


diagram.

ˆ The plane containing the axis of the deformed beam is the flexure
plane and the track on the cross-section is the flexure axis f , parallel to the
gradient ḡ.

ˆ The plane, parallel to the axis of the beam, on which the linear strains
ε are equal to zero, and therefore even the normal stresses σ, is the neutral
plane and the track on the cross-section is the neutral axis n of the flexure.

The gradient ḡ of dilatations is orthogonal to the neutral axis n, directed


to the part of rising strains, and has magnitude

εmax − ε0
|g| =
dmax

34
Figure 3.2: Generic cross-section and its relative stress diagram

The straight line n∗ is the parallel to the neutral axis passing through the
origin O, and dmax is the distance from n∗ to the points of the cross-section
in which there is the maximum strain.
The normal stresses on the cross-sections are provided by the elastic
relationship:

σ(r̄) = E(r̄) ε(r̄) = E(r̄) (ḡ · r̄ + ε0 ) = E(x, y) (gx x + gy y + ε0 )

The values of ḡ and ε0 are calculated by imposing the equilibrium


conditions to the translation along the axis of the beam, and the rotation
with respect to the pole O.

Equilibrium conditions :

Z
N= σ(r̄) dA
Z A Z Z
M̄f = r̄ × σ(r̄) k̄ dA ⇐⇒ k̄ × M̄f = σ(r̄) r̄ dA
A A A

The equilibrium conditions require that the resultant and resultant


moment with respect to O of the distribution of normal stresses σ on any
cross-section are respectively equal to N and Mf .

35
Expressing the normal stresses σ in terms of strains, we have that
Z Z Z
N= σ(r̄) dA = ( E(r̄) r̄ dA) · ḡ + ( E(r̄) dA) · ε0
Z A AZ Z
k̄ × M̄f = r̄ × σ(r̄) dA = ( E(r̄) r̄ ⊗ r̄ dA) · ḡ + ( E(r̄) r̄ dA) · ε0
A A

The moments of order 0, 1, 2 of the distribution of elastic moduli are


given by
Z
A(E) = E(r̄) dA Elastic area (Moment of order 0)
ZA
S̄0 (E) = E(r̄) r̄ dA Elastic static moment (Moment of order 1)
ZA
J˜0 (E) = E(r̄) r̄ ⊗ r̄ dA Elastic inertial moment (Moment of order 2)
A

whose components are


Z Z
Sx (E) = E x dA Sy (E) = E y dA
A A
Z Z
Jx (E) = E x2 dA Jy (E) = E y 2 dA
ZA A
Jxy (E) = E xy dA
A

Therefore the equilibrium conditions are

N = S̄0 ḡ + A(E) ε0
k̄ × M̄f = J˜0 (E) · ḡ + S̄0 (E) ε0
Note that

ˆ The vectorial product of k̄ and another vector orthogonal to it, so,


parallel to the plane of the section, provides the vector k̄ × ā perpendicular to
the beam axis and rotated with respect to ā of π2 counterclockwise according
to an observer arranged along k̄.

Such a consideration may then be applied to the bending moment vector


M̄f and the position vector r̄ in the cross-section plane.

π
Denote by R̃ the tensor making the counterclockwise rotation of 2
in the
plane of the cross-section.

36
The matrix associated to it with respect to the basis {i, j} is

" #
0 −1
R̃ =
1 0

Considering k̄ × M̄f and k̄ × r̄ as bi-dimensional vectors which belongs


to the plane of the cross-section, it is possible to write

k̄ × r̄ = R̃ r̄ k̄ × M̄f = R̃ M̄f

In terms of components is therefore

" # " #
−y −My
k̄ × r̄ = R̃ r̄ = k̄ × M̄f = R̃ M̄f =
x Mx

Figure 3.3: Bending moment applied on beam model and its relative
cross-section
So the equilibrium conditions could be written in terms of components
as

N = gx Sx (E) + gy Sy (E) + ε0 A(E)


−My = gx Jx (E) + gy Jy (E) + ε0 Sx (E)
Mx = gx Jxy (E) + gy Jy (E) + ε0 Sy (E)

37
or in terms of matrix

    
A(E) Sx (E) Sy (E) ε0 N
 Sx (E) Jx (E) Jxy (E)   gx  =  −My 
    

Sy (E) Jxy (E) Jy (E) gy Mx


The elastic center or center of gravity G of the cross-section is by
definition the point that has the position of the weighted average of the
longitudinal fibers positions weighted on the elastic moduli.

The elastic center of gravity is therefore identified by the vector

S̄0 (E)
r̄G =
A(E)
of components

Sx (E) Sy (E)
xG = yG =
A(E) A(E)
The solution of the linear system leads more conveniently to assume the
origin coincident with the elastic center, because in this case the elastic
static moment vector will be equal to zero S̄0 (E).

With respect to the new origin G the bending moment becomes

M̄Gf = M̄f − r̄G × (N k̄)


Indicating with J˜G the elastic inertial moment tensor with respect to
the elastic center, and with JGx , JGy , JGxy its components, the whole system
decouples itself reducing to

N = A(E) ε0 ⇐⇒ N = A (E) ε0
k̄ × M̄Gf = J˜G (E) ḡ ⇐⇒ -MGy = gx JGx (E) + gy JGxy (E)
MGx = gx JGxy (E) + gy JGy (E)
or in terms of matrix

    
A(E) 0 0 ε0 N

 0 Jx (E) Jxy (E)   gx  =  −My 
  

0 Jxy (E) Jy (E) gy Mx

38
The parameters ε0 and ḡ are obtained, therefore, by the relations

N
ε0 = ḡ = J˜G−1 (E) (k̄ × M̄Gf )
A(E)

Note that the elastic area A(E) and elastic inertial moment J˜0 (E) are both
positive.
The elastic inertial tensor, being defined positive is invertible.

ˆ The inverse tensor J˜G−1 (E) is called flessional deformability tensor of the
cross-section.

Strains and normal stresses are given by the following formulas

N
ε(r̄) = + [J˜G−1 (k̄ × M̄Gf )] · r̄
A(E)
N
σ(k̄) = E(k̄) + E(k̄) [J˜G−1 (k̄ × M̄Gf )] · r̄
A(E)

If x and y axes are the principal axes for the elastic inertial tensor of the
section, the inertial product JGxy (E) vanishes.

So the equilibrium equations are

−MGy = gx JGx (E)


MGx = gy JGy (E)
N = A(E) ε0

and therefore fully decoupled as follows

    
JGx (E) 0 0 gx −MGy

 0 JGy (E) 0   gy  =  Mx 
  

0 0 A(E) ε0 N

39
So in the case of non-homogeneous sections strains and stress of the
longitudinal fibers have then the simplified expression, well known as
Navier ’s formulas

N MGy MGx
ε(x, y) = − x+ y
A(E) JGx (E) JGy (E)

!
N MGy MGx
σ(x, y) = E − x+ y
A(E) JGx (E) JGy (E)

Fig. 3.4 shows the typical diagrams of longitudinal strains and normal
stresses on a beam in bending where longitudinal fibers have different elastic
moduli.

Figure 3.4: Stress and strain diagrams of a generic cross-section composed


by two different elastic Young’s moduli

40
3.2 Simple flexure
If the normal force N is null, the cross-section is loaded by a simple flexure.

The neutral axis passes through the elastic center of gravity G of the
section.
By placing the origin in G and choosing the principal inertial axes x and y,
the dilatation of the longitudinal fibers has the expression

MGy MGx
ε(x, y) = − x+ y
JGx (E) JGy (E)

and the normal stresses are

!
MGy MGx
σ(x, y) = E − x+ y
JGx (E) JGy (E)

41
3.2.1 Straight flexure
Straight flexure occurs if the bending moment has the direction of one of the
principal inertial axes of the cross-section.

In this case, in fact, assuming the y-axis as axis of bending, is

MGx MGx
ε(y) = y σ(y) = E y
JGy (E) JGy (E)

Figure 3.5: Straight flexure on a generic cross-section

The neutral axis thus coincides with the x -axis and is orthogonal to the
interaction axis.

42
3.2.2 Skew flexure
In the case the bending moment does not have the direction of one of the
principal inertial axes, the flexure axis f and the interaction axis s do not
coincide.

The interaction is then said skew flexure to emphasize the fact that the
axis of the beam is deflected in a plane inclined with respect to that of
interaction.

Figure 3.6: Skew flexure on a generic cross-section

43
3.3 Centered normal force
The interaction is called normal centered force if the normal force N is
different from zero and the resultant moment with respect to the center of
gravity is equal to zero.
The force is tensile or compressive depending on whether N is positive or
negative.

The dilation of the longitudinal fibers of the beam is constant and equal to

N
ε=
A(E)

while that of the transversal fibers is

νN
εt = −
A(E)

The normal stress is proportional to the Young’s modulus

E(r) N
σ(r) =
A(E)

44
3.4 Eccentric normal force
The interaction is called eccentric normal force or composed flexure when
both, the normal force N and the bending moment M̄Gf , are different from
zero.
The central axis of the interaction is parallel to the beam axis.
Its intersection C with the plane of the section is called center of the
interaction.
If r̄C is the position vector of C with respect to the elastic section center
GE , the resultant moment vector with respect to C is given by

M̄Cf = M̄Gf − r̄C × N k̄

and therefore, by imposing that M̄Cf = 0 is obtained

M̄Gf
r̄C × k̄ =
N

Figure 3.7: Eccentric normal force on a generic beam model and its relative
cross-section

Pre-multiplying by k̄ is possible to obtain the expression for the position


vector r̄c of the interaction center

k̄ × M̄Gf R̄ M̄Gf
r̄C = =
N N

45
which, written in terms of components, provides the coordinates of the
center of interaction

MGy MGx
xc = − yc =
N N
The distance between |r̄c | and the elastic center G is called eccentricity.
From the following expression of the position vector rc of the interaction
center

N r̄C = k̄ × M̄Gf

can be deduced the following formulas respectively about the longitudinal


strains and the normal stresses on the cross-section

N
ε(r̄) = + [J˜G−1 (E) N r̄C ] · r̄
A(E)
" #
N
σ(r) = E(r̄) + [J˜G−1 (E) N r̄C ] · r̄
A(E)

that in terms of the components are written as

!
N xc yc
ε(x, y) = 1 + 2x + 2y
A(E) ρx ρy
!
EN xc yc
σ(x, y) = 1 + 2x + 2y
A(E) ρx ρy

The neutral axis of the composed flexure n is then defined by the equation

!
xc yc
1 + 2x + 2y = 0
ρx ρy

46
3.5 Homogenous sections
In the case in which the longitudinal fibers of the beam are characterized by
a single Young’s modulus the diagram of normal stresses is proportional to
that one of strains and therefore has a linear trend.

N
ε(r̄) = + [(E J˜G )−1 (k̄ × M̄Gf )] · r̄
EA
N
σ(r̄) = + [(J˜G )−1 (k̄ × M̄Gf )] · r̄
A

Having indicated with J˜G the geometrical inertial central tensor of the
section evaluated with respect to the elastic center G

Z
J˜G = r̄ ⊗ r̄ dA
A

whose components are

Z Z Z
2 2
JGx = x dA JGy = y dA JGxy = xy dA
A A A

in the principal inertial reference the strains and normal stresses in terms
of components have the following expressions

N MGy MGx
ε(x, y) = − x+ y
EA EJGy EJGx
N MGy MGx
σ(x, y) = − x+ y
A JGy JGx

47
Here is a summary of what has been shown so far

Mechanical behavior of homogeneous sections

Eccentric normal force (N 6= 0, rc 6= 0)

N MGy MGx
ε(x, y) = − x+ y
A E JGx E JGy E

N MGy MGx
σ(x, y) = − x+ y
A JGx JGy

Centered normal force (N 6= 0, M = 0)

N N
σ= ε=
A AE

Simple flexure (N 6= 0, M 6= 0)

ˆ Skew flexure

MGy MGx
ε(x, y) = − x+ y
JGy E JGx E

MGy MGx
σ(x, y) = − x+ y
JGy JGx

ˆ Straight flexure (assuming the x -axis as neutral axis


and the y-axis as interaction axi s)

MGx MGx
ε(y) = y σ(y) = y
JGy E JGy

48
Mechanical behavior of non-homogeneous sections

Eccentric normal force (N 6= 0, rc 6= 0)


!
N MGy MGx
ε(x, y) = E − x+ y
A (E) JGx (E) JGy (E)

N MGy MGx
σ(x, y) = − x+ y
A (E) JGx (E) JGy (E)

Centered normal force (N 6= 0, M = 0)

N N
σ=E ε=
A(E) A (E)

Simple flexure (N 6= 0, M 6= 0)

ˆ Skew flexure

MGy MGx
ε(x, y) = − x+ y
JGy (E) JGx (E)
!
MGy MGx
σ(x, y) = E − x+ y
JGy (E) JGx (E)

ˆ Straight flexure (assuming the x -axis as neutral axis


and the y-axis as interaction axi s)

MGx MGx
ε(y) = y σ(y) = E y
JGy (E) JGy (E)

49
Chapter 4

Thermal behavior

4.1 Analysis of thermal loading effects


As already discussed in Chapter 1, the behavior of a structural element due
to a thermal load can vary depending on its constraints configuration, and
the thermal load type (constant or variable temperature):

a large variety of stress states can exist, large compressions where


restrained thermal expansion is dominant and very low stresses where the
expansion and bowing effects balance each other; in the cases where thermal
bowing dominates, tension occurs in laterally restrained and rotationally
unrestrained members, while large bending moments occur in rotationally
restrained members.

Most situations in structures under thermal loads have a complex mix of


mechanical strains due to mechanical loads, and mechanical strains due to
restrained thermal expansion.
These lead to combined mechanical strains which often far exceed the yield
values, resulting in extensive plastification.
The deflections of the structure, by contrast, depend only on the total
strains, so these may be quite small where high restraint exists, but they are
associated with extensive plastic straining.
Alternatively, where less restraint exists, larger deflections may develop, but
with a lesser demand for plastic straining and hence a lesser destruction of
the stiffness properties of the materials.
These relationships, which indicate that larger deflections may reduce
material damage and correspond to higher stiffnesses, or that restraint may
lead to smaller deflections with lower stiffnesses, can produce structural

50
situations which appear to be quite counter-intuitive if viewed from a
conventional structural engineering perspective.
The behavior of a common beam model due to the constraint configuration
and the type of thermal load has already been treated in sections 1.5, 1.6
and 1.7.

In the whole chapter the structural behavior with respect to low thermal
loads, with no fire, will be illustrated.

As it has already been underlined thermal effects on beams is an


argument often ignored, but its importance is remarked by the following
pictures:

51
Figure 4.1: Thermal crack on a concrete bridge

Fig. 4.1 : Due to the temperature difference the cooler portion contracts
more than the warmer one, which restrains the contraction. If the restraint
results in tensile stresses that exceed the in-place concrete tensile strength
then thermal cracks appears.

Figure 4.2: Closely spread thermal cracks on asphalt

Fig. 4.2 : The formation of longitudinal cracks is another typical distress


associated with high temperature placement. Closely spaced transverse
cracks and longitudinal cracks due to thermal shock are more prone
to develop into spalling and punchout distresses in the long term as a
consequence of solar irradiation.

52
Figure 4.3: Vertical or near vertical cracks in poured concrete

Fig. 4.3 : The cracking is occurring at a building corner at a garage door.


There is greater risk that the builder didn’t get the garage footers below the
frost line or that in addition, the un-heated garage is more vulnerable to
frost damage (only in freezing climates of course).

Figure 4.4: Diagonal cracks in poured concrete foundations

Fig. 4.4 : Concrete walls tend to display vertical cracks more similar to
that of Fig. 4.3, but frost heaving at a corner of a concrete wall can produce
diagonal cracks or breaks in that location.

These figures, showing cracks due to thermal loads, have been introduced
in order to illustrate briefly the dramatical consequences of thermal loads
on common real structures.
Therefore a correct valuation of the stress state comes to be fundamental.

The idea is to create a mathematical model based on Navier’s theory of


beams (ref. chapter 3) including thermal loads, that will be introduced in
the next section.

53
4.2 Navier’s theory including thermal loads
The aim is to illustrate a general method that allows to evaluate the effects of
arbitrary changes of thermal scenery interesting the cross-section of a beam.
The context is that of the technical theory of beams based on the assumption
of conservation of planar sections (Euler-Bernoulli hypothesis).

4.3 Measurements of temperature variations


The data of the problem are the measurements of temperature variations
in a set of measurement points which allow to interpolate satisfactorily the
continuous field of temperature variations.

Under conditions of stationary heat flow and thermal conductivity


characteristics homogeneous into the section, the field of temperature
variations is taken with linear trend, described by measurements made at
the vertices of a non-degenerate triangle { θi , i = 1, 2, 3 } .
By assuming a fixed Cartesian reference { O, x, y } orthogonal to the
plane of the cross-section, from the solution of the linear problem

    
x1 y1 1 gθ x θ1

x
 2 y2 1 
g
  θy 

=  θ1 
 

x3 y3 1 θ0 θ3

can be deduced the linear expression of the variation of the temperature


field

θ(r̄) = g¯θ · r̄ + θ0

where the vector g¯θ is the gradient of the temperature field changes and
θ0 represents the value of temperature field.

54
4.4 Elastic equilibrium
The hypothesis of conservation of planar sections implies that the dilation
field of the longitudinal fibers of the beam has a linear expression
ε(r̄) = ḡ · r̄ + ε0
The field of axial elastic dilatations induced in the cross-section, denoted
by εel , is derived from the stress-strain relationship under uni-axial stress
state.
The field of thermal dilatations, denoted as εθ , is proportional to the
coefficient of thermal expansion field α and to the temperature change field
θ.

σ(r̄)
εel (r̄) = εθ (r̄) = α(r̄) θ(r̄)
E(r̄)
The geometrical dilatation field ε is the sum of the elastic and thermal
dilatations, so that
ε(r̄) = εel (r̄) + εθ (r̄) = ḡ · r̄ + ε0
Once the geometrical dilatation has been computed, the field of elastic
expansions of the longitudinal fibers of the beam is found out by subtracting
thermal dilatations from geometrical ones

εel (r̄) = ḡ · r̄ + ε0 − εθ (r̄) = ḡ · r̄ + ε0 − α(r̄) θ(r̄)

The field of normal stresses is then determined by multiplying εel by the


elastic Young’s moduli field E of the longitudinal fibers, as follows
σ(r̄) = E(r̄)(ḡ · r̄ + ε0 ) − E(r̄) α(r̄) θ(r̄) = E(r̄)(ḡ · r̄ + ε0 ) − σθ (r̄)

where the field of thermal variations is equivalent to a field of fictitious


stresses defined as

σθ (r̄) = E(r̄) α(r̄) θ(r̄)

Stresses at the interface of the cross-section are evaluated by subtracting


geometrical dilatations ε and thermal thermal dilatations εθ , all multiplied
by the Young’s modulus. From this difference has been obtained σθ , which
is a computational entity with the dimension of a stress, but it is not a stress.

55
By recalling the definitions of normal force N and bending moment
Mf already viewed in Chapt. 3, and defining the contributions due to
thermal loads, Nθ and Mθ , which are obtained by the fictitious stresses σθ
hence they are fictitious as well, it is possible to derive the following relations

Z Z
N= σ(r̄) dA Nθ = σ θ (r̄) dA
A A
Z Z
Mf = σ(r̄)r̄ dA Mθ = σ θ (r̄)r̄ dA
A A

The parametrs ḡ and ε0 can be derived by solving the follwing linear


problem
    
A(E) Sx (E) Sy (E) ε0 N + Nθ
 Sx (E) Jx (E) Jxy (E)   gx  =  −My − Mθy 
    

Sy (E) Jxy (E) Jy (E) gy Mx + Mθx

The moments of order 0, 1, 2 of the distribution of elastic moduli are


given by

Z
AE = E(r̄) dA Elastic area (Moment of order 0)
A
Z
S̄E = E(r̄) r dA Elastic static moment (Moment of order 1)
A
Z
J˜E = E(r̄) r̄ ⊗ r̄ dA Elastic inertial moment (Moment of order 2)
A

Therefore it is possible to write the elastic equilibrium problem as

" #" # " #


AE S̄E ε0 N + Nθ
=
S̄E J˜E ḡ M̄ + M̄θ

56
If x and y axes are the principal axes for the elastic inertial tensor of the
section, the inertial product JGxy (E) vanishes and the normal stresses are
given by the Navier ’s formula including thermal loads, as follows

N + Nθ MGy + MGθy MGx + MGθx


ε(x, y) = − x+ y
A(E) JGx (E) JGy (E)
!
N + Nθ MGy + MGθy MGx + MGθx
σ(x, y) = E − x+ y
A(E) JGx (E) JGy (E)

57
Then simply by summing respectively stresses and strains, it is possible
to evaluate characteristics of interaction due to mechanical and thermal
loads in homogeneous and non-homogeneous sections

Mechanical and Thermal behavior of homogenous sections

Eccentric normal force (N + Nθ 6= 0, rc 6= 0)

N + Nθ MGy + MGθy MGx + MGθx


ε(x, y) = − x+ y
AE JGx E JGy E

N + Nθ MGy + MGθy MGx + MGθx


σ(x, y) = − x+ y
A JGx JGy

Centered normal force (N + Nθ 6= 0, M + Mθ = 0)

N + Nθ N + Nθ
σ= ε=
A AE

Simple flexure (N + Nθ 6= 0, M + Mθ 6= 0)

ˆ Skew flexure

MGy + MGθy MGx + MGθx


ε(x, y) = − x+ y
JGy E JGx E

MGy + MGθy MGx + MGθx


σ(x, y) = − x+ y
JGy JGx

ˆ Straight flexure (assuming the x -axis as neutral axis


and the y-axis as interaction axi s)

MGx + MGθx MGx + MGθx


ε(y) = y σ(y) = y
JGy E JGy

58
Mechanical and Thermal behavior of non-homogeneous sections

Eccentric normal force (N + Nθ 6= 0, rc 6= 0)


!
N + Nθ MGy + MGθy MGx + MGθx
ε(x, y) = E − x+ y
A (E) JGx (E) JGy (E)

N + Nθ MGy + MGθy MGx + MGθx


σ(x, y) = − x+ y
A (E) JGx (E) JGy (E)

Centered normal force (N + Nθ 6= 0, M + Mθ = 0)


!
N + Nθ N + Nθ
σ=E ε=
A(E) A (E)

Simple flexure (N + Nθ 6= 0, M + Mθ 6= 0)

ˆ Skew flexure

MGy + MGθy MGx + MGθx


ε(x, y) = − x+ y
JGy (E) JGx (E)
!
MGy + MGθy MGx + MGθx
σ(x, y) = E − x+ y
JGy (E) JGx (E)

ˆ Straight flexure (assuming the x -axis as neutral axis


and the y-axis as interaction axi s)

MGx + MGθx MGx + MGθx


ε(y) = y σ(y) = E y
JGy (E) JGy (E)

59
Chapter 5

An application of the theory

On the basis of what has been shown previously it comes to be useful to


illustrate an application of the theory.

The main aim is to describe a common example of material under


temperature variations, apply to it the theory explained in the previous
chapter and finally analyze its behavior in terms of displacement.

A common and very interesting case of thermal loads’ application is


provided by bimetallic strip subjected to uniform temperature variation.

5.1 Bimetallic strip subjected to tempera-


ture rise
The strip consists of two layers of different metals which expand at different
rates as they are subjected to a temperature increase, usually steel and
copper, but also steel and brass.
Its role is to convert a temperature change into mechanical displacement.

The bimetallic strips, first realized in 1759, nowadays are still commonly
used in thermostats, fire alarms, clocks, thermometers, heat engines and
electrical devices, like circuit breakers.

60
The strips are melted together throughout their length.
The two metals, when subjected to a temperature rise, tend to expand, and
because of their different coefficients of thermal expansion, display different
dilatations.

As consequence of this difference in terms of expansion between the two


materials constituting the bimetallic strip, if the two metals are connected,
the whole structure tends to deflect.
An example of bimetallic behavior is illustrated in Fig. 5.1, where the metal
with the higher coefficient of thermal expansion is on the upper side.

Figure 5.1: Bimetallic strip subjected to temperature increase

As follows it is defined the relative change of curvature along the arc


length of the beam, rather the increase of curvature

6 E1 E2 (h1 + h2 )h1 h2 ε
κ=
E12 h41 + 4 E1 E2 h31 h2 + 6 E1 E2 h21 h22 + 4 E1 E2 h32 h1 + E22 h42

Where E1 and h1 are the Young’s modulus and height of material one
and E2 and h2 are the Young’s modulus and height of material two.
ε is the misfit strain, calculated by:

ε = (α1 − α2 ) θ

Where α1 is the coefficient of thermal expansion of material one and α2 is


the coefficient of thermal expansion of material two and θ is the temperature
variation.

61
5.2 Application of Navier’s theory including
thermal loads
The cross-section of the bimetallic strip is composed by nickel on the upper
side, and copper on the lower one, as in Fig. 5.2, where O is the elastic center
of the whole section weighted on the elastic moduli.
Nickel and copper strips of rectangular shape, of basis (B) 10 mm and height
(H) 2 mm, have equal dimensions.

Figure 5.2: Cross-section of bimetallic strip

The bimetallic strip is subjected to a uniform temperature variation


θ = 160 °C .
The two materials, nickel and copper, have respectively coefficients of
thermal expansion αN i and αCu , and Young’s moduli EN i and ECu .
The values of these lasts are reported in the following table.

- α [°C−1 ] E [GP a]
Nickel 1.30 × 10−5 220
Copper 1.70 × 10−5 130

The fictitious stresses equivalent to thermal dilatations are given by

σθ, N i (r̄) = EN i (r̄) αN i (r̄) θ(r̄)

σθ, Cu (r̄) = ECu (r̄) αCu (r̄) θ(r̄)

62
The geometrical areas of nickel and copper laminae are both equal to
A = B × H.
The fictitious normal force equivalent to the thermal dilatation is calculated
as follows
Z
Nθ = E α θ dA = θA (αN i EN i + αCu ECu )
A

Nθ = 160×20×(1.30×10−5 ×220×103 + 1.70×10−5 ×130×103 ) = 16224 N


Fictitious bending moments equivalent to the thermal expansion are
evaluated with respect to the geometrical center of the whole section G.
Z H Z 0
B H2
Mθy = B E α θ y dy + B E α θ y dy = θ (αN i EN i − αCu ECu )
0 −H 2
10 × 22
Mθy = 160 (1.30 × 10−5 × 220 × 103 − 1.70 × 10−5 × 130 × 103 )
2
Mθy = 2080 N mm
Hence the cross-section is subjected to normal force and simple flexure
(ref. Sect. 3.4).

The moments of order 0, 1 and 2 of the distribution of Young’s moduli


are given by

A(E) = B H (EN i + ECu ) = 2 × 10 (220 + 130) × 103 = 7 × 106 N


B H2 10 × 22
Sy (E) = (EN i − ECu ) = (220 − 130) × 103 = 1.80 × 106 N mm
2 2
B H3 10 × 23
Jy (E) = (EN i + ECu ) = (220 + 130) × 103 = 9.33 × 106 N mm2
3 3
By imposing the elastic equilibrium conditions, the dilatation ε0 of the
fiber passing through the origin of reference and the bending curvature
along the y-axis gy can be deducted, as follows
( (
Nθ = A(E) ε0 + Sy (E) gy 16224 = 7 × 106 ε0 + 1.80 × 106 gy
Mθy = Sy (E) ε0 + Jy (E) gy 2080 = 1.80 × 106 ε0 + 9.33 × 106 gy
The solutions of the system above are

ε0 = 2.38 × 10−3 gy = −2.36 × 10−4 mm−1

63
Considering that the bimetallic strip structural scheme is a shelf, its
deformation due to the temperature variation is reported in Fig. 5.3.

Figure 5.3: Bending of a shelf

Assuming that the length of the strip is l = 10 cm, and being the rotation
uniform and the curvature constant along the strip, the rotation center is in
the middle of the shelf.
It is possible to calculate firstly the rotation of the free ending section φ,
and secondly the lowering v.

φ = gy × l = −2.36 × 10−4 × 100 = −2.36 × 10−2


l 100
v =φ× = −2.36 × 10−2 × = −1.18 mm
2 2
The diagram of geometrical dilatation is linear and varies with the
following law

ε(y) = gy · y + ε0 = −2.36 × 10−4 · y + 2.38 × 10−3


In order to evaluate the stresses comes to be necessary to firstly calculate
the dilatations εθ due to the temperature variation, in nickel and copper, and
then compute the fictitious stresses σθ equivalent to the thermal expansion

εθ (y) = α(y) θ

εθ (y) = 1.3 × 10−5 × 160 = 2.08 × 10−3 In nickel


εθ (y) = 1.7 × 10−5 × 160 = 2.72 × 10−3 In copper

64
σθ (y) = E(y) α(y) θ

In order to draw the geometrical dilatations diagram (ref. Fig. 5.4) it


is convenient to evaluate dilatations at the upper and lower sides of each
homogeneous layer

ε (y = 2 mm) = −2.36 × 10−4 × 2 + 2.38 × 10−3 = 1.90 × 10−3


ε (y = −2 mm) = 2.36 × 10−4 × 2 + 2.38 × 10−3 = 2.85 × 10−3

The elastic dilatations are given by the difference of geometrical


dilatations and the dilatations due to the thermal load.
Then the stresses can be calculated by multiplying the elastic dilatation by
the Young’s modulus (different for the two metals).
In order to draw the stress and elastic dilatation diagrams have been
derived stresses and elastic dilatations at the upper and lower sides of each
homogeneous layer, and on the interface for the two metals.
Compression stresses have been assumed negative, and traction ones positive.

εel (y) = gy · y + ε0 − α(y) θ

εel (y = 2 mm) = 1.90 × 10−3 − 1.30 × 10−5 × 160 = -1.80 ×10−4


εel (y = 0+ ) = 2.38 × 10−3 − 1.30 × 10−5 × 160 = 3.00 × 10−4
εel (y = 0− ) = 2.38 × 10−3 − 1.70 × 10−5 × 160 = −3.40 × 10−4
εel (y = −2 mm) = 2.85 × 10−3 − 1.70 × 10−5 × 160 = 1.30 × 10−4

σ (y) = E(y)(gy · y + ε0 ) − E(y) α(y) θ

σ (y = 2 mm) = −1.80 × 10−4 × 220 × 103 = −39.60 N/mm


σ (y = 0+ ) = 3.00 × 10−4 × 220 × 103 = 66.00 N/mm
σ (y = 0− ) = −3.40 × 10−4 × 130 × 103 = −44.20 N/mm
σ (y = −2 mm) = 1.30 × 10−4 × 130 × 103 = 16.90 N/mm

65
The two materials are welded along the whole length of the strip, so that
the dilatation diagram of longitudinal fibers must be continuous trough the
interface between nickel and copper.

In Fig. 5.4 are illustrated respectively:


ˆ The uniform temperature variation trough the depth on the strip (θ)
ˆ The dilatations due to the temperature variation (α θ)
ˆ The linear geometrical dilatation diagram (ε)
ˆ The elastic dilatation diagram (εel )
ˆ The stresses diagram (σ).

Figure 5.4: Dilatation and stress diagrams at the cross-section of the


bimetallic strip

With reference to Fig. 5.4 the θ diagram is constant since temperature


rise is uniform, and the discontinuity of the α θ diagram is given by the
difference in terms of α of the two materials.
The ε diagram is linear because of the assumption of conservation of planar
section, while εel and σ diagrams are only piecewise linear with an essential
discontinuity in correspondence of the interface between the two metals with
different Young’s moduli and coefficient of thermal expansion.

Since the characteristics of interaction are null, the normal stress field
on the cross-section is globally self-balanced.

If mechanical interactions are acting in the strip, their moments of


order 0 and 1 would have to be added to the ones of the fictitious stresses
equivalent to thermal increase, in order to solve the thermoelastic problem.

66
Bibliography

[1] Giovanni Romano: ”Scienza delle Costruzioni”, Tomo I, Tomo II,


Tomo III, Hevelius, Benevento, (2002)

[2] A.S. Usmani, J.M. Rotter, S.Lamont, A.M. Sanad, M. Gillie: ”Fun-
damental principles of structural behaviour under thermal effects” Fire
Safety Journal 36 (2001) 721-744

[3] Todd A. Helwig, Karl H. Frank, Michael D. Engelhardt, John L.


Tassoules, Eric B. Becker : Effects of thermal loads on Texas steel
bridge

[4] Barbara Klemczak, Agnieszka Knoppik-Wrobel: ”Early age thermal


and shrinkage cracks in concrete structures - Description of the prob-
lem” (2/2011)

[5] Paul Acker, Jean-Michel Piau, Isabelle Radouant: ”Modelling thermal


and hygrometric effects in concrete” (05/2014)

[6] Mustafa K. Badrah, Mansour N. Jadid : ”Investigation of Developed


Thermal Forces in Long Concrete Frame Structures” The Open Civil
Engineering Journal, 2013, 7, 210-217

[7] René de Borst, Paul P.J.M. Peeters : ”Analysis of concrete structures


under thermal loading” (05/1989)

[8] J. Qiu, J. Tani, J. Takagi : ”An intelligent piezoelectric composite


material without bending deformation” Journal of technical physics
(1994)

[9] Hemangi K. Patade, Dr. M. A. Chakrabarti ”Thermal Stress Analysis


of Beam Subjected To Fire” nt. Journal of Engineering Research and
Applications (10/2013)

67

View publication stats

You might also like