You are on page 1of 11

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/281294489

Optimun Design of Reinforcement in Concrete


Plates and Shells

Conference Paper · July 2002

CITATIONS READS

5 40

2 authors:

Antonio Tomas Pascual Martí-Montrull


Universidad Politécnica de Cartagena Universidad Politécnica de Cartagena
50 PUBLICATIONS 71 CITATIONS 52 PUBLICATIONS 253 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Antonio Tomas
letting you access and read them immediately. Retrieved on: 07 October 2016
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", Vienna, Austria, July 2002. (Pre-print version)

Optimun Design of Reinforcement in Concrete Plates and Shells

A. Tomás* and P. Martí

Department of Structures and Construction


Cartagena Technical University, Campus Muralla del Mar, 30202 Cartagena, Murcia, Spain
e-mail: antonio.tomas@upct.es

Key words: optimization, reinforcement, concrete, shells, plates

Abstract
The use of finite element program packages as a design tool for reinforced concrete structures has
increased considerably in recent years. All structural codes of practice recognize the assumption of
linear elastic behavior at a design stage, however, the absence of universally accepted solutions for the
design of reinforcement in plates and shells and, therefore, the scarcity of provisions in the structural
concrete codes, gives rise to the problem of calculating the required reinforcement in these structures.
The constant development of the computers' performance and storage capacity combined with
powerful numerical methods revealed the need of a standard procedure to design elements subject to
membrane and flexure forces. In the present study, the amount of reinforcement is optimized locally
for each finite element, from the equilibrium between applied and internal forces, which leads to an
indeterminate system of nonlinear equations. The biaxial behavior of concrete and different lever arms
for all the internal forces are included in the formulation. Ideal plastic behavior is assumed for both
materials. The analysis and optimization modules of the finite element program ANSYS are used. The
objective function is the combination of the tensile forces in the reinforcement, which are to be
positive and are a function of the concrete strut directions in top and bottom layers in the element.
These directions are taken as design variables of the problem. The accuracy of the optimum solution
reached depends on the different tolerances selected by the designer. One of these tolerances refers to
the convergence of the thickness of each layer on the expression proposed by the structural concrete
code used in the problem. A numerical example is given and compared to the results provided by some
authors.
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

1 Introduction

Traditionally, the failure conditions of the materials in the field of concrete shells have been applied to
the stresses at the structure points. Even though this check methodology has been adopted by most of
researchers, the basic hypotheses and methods to solve the problem are still open to debate at present.
As proof of this, it can be seen that the references in the design codes referring to the design of plates
and shells are quite scarce and, in the majority of cases, they are dealt with rather superficially:
Eurocode 2-1991 (EC2) does not include any reference to shells, and only refers to plates that are
loaded on their plane, while the ACI 318-99 code considers any method of design which assures
sufficient strength with equilibrium is considered applicable. In the Model Code CEB-FIP 1990
(MC90) there is a more in depth treatment, given that design hypotheses are included which are based
on the use of layer models [1], resisting the external membrane forces and the internal shear force.
Nevertheless, there are design methods which use plate or shell elements, where each individual
element is considered as a dimensioning unit that has to resist their nodal forces. The objective is to
reach the equilibrium between the external and internal forces due to the reinforcement and the
concrete contributions. To this end, calculus algorithms are used which provide the quantity of
reinforcement at the outer layers of the element in two orthogonal directions [2][3], and even strategies
have been developed which enable the obtention of more rational reinforcement distribution, with less
weight in the concrete plates [4].
The development of the optimization techniques, which are used in an environment of progressive
advances in the power of the computers, as well as in their graphic capacity, can be very effective
tools, enabling us to obtain alternative reinforcement distribution, satisfying the design conditions
(stress constrictions, compatibility conditions, construction prescriptions, etc.) in an optimal way
(minimum weight, minimum stress level, etc.) [5].
In this paper, we will adopt a design formulation of reinforced concrete shell elements in which
the amount of reinforcement is determined optimally in each mesh element which models the
geometry of the problem, starting from the equilibrium that is set between the applied forces and the
internal forces. The formulation of the method includes the biaxial behavior of the concrete and
different lever arms of the reinforcement, assuming ideal plastic behavior for both materials [6]. The
ANSYS finite element program has been used in order to analyze the structure and to obtain the forces
in the shell elements. Furthermore, the formulation has been implemented by means of user routines
within the optimization module of the program.

2 Formulation and solution of the structural optimum design problem

The aim of the structural optimum design is to obtain a design, that is to say, a set of values for the
design variables, which minimizes an objective function, and which complies with a series of
constrictions that depend on the variables.
The design variables of a structure can be: properties of the cross section of the elements (surface
areas, thicknesses, inertia moments, etc.); structure geometry parameters; structure topology (nodes
and element connections), and properties of the material of the structure. The type of optimization to
be carried out will depend on the type of variables that are being considered. Traditionally, the design
of minimum weight structures has been sought after, which has led to the fact that the most common
objective function is the weight of the structure. Nevertheless, in other applications the weight is not
the determining factor, and other objective functions are resorted to, such as the cost, the reliability,
the stiffness, etc. The constrictions are the conditions that the design must be comply with in order for
it to be taken as valid.
In mathematical terms, the optimum design problem can be formulated as follows:

2
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

To find the variable vector of design x which


minimizes: f (x)
subject to: hj (x) = 0 j = 1, 2, ..., mi
gk (x)  0 k = 1, 2, ..., md
xiL xi  xiU i = 1, 2, ..., n
where: x is the n-dimensional vector of the design variables; f (x) is the objective function; hj (x) is the
j equality design constriction; gk (x) is the k inequality design constriction; mi is the number of equality
constrictions; md is the number of inequality constrictions; n is the number of variables, and xiL (xiU) is
the lower limit (upper) of the variable i.
The problem having been approached in this way can be solved, transparently for the user, by
means of several different methods (Mathematical Programming, Genetic Algorithms, etc.).

3 Formulation of the optimum design problem of reinforcement in shell


elements

3.1 Description of the material

Concrete is a complex material that requires a large number of parameters in order to provide a
complete description of its constitutive equations. Nevertheless, in the design model that has been
adopted [6], and which is quite sufficient to meet the proposed aim, concrete is considered as a rigid-
plastic behaving material, which is characterized by a single parameter, the fc uniaxial concrete
compressive strength measured on cylinders. With respect to the general case of the design, the choice
of fc is a delicate matter, for which reason the values suggested in MC90 for the design strength of
concrete will be adopted:
Non-cracked zones
 f 
f cd 1  0,851  ck  f cd (1)
 250 
Cracked zones where the compressive strength may be reduced by the effect of transverse tension
from the reinforcement, and by the need to transmit force through the cracks
 f 
f cd 2  0,601  ck  f cd (2)
 250 
where:
fck is the characteristic cylinder compressive strength of the concrete in MPa, and
fcd is the design cylinder compressive strength of the concrete in MPa.
The strength is increased for concrete that is subject to a biaxial compression state, it being
possible to reach an increase of up to 16 % under a stress state of the same order in both directions [7].
This increase, due to the confinement effect, can be formulated by multiplying the compressive
strength by a coefficient K
1  3,65 
K , with   2 ;  2   1 (3)
1   2
1
where: 2 is the failure second principal stress and 1 refers to the failure minor principal stress,
considering the compressions with a negative sign.

3
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

As is normal when dealing with the design of concrete structures, neither the tensile strength
of the concrete, nor the aggregate interlock phenomena will be taken into account.
A rigid-plastic behavior has been considered for reinforcement steel, with a maximum stress
that is equal to the yield limit. The reinforcement only resists uniaxial forces, which gives rise to the
fact that the dowel action of the rebars is not taken into account. The contribution of the reinforcement
in compression resistance has been discarded as a consequence of the fact that its effect is slight in
comparison to the surrounding concrete. Therefore, the reinforcement is designed solely to resist
tensile forces, and the effects related to the adhesion and the anchoring of the rebars have not been
taken into account. The orthogonal distribution of the reinforcement has been considered.

3.2 General approach

The flexural and membrane forces which act on the sides of a shell element (Figure 1) must be in
equilibrium with the internal compression forces in the concrete and the tensile forces in the
reinforcement. In general, the principal directions of the membrane and flexural forces do not coincide.
Figure 2 shows a model that is used for a shell element the reinforcement set parallel to the x and
y axes, and also shows the acting internal forces. The tensile forces of the reinforcement are identified
as Nsxt, Nsyt, Nsxb, Nsxt, with the subscripts x and y, which refer to the respective axes, and the subscripts
t and b which refer to the top and bottom layers, respectively. Different lever arms of the
reinforcement (Figure 2a) have been included in the formulation. We can suppose a vertical failure
plane parallel to the direction of the cracks on the top layer, the normal vector of which forms an angle
t with the x axis, and which is contained in the xy plane. The thickness of the compressed top layer is
represented by at, within which the uniform distribution of the stresses (rectangular stress block) is
assumed, the resultant of which is Nct, and is parallel to that normal vector. In the same way, with
respect to the bottom layer, b refers to the normal vector of the failure plane, ab to the thickness of the
uniform distribution of the stresses, and Ncb with their resultant.
Mxy Mxy
My Nxy
x x

Mx
Nxy
Mxy Mxy Nx Nx

Mx Nxy

My Nxy

y Mxy y Ny

Figure 1: Flexural and membrane forces in shell element (per unit length).

This formulation, which was originally developed for the case of reinforcement that is needed
in the top and bottom layers [8], was later widened to cover three other cases: reinforcement that is
only needed at the bottom layer, reinforcement that is only needed at the top layer, and unnecessary
reinforcement [2][6]. Even though the phenomena described is relatively simple, this is not the case
with its mathematical solution, given that the development of a formulation is required which leads to
an indeterminate system of non-linear equations for each case.

4
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

y x y
x
z
z
hyt hxt t at
hyb hxb
ab
Nsyt
Nsxt Ncb
Nct b

Nsxb x
Nsyb

(a) (b)
Figure 2: (a) Model of a shell element including reinforcement. (b) Internal forces.

3.3 Objective function

The objective function that is used is the sum of the tensile forces of the reinforcement, where its
expression for each case is as follows:
Reinforcement needed for top and bottom layers
N tot  N sxt  N syt  N sxb  N syb (4)

Reinforcement needed only for the bottom layer


N tot  N sxb  N syb (5)

Reinforcement needed only for the top layer


N tot  N sxt  N syt (6)

On optimizing the tensile forces, the reinforcement cross-section area is also optimized, given its
relation by means of the steel design strength fyd, and therefore, the weight of the steel.
All of the objective functions that are used have been normalized by dividing their value by the
initial value adopted in the optimization process, thus they take on values approaching the unit, which
have similar range of values as the design variables. This technique provides good results when non-
linear mathematical programming is used in order to solve the optimum design problem.
A tolerance value of 10-4 has been defined in order for the optimization process to be stopped in
some of the following two cases, whenever the variation in the objective function in two consecutive
iterations is lower than the tolerance, or whenever the variation in the objective function between the
last iteration and the minimum value obtained in a previous iteration be less than the tolerance.

3.4 Design variables

Reinforcement needed in the top and bottom layers. As has been stated above, the difficulty of the
problem resides in trying to solve the equilibrium equation system. Even though the unknowns that we
are interested in are the tensile forces of the reinforcement (Nsxt, Nsyt, Nsxb and Nsxt), there are another
four unknowns (at, ab, t and b), thus we have a system of six equilibrium equations with eight
unknowns. This implies searching for t and b values in order for the total amount of reinforcement to

5
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

be minimal in order to satisfy the equilibrium equations. Therefore, the t and b angles have been
employed as optimization variables.
The MC90 states that by adopting 45º for the compressed truss of the concrete, a minimum local
reinforcement is obtained. Therefore, the values t = b = /4 phase shifted by 2, have been used as
initial values for the design variables in the various validation examples of the procedure, the results
obtained being satisfactory. The phase shift is used as a strategy to avoid negative or zero values in the
design variables, given the fact that the ANSYS program does not allow their use in its optimization
module. In this way, the variation ranges are:
3/2  b  2 (7)

2  t  5/2 (8)
The design variables are forced to adopt the extreme values of the variation intervals in case they
be too close to them; for example, an angle of between 10º and 15º. This is done for two basic reasons.
The first is concerned with taking into consideration the recommendation that is set forth in the MC90
with respect to the fact that the tensile and compression forces must be separated by at least 15º, given
that it is not reasonable to accept that, for two very near directions as regards the angle, we could come
across tension in the reinforcement and compression in the concrete, given that this would infringe
upon the displacement compatibility condition. The second reason is to make sure of the convergence
of the optimization process in the solution of a non-linear equation system. Then we can, within the
framework of the formulation proposed, interpret that for t and b values which are close to 0 or /2,
the direction of the force of compression in the concrete tends to be parallel to any of the
reinforcement directions, which leads to a compression force in the reinforcement. Given that this in
unacceptable according to the formulation that has been proposed, if this situation occurs it will imply
that that reinforcement is not needed, thus enabling the achieving of the convergence.
The design variables have been normalized by dividing their value by the initial value that has
been chosen in the optimization process. These normalized values of the design variables are used by
the optimization algorithm, for that reason we also refer to them as optimization variables.
Finally, a tolerance value of 10-6 has been defined in order for the optimization process to come to
a halt in case the variation in the design variables in two consecutive iterations be less than this value.
Reinforcement needed only at the top layer. In this case, the unknowns that we need are the tensile
forces of the reinforcements Nsxb and Nsyb. Nevertheless, there are another six unknowns, such as at, ab,
b, Ncxt, Ncyt and Ncxyt, so we have a system of seven equations (the six equilibrium equations and an
check equation of the maximum compressive stress at the top layer) with eight unknowns. This implies
looking for the b values in order for the total amount of reinforcement to be minimal and to satisfy the
seven equation system. Therefore, the b angle has been used as an optimization variable.
b = 7/4 is adopted as the initial value of the design variable, i.e. /4 phase shifted by 2 for
the reasons that have been given in the previous section. The range of variation is that which is
adopted in (7). In the same way, for values of b close to 3/2 or 2, the direction of the compressive
force in the concrete is very close to the direction of the reinforcement, which gives rise to it being
unnecessary. The variation tolerance of the design value takes on the value 10-6.
Reinforcement needed only at the bottom layer. This case is analogous to the one described above
for the top layer. Here the angle t has been used as an optimization variable, and t = 9/4 has been
adopted as the initial value of the design variable, i.e. /4 phase shifted by 2. The variation range is
that which is adopted in (8). In the same way, t values that are close to 2 or 5/2 gives rise to the
fact that the reinforcement is not necessary. Once again, the tolerance adopted is 10-6.

6
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

3.5 Constrictions

We will also call the constrictions state variables, given that this is the nomenclature used in the
ANSYS program, whenever they are parameters that define the design space, and therefore, the
validity status of the design in each iteration. As with the objective function and the design variables,
the constrictions have been normalized adopting similar values between them and close to the unit,
thus managing to get the mathematical programming algorithm evolve satisfactorily. A tolerance value
of 10-6 has been defined, for which the optimization process comes to a halt whenever it comes up a
lower variation in the constrictions between two consecutive iterations.
Inequality constrictions. The inequality constrictions used consist in limiting the tensile forces of the
reinforcement (Nsxt, Nsyt, Nsxb and Nsxt) in order for them not to take on negative values, given that this
would imply compressive forces in the reinforcement, which is unacceptable according to the
formulation being proposed. The constrictions are:
N sij 0 (9)

where Nsij is the tensile force of the reinforcement in the i (x or y) direction in the j (t top or b bottom)
layer. The constrictions can be normalized in the following way:
N sij
gijN  0 (10)
N sij , max
where Nsij,max refers to the greater initial tensile force.
Equality constrictions. In the various cases that have been analyzed, the values of the thicknesses of
the top and bottom layers (at y ab) come up in the equation system. These unknowns will be treated as
constrictions or state variables in such a way that their value will converge towards the thickness that
is needed to resist the compressive force in the layer. The design condition is:
a j  a j, N (11)

where aj refers to the thickness of the j layer in each iteration and aj,N is the thickness that is needed to
resist the compressive force in the j layer, which gives us:
N cj
a j, N   (12)
Kf cd 1
The relevant constriction will have the form:
h aj  a j  a j , N  0 (13)

which can be normalized as follows:


a j  a j ,N
h aj  0 (14)
Da ,max
where Da,max refers to the greater difference between the initial aj and aj,N values.
Actually, the restriction that is applied is not a strictly equality constriction, given that the
convergence of aj towards the value aj,N is considered to have been reached when a certain tolerance T,
which has been set by the designer, is satisfied. This is a function of the degree of accuracy that is
desired, which besides has a bearing on the decreasing of the values achieved for the objective
function. Therefore, the relevant restriction is formulated in the following manner:
0  h aj  T (15)

where it is also necessary to normalize the tolerance by dividing it by Da,max. We deem it acceptable to
adopt a hundred-thousandth part of the thickness h of the element, T = 10-5h.

7
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

4 Example

By way of example in order to obtain the optimum design of reinforcement, the concrete plate used in
[6] has been employed so as to be able to compare the results that are obtained. What we have is a
rectangular plate which is simply supported in the four edges. It is 0,15 m thick and has a span of 5,0 
6,0 m2. It is subject to a vertical load p of 15,0 kN/m2, which is uniformly distributed, and which
includes its own weight. It is also subject to an uniform linear load q around the edges of the plate
(Figure 3a). We will use two q values in our study, supposing a simple flexural state in the plate (q = 0
kN/m), and supposing a compressive and flexural state with a value of q = 150 kN/m. The materials
are 20 MPa compressive strength concrete and 400 MPa tensile strength steel. The distance from the
center of gravity of the layers of reinforcement to the exterior plate fibers is 0,025 m. The mesh of the
finite element model is represented in Figure 3b.
y q

q p = 15 kN/m2 q 6m

405 nodes
x 120 shell elements
q
5m 2430 d.o.f.

(a) (b)
Figure 3: (a) Geometry and loads for a simply supported plate. (b) Finite element model.

(a) (b)
Figure 4: Reinforcement densities parallel to x axis (q = 0 kN/m). (a) Top layer. (b) Bottom layer.
The reinforcement densities in both directions for each plate layer in the case of simple flexural
state are represented in Figures 4 and 5, where the units are expressed in mm2/m. The geometry to be
seen in Figure 5 reveals how the reinforcement densities are grouped following the principal stress
directions, turning out to be non-existent for the most part of the top layer, given that there fitting is
not needed, and obtaining greater quantities in the reinforcement according to the x axis with respect to
the y axis as a result of the difference in length of their sides.
In Figures 6 and 7 we are given comparative graphs between the reinforcement values obtained
by Lourenço and Figueiras [6] resolving a non-linear equation system and the optimum reinforcement

8
A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

in accordance with the method being proposed, for the two reference cases according to the value that
q takes.

(a) (b)
Figure 5: Reinforcement densities parallel to y axis (q = 0 kN/m). (a) Top layer. (b) Bottom layer.
Asy,t Lourenço (1993) Asy,b Lourenço (1993) Asy,b Lourenço (1993) Asy,b Tomas (2002)
Asy,t Tomas (2002) Asy,b Tomas (2002)
0
600 -50 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5

400 -100
-150
200 A sy (mm2/m) -200
A sy (mm2/m)
Top (+)
Top (+) 0 -250
Bottom (-)
Bottom (-) 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 -300
-200
-350
-400
-400
-600 -450
x axis (m) x axis (m)

(a) (b)
Figure 6: y-Reinforcement in a cross-section parallel to x axis (q = 0 kN/m).
(a) Cross-section close to the supports. (b) Cross-section at mid-span.
Asy,t Lourenço (1993) Asy,b Lourenço (1993) Asy,b Lourenço (1993) Asy,b Tomas (2002)
Asy,t Tomas (2002) Asy,b Tomas (2002)
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5
300 0
-20
200
-40
100 -60
A sy (mm2/m) 2
A sy (mm /m) -80
Top (+) 0 Top (+) -100
Bottom (-) 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 Bottom (-) -120
-100
-140
-200 -160
-180
-300 -200
x axis (m) x axis (m)

(a) (b)
Figure 7: y-Reinforcement in a cross-section parallel to x axis (q = 150 kN/m).
(a) Cross-section close to the supports. (b) Cross-section at mid-span.
In the case of simple flexural state (q = 0 kN/m) important reinforcement differences arise at the
sections that are near the supports, which by expressing them in percentage terms we get mean values
to the order of 34 % for the top reinforcement and to the order of 10 % for the bottom reinforcement at

9
400

A. Tomás, P. Martí, "Optimun Design of Reinforcement in Concrete Plates and Shells ", in H.A. Mang, F.G.
Rammerstorfer, J. Eberhardsteiner, (Editors), "Proceedings of the Fifth World Congress on Computational
Mechanics", July 2002. (Pre-print version)

the section beside the supports (Figure 6a). On analyzing the sections that are situated nearer the center
of the span, we can notice that the fitting of top reinforcement is not necessary, which implies that the
optimization process has less variables with which to achieve the optimum reinforcement in the center
section (Figure 6b).
In the case of compressive and flexural state (q = 150 kN/m) the reinforcement differences are
less, which is quite logical if we bear in mind that the need for reinforcement is less in elements that
are subject to a compressive and flexural stress state respect to those that are subject to simple flexure.
On concreting these differences percentage-wise, we get 5 % less reinforcement in the section that is
beside supports (Figure 7a) and, in the same way as the previous case, a practically non-existent
difference in the central section (Figure 7b).

5 Conclusions

Traditionally, computers have been used within the structure design process to analyze the response of
a structure (defined by the user) and to check its safety with respect to given loads. The use of
optimization techniques in the structure design process widens the field of computer use and enables
the user to obtain optimum designs for the design conditions that he or she has set.
In this paper a reinforcement optimum design formulation has been presented for concrete shell
elements, for the solution of which a finite element program has been used in which the formulation
has been implemented by means of user routines within the program optimization module. It may be
concluded on the basis of the results that have been obtained, that the differences observed between the
amount of reinforcement obtained by using optimization techniques and that obtained by means of
traditional methods may be considerable, it being possible to achieve savings of over 30% in some
sections of the structural element, and savings of between 10% and 15% in the whole assembly.

References

[1] P. Marti, Dimensioning and detailing, IABSE, 62, Stuttgart, Germany (1991), 411-443.
[2] P. B. Lourenço and J. A. Figueiras, Solution for the design of reinforced concrete plates and
shells, J. Struct. Eng., 121(5), (1995), 815-823.
[3] C.-S. Min, Design and ultimate behavior of reinforced concrete hyperbolic paraboloid saddle
shell, R. Astudillo and A. J. Madrid eds., Proc. 40th Cong. IASS, Madrid, 1999, CEDEX, Madrid
(1999), Vol. 1, pp. A89-A98.
[4] S. M. R. Tabatabai, Finite element-based elasto-plastic optimum reinforcement dimensioning of
spatial concrete panel structures, Ph.D. Thesis, Institute of Structural Engineering, ETH Zurich:
Birkhäuser Verlag Basel, 1996.
[5] P. Martí, A. Tomás and M.S. Torrano, The application of optimization techniques in structural
concrete teaching, ACHE ed., 1st ACHE Conference in structural concrete teaching, Madrid,
2001, ACHE, Madrid (2001), pp. 203-209 (in Spanish).
[6] P. B. Lourenço and J. A. Figueiras, Automatic design of reinforcement in concrete plates and
shells, Engineering Computations, 10, (1993), 519-541.
[7] G. Hofstetter and H. A. Mang, Computational Mechanics of Reinforced Concrete Structures,
Vieweg, Braunschweig/Wiesbaden, Germany (1995).
[8] A. K. Gupta, Combined membrane and flexural reinforcement in plates and shells, J. Struct. Eng.,
112(3), (1986), 550-557.

10

You might also like