Escolar Documentos
Profissional Documentos
Cultura Documentos
2/1/2007
Genetic algorithms (GA), a search method inspired by Darwins theory of evolution, offer an optimisation tool has been used very successfully to solve a variety of engineering problems. The search process it implements starts with a set of one or more chromosomes (initial population) and, by applying selection and reproduction operators, iteratively evolves the population into hopefully better ones, until a stopping criteria is reached. This article investigates lattice-reinforced joist slab cost optimisation problems using a GA with continuous variables. The problem considered concerns one-way slabs, continuous over two spans, which only the in-situ concrete characteristics and joist spacing are varied. The design variables are: concrete layer thickness, concrete layer strength, reinforcement, distance between joists and degree of redistribution of the continuous slabs negative moments. The search for a solution includes an investigation into the use of discrete variables for data representation. To obtain results that allow for a comparative empirical analysis, these problems are also evaluated by a conventional optimisation method. The results indicate that the GA method is a viable optimisation tool for solving lattice-reinforced joist slab cost minimisation problems.
Vanessa Cristina de Castilho Universidade Federal de ndia, Brazil Uberla Maria Cristina Vidigal de Lima Universidade Federal de ndia, Brazil Uberla
Notation
h Map
Mel f (x ) x1 x2 x3 x4 degree of redistribution negative bending moment after the redistribution negative bending moment for elastic-linear material cost function, in R$/m2 cast-in-place concrete, in m compressive strength of the concrete, in MPa distance between joists axes, in m degree of redistribution, in %
than that of optimisation. In the area of precast concrete structures, the optimisation of elements is of major interest due to the way these elements are produced. In Brazil, one of the most common applications of precast concrete elements is in buildings slabs. These slabs usually consist of latticereinforced joists, prestressed concrete joists, TT panels and hollow core panels. In precast concrete elements, part of the problem is the transitory stages of production, transportation and assembly, which may impose more unfavourable loads on these elements than on cast-in-place structures. These stages consist of:1 (a) production execution of precast concrete elements; (b) transportation moving from the production to the building site; (c) assembly placement of the elements in their permanent position and execution of the connections. A robust solution to the problem of cost minimisation of precast concrete structures requires that all these stages be taken into account. This article analyses the cost optimization of lattice-reinforced joist slabs based on genetic algorithms (GAs).2,3 The problem variables are the height and strength of the concrete, the reinforcement, the distance between joists and the coefcient of redistribution of the negative moments of continuous slabs. Based
on these factors, the inuence of the continuous reinforcement is evaluated in the structures nal cost.4,5 The variables are dened initially as continuous and later as discrete, since it is much easier to associate a discrete set of values to the variables values because they are more easily applied. To investigate the potential of the GA, the same problems were addressed using a conventional method, the Excel Solver, which uses the nonlinear optimisation code Generalized Reduced Gradient (GRG2).
Genetic algorithm
The genetic algorithm (GA) is an optimisation and search method based on the concepts of genetics, i.e. the evolutionary mechanisms of populations of live beings. GAs were inspired on the principle of natural selection and survival of the ttest established in 1859 by Charles Darwin in his book The Origin of the Species. According to Darwins theory, in any given population, the individuals possessing good genetic characteristics have better chances for survival and reproduction than the less t individuals, who tend to disappear over time. The GA simulates biological evolution by means of a multidirectional search within the space of potential solutions for the problem. This algorithm maintains a constant number of potential solutions (population), modifying the population in each successive generation
Introduction
Researchers in the eld of structural engineering have always attempted to devise optimised design solutions. Indeed, few topics have received more attention in structural analyses
so that good solutions can reproduce and pass on to the next generation, while bad solutions are discarded. GA generally uses probabilistic transition rules to select some solutions for reproduction and others to be discarded. The basic principles of GA basic were established in Holland6 and are mentioned in many bibliographical references.7 11 Each individual in a population (called a chromosome) usually corresponds to a point in the search space and represents a possible solution to the problem this solution is also called a hypothesis. The GA can explore the space of possible solutions to seek the best one by applying its reproduction mechanism on the individuals of the current population. Instead of starting from a single point (or potential solution) in the search space, a GA is initialised with a population of potential solutions. These potentials solutions are normally generated randomly and represent dispersed points in the search space. A typical GA uses three operators selection, crossover and mutation to guide the population (through several generations) toward a convergence at the global optimal point. After the selection, crossover and mutation operators have been applied, a new population is formed. The process is repeated until a given number of generations have been created or another stop criterion is reached. These characteristics, organised into a procedure, can be rewritten as the pseudo-code shown in Figure 1. This optimisation tool is widely employed in the search for solutions to innumerable problems in the eld of engineering.12 17
The concrete base is moulded in metal moulds to ensure the quality of the concrete, which is applied in 2 3 cm thickness, using concrete with small rich aggregates in a aggregate cement paste to avoid the vibration operation (Figure 3). Lattice-reinforced joists are usually produced in lengths varying from 80 300 mm, with possible one-centimetre variations. Each type of joist is identied by a set of symbols, as illustrated in the example below,24 where TR-08634: TR characterises the latticereinforced joist; 08 indicates the 8 cm high lattice reinforcement; 6 indicates the f6 mm gauge of the lattices upper bar; 3 indicates the f3 . 4 mm gauge of the diagonals; and 4 indicates the f4 . 2 mm gauge of the lower bars. Lattice-reinforced joists can be produced by order to include additional reinforcements introduced in the concrete base without complicating the production process. El Debs1 points out that the use of this type of lattice-reinforced joist favours the application of bi-directional reinforcement slabs. The Brazilian Civil Construction Committee is currently studying the elaboration of the Code for Prefabricated Slabs. Therefore, the calculations normally used by design engineers to dimension and check lattice-reinforced joist slabs are those published by manufacturers of precast elements, according to Diniz,25 Lima26 and Pereira.27
concrete base
S(t ) population of chromosomes in generation t. initialize S(t ) evaluate S(t ) while (not termination-condition) begin select S(t ) from S(t 1) crossover S(t ) mutation S(t ) evaluated S(t ) end
lower bars
additional reinforcement
minimize f (x ) Such that: gi (X) = 0, i = 1, neq Lj < Xj < U j, j = 1, n where Lj and Uj are the lower and upper limits from xj.
Structural Concrete
2007 8 No 1
Table 1 Fabrication costs Material Raw material Concrete: R$/m3 Reinforcement: R$/kg Inll blocks: R$/m3 Factory activity: R$/m3 Administration: R$/m3 123.75 1.80 2.00 Labour 4.40 0.25 2.00 4.40 Equipment 4.00 1.20 1.67
4 Figure 4 Difference between Mel and Map The continuous reinforcement of the lattice-reinforced joist was determined using as the variable the degree of the bending moments redistribution, taking care not to exceed the slabs bearing capacity.28 The degree of redistribution h, expressed as a percentage (Figure 4), is given by Map 100(%) h 1 Mel where Map is the negative bending moment after the redistribution and Mel is the negative bending moment for elastic-linear material.
Table 2 Assembly costs Material Erection: R$/m Cast-in-place concrete: R$/m3 Reinforcement: R$/kg Administration: R$/m3
3
Labour
Equipment
10.4 (24.75 . fck,capa 74.25) 110.75 8.18 20% of the overall costs involved in the assembly, cast-in-place concrete and complementary reinforcement
Solution of the cost optimization problem of lattice-reinforced joists slabs Costs and input for latticereinforced loists
The denition of the cost minimisation function for the precast concrete elements studied here was based on the costs of fabrication, external transport and application. Listed below are the partial values of the inputs in each stage. Note that these values correspond to an initial analysis of the problems evaluated in this work. All prices are given in Brazilian Real (R$). The conversion rate is 1.00USD 2.14BRL and 1.00EUR 2.79BRL (at 16 May 2006). The costs considered here are the same as those adopted by Castilho:29 (a) Fabrication costs. The costs involved in the production of precast concrete elements correspond to the cost of raw material (concrete, reinforcements, inll blocks), factory activities (the cost of this phase corresponds to the activities that take place after moulding but before delivery of the product), and administrative costs
(the overall costs involved in administrative tasks and the salaries of the people involved in the job). A breakdown of these costs is given in Table 1. (b) External transport cost. This is the cost of transporting the joist from the factory to the construction site, comprising labour, trucks, fuel, insurance and maintenance costs. For the purposes of this study, we considered a construction site located 100 km from the factory. External transport costs (R$=m3 ) 52 (c) Assembly costs. The costs involved in the assembly refer to the assembly of the joist, the cast-in-place concrete, the complementary reinforcement and administrative costs, as indicated in Table 2.
es,4 and the TR-12645 and TRby Magalha 16645 available used in Brazilian market, all with the same loading conditions. The lattice-reinforced joist was optimised under the same loading conditions adopted es:4 by Magalha (a) Distributed uniformly along the element, the loads were based on the concretes dead weight (gc 25 kN/m3), expanded polystyrene (EPS) inll blocks (ge 0.12 kN/m3), a permanent load of 0.5 kN/m2 and a live load of 2.0 kN/m2. (b) The slabs were dimensioned considering a concrete base compressive strength of 20 MPa and a CA 60 reinforcement (yield strength equal to 552 MPa). (c) The resistance factor used for concrete is 1.4 and 1.15 for steel. The load factor used in the analysis is 1.4. Figures 5 and 6 depict, respectively, the static scheme and the slab section. The slabs
4m
4m
Structural Concrete
2007 8 No 1
EPS 10
EPS 10
EPS
x1 HEPS He 3
The height of the inll blocks is a function of the slabs total span: (a) for a total span of 8 m (4 m 4 m): HEPS 8 cm; and (b) for a total span of 12 m (6 m 6 m): HEPS 12 cm. The process of designing the slab with lattice-reinforced joists was the same as that es.4 The design criteria described by Magalha were based on the joists service limit state and ultimate limit state.
25
13 EPS
additional reinforcement
3 15
The various costs were added up to nd the function that represents the joists total production cost, considering the stages of execution, transportation and application. Further details regarding the nal expression of the function (equation (1)) are described by Castilho.29 (0 00122 0 127 x1 x3 0 0127 x1 0 0158 x3 0 00395 x3 1 115 (100 x4 ) 0 01)2 3 f (x ) 2 x3 (0 219 25 x3 x1 7 6 4 2 50 x1 2 50 x3 ) 5 (x1 0 065) 100 (12 50 x1 x3 0 000624 1 248 x1 ) x3 120 (2 475 x2 74:25) 0 12 50 x x 0 000624 1 1 3 C B 1 248 x1 C B @ A x3 34 3 x1 14 57 x2 x3 (88 8 x1 5 457) (22 4 x1 1 456) ((100 x4 ) 0 01= 4633 x1 284 713) x1 x3 0 00005 0 1 x1
nal section was completed with EPS inll blocks and a layer of concrete. The cost function was represented by four variables: x1 representing the height of the cast-in-place concrete; x2 representing the concretes compressive strength; x3 representing the distance between joists;
and x4 representing the degree of redistribution. Three joists with heights (He) of 8 cm, 12 cm and 16 cm (TR-08634, TR-12645 and TR-16645) were analysed to nd a solution for the optimization problems of latticereinforced joists loaded as mentioned earlier.
Table 4 Values of variables, reinforced areas and cost function for two spans 2 TR-08634 TR08634 Span 8 m GA (cont_var) x1: m x2: Mpa x3: m x4: % Cost function: R$/m2 Negative reinforcement: 1024 m2 Additional reinforcement/ joists: 1024 m2 0.040 19.3 0.598 40 19.96 1.90 1.19 GA (disc_var) 0.040 20.0 0.600 40 20.58 1.97 1.19 Excel Solver 0.040 19.0 0.600 40 20.34 1.96 1.19 Span 12 m GA (cont_var) 0.094 23.0 0.373 22 39.55 2.76 1.20 GA (disc_var) 0.100 21.0 0.370 38 38.84 2.07 1.23 Excel Solver 0.085 21.8 0.378 40 35.56 2.12 1.56
(1)
where x1 is cast-in-place concrete, in m; x2 is compressive strength of the concrete, in Mpa; x3 is distance between joists axes, in m; and x4 is degree of redistribution, in %. Note that, after several tests, this function was found to incorporate all the problem variables, offering the best representation of the lattice-reinforced joist cost minimisation
Structural Concrete
2007 8 No 1
Table 5 Values of variables, reinforced areas and cost function for two spans TR-12645 TR-12645 Span 8 m GA (cont_var) x1: m x2: Mpa x3: m x4: % Cost function: R$/m2 Negative reinforcement: 1024 m2 Additional reinforcement/ joists: 1024 m2 0.040 18.2 0.599 40 21.82 2.47 0.72 GA (disc_var) 0.040 19.0 0.600 40 22.06 2.59 0.86 Excel Solver 0.040 18.2 0.600 40 21.48 2.48 0.86 Span 12 m GA (cont_var) 0.040 19.5 0.597 40 23.60 2.64 1.86 GA (disc_var) 0.040 20.0 0.580 38 24.60 2.72 1.84 Excel Solver 0.040 20.0 0.600 31 22.85 3.13 1.81
and TR-16645 lattice-reinforced joists with heights (He) of 8 cm, 12 cm, and 16 cm, respectively, for two slab spans: 8 m (4 m 4 m) and 12 m (6 m 6 m). For purposes of reference, these tables also list the values of the variables for the conventional optimization method (Excel Solver), and the values evaluated of the nal areas of the negative (at middle support) and additional reinforcement (one joist) for each case. The reinforcement areas (negative and additional) are not variables of the optimisation method. Transverse reinforcement is not required. Tables 4, 5 and 6 indicate that, for larger spans of the same joist, there is: (a) an increase in the height of the concrete layer (in the case of TR-08634), which was expected, since this increase leads to an increase of the compression ange, thereby enhancing the slabs strength; (b) an increase in the concrete layer strength. Thus, increasing the span requires increasing the slabs strength, which is also achieved by increasing the strength of the cast-in-place concrete; (c) a reduction in the distance between joists, in the case of the TR-08634 joist, which was also expected, although there was no alteration in the distance between the other joists. The increased height of the compression ange resulted in a shorter distance between joists and, hence, a reduction in the area of concrete, which was reected in a reduction of the cost of the cast-in-place concrete; (d) an increase in the negative and additional reinforcements, as expected. A further analysis of the results shown in the above tables indicates that, when the height of the lattice is increased, the results for the same span are inverted. As can be seen: (a) in the case of the TR-08634 and TR-12645 joists with a 12 m span, the height of the concrete layer decreased, and the distance between joists increased. Conversely, as the height of the compression ange increased, the distance between joists decreased to obtain the lowest cost for concrete;
problem. These results are part of an initial study of the negative moments in continuous slabs with lattice-reinforced joists. Therefore, the slab cost optimisation problem boils down to the minimisation problem f (x) (x1, x2, x3, x4), which is subject to service limit state and ultimate limit state restrictions. In addition to these restrictions, the continuous and discrete variables should satisfy the inequalities presented in Table 3.
GA, based on the following characteristics: elitism (1 individual), a population of 200 individuals, representation by real numbers, uniform crossover, competition selection strategy, and stop criterion delimited in 3000 generations. Since the GA is highly sensitive to the initial population, the data describing the results of each experiment represent the average values obtained using ve randomly selected initial populations (average of ve runs). Tables 4, 5 and 6 present the values obtained for each variable and for the cost function, using GA through continuous (cont_var) and discrete (disc_var) variables. The analysis involved TR-08634, TR-12645
Table 6 Values of variables, reinforced areas and cost function for two spans 2 TR-16645 TR-16645 Span 8 m GA (cont_var) x1: m x2: MPa x3: m x4: % Cost function: R$/m2 Negative reinforcement: 1024 m2 Additional reinforcement/ joist: 1024 m2 0.040 16.5 0.599 40 22.66 2.86 0.68 GA (disc_var) 0.040 17.0 0.600 40 22.08 2.95 0.68 Excel Solver 0.040 17.0 0.600 40 22.08 2.95 0.68 Span 12 m GA (cont_var) 0.040 18.5 0.599 40 25.19 3.21 1.47 GA (disc_var) 0.040 19.0 0.600 40 25.47 3.30 1.47 Excel Solver 0.040 18.9 0.600 40 25.32 3.28 1.47
Structural Concrete
2007 8 No 1
40
35 cost: R$/m2
30 span = 8 m span = 12 m 20
25
Most conventional optimisation methods do not work with discrete variables, which is a signicant disadvantage. Another disadvantage involved the initial points of the design variables. In the conventional method, the initial values of the variables must be provided for the initial processing. In our analysis of these problems, the initial values corresponded to those obtained via the GA (continuous variables). Different initial values were adopted to evaluate the sensitivity of the Excel Solver optimisation method. In general, the results (GA and Excel Solver) converged to an optimum. In some cases, the algorithm did not converge or showed no signicant improvement at all using GA with discrete variables.
Conclusions
(b) a reduction in the strength of the concrete layer since, under the same load situation, a lower slab strength was found to be necessary to offset the loads. Figures 7, 8 and 9 illustrate the behaviour of the cost function for each generation, using the GA (cont_var), for the three cases listed in the tables. Note that the curves depict the same tendency: up to the generation of 500 individuals, the values of the cost function remained practically unchanged. The results indicate that most of the values obtained both with the GA, in the case of continuous variables, and with the conventional optimization method were practically the same. It is worth noting that the values obtained by the GA with discrete variables were compatible with those of the Excel Solver, leading us to conclude that the GA was effective in nding the solution for the lattice-reinforced joist cost optimisation problems. This work investigated the use of GAs to nd solutions for the cost optimisation of latticereinforced joists. The problems involved slabs with lattice-reinforced joists for TR-08634, TR12645 and TR-16645 lattices with span lengths of 8 m and 12 m. In addition, an analysis was made of the design variables, some continuous and others discrete. For purposes of comparison, the same problems were evaluated using a conventional method (Excel Solver). The results of the cost function and the variables, in the case of TR-08634, indicate that there was an increase in the height of the concrete layer, leading to an increase of the compression ange and an increment in the slabs strength. The longer span led to increased strength in the concrete layer. This, in turn, led to the need for increasing strength in the slab, which was also obtained by increasing the strength of the cast-in-place concrete. Moreover, the distance between these same joists decreased. Increasing the height of the compression ange led to reduction in the distance between joists and, hence, a smaller area of concrete, resulting in a reduction of the cost of cast-in-place concrete. The other lattices showed no change in the distance between joists. All the lattices analysed showed an increase in the negative and additional reinforcements.
25
24 cost: R$/m2
23
22
21
Structural Concrete
2007 8 No 1
30 29 28 27 cost: R$/m2 26 25 24
14.
15.
span = 8 m span = 12 m
16.
18.
Based on these analyses, we found that most of the values of the variables for the GA (discrete variables) and for the Excel Solver tended toward the same results. Therefore, we concluded that the GA optimisation tool efciently solved the cost optimization problems of the lattice-reinforced joists analysed here.
-moldados tipo nervuras com elementos pre o trelic armac a ada. (In English: Study of negative bending moment of one-way slabs continuous over two spans with lattice-reinforced joists). o Paulo, Sa o Masters thesis, Universidade de Sa Carlos, 2001. 5. Merlin, A. J. Momentos etores negativos nos apoios de lajes formadas por vigotas de concreto protendido. (In English: Study of negative bending moment of one-way slabs continuous over two spans with prestressed concrete joists). Masters thesis, Escola de Engenharia de o Carlos, Universidade de Sa o Paulo, Sa o Sa Carlos, 2002. 6. Holland, J. H. Adaptation in Natural and Articial Systems. University of Michigan Press, Ann Arbor, MI, 1975. 7. Goldberg, D. E. Genetic Algorithms in Search, Optimization and Machine Learning. AddisonWesley, Reading, MA, 1989. 8. Michalewicz, Z. Genetic Algorithms Data Structures Evolution Programs. Springer, Berlin, 1996. 9. Gen, M. and Cheng, R. Genetic Algorithms and Engineering Design. John Wiley & Sons, New York, 1997. 10. Beasley, D. et al. An overview of genetic algorithms: Part 2. Research Topics, University Computing, 1993, 15, No. 4, 170 181. 11. Beasley, D. et al. An overview of genetic algorithms: Part 1. Fundamentals, University Computing, 1993, 15, No. 2, 58 69. 12. Coello, C. C. et al. Optimal design of reinforced concrete beams using genetic algorithm. Expert
19.
20.
21.
References
-moldado: funda1. El Debs, M. K. Concreto pre es. (In English: Precast conmentos e aplicac o crete: fundamentals and applications). Projeto o Carlos, 2000. REENGE, EESC-USP, Sa 2. Castilho, V. C., Nicoletti, M. and El Debs, M. K. An investigation of the use of three selectionbased genetic algorithm families when minimizing the production cost of hollow core slabs. Computer Methods in Applied Mechanics and Engineering, 2005, 194, No. 45 47, 4651 4667. o de Algoritmos 3. Castilho, V. C. et al. Aplicac a ticos na otimizac o estrutural dos elemenGene a -moldado. (In English: tos de concreto pre Genetic Algorithms application on precast concrete elements optimization). In XXIX Jornadas Sudamaricanas de Ingenieria Estructural, Ponta Del Leste, Anais (CD-ROM), Ponta Del Leste, 2000. es, F. L. Estudo dos momentos etores 4. Magalha negativos nos apoios de lajes formadas por
22.
23.
24.
25.
Systems with Applications, 1997, 12, No. 1, 101 108. Koskisto, O. J. and Ellingwood, B. R. Reliabilitybased optimization of plant precast concrete structures. Journal of Structural Engineering, ASCE, 1997, 123, No.3, 298 304. o de algoritmos Lemonge, A. C. C Aplicac a ticos em otimizac o estrutural. (In English: gene a Genetic algorithms applications on structural optimization). PhD Thesis, COPPE, Universidade Federal do Rio de Janeiro, 1999. Rajev, S. and Krishnamoorthy, C. S. Discrete optimization of structures using genetic algorithm. Journal of Structural Engineering, ASCE, 1992, 118, No. 5, 12331250. Koumousis, V. K. and Arsenis, S. J. Genetic algorithm in optimal detailed design of reinforced concrete members. Computer-Aided Civil and Infrastructure Engineering, 1998, 13, 43 52. Prakash, A. et al. Optimum design of reinforced concrete sections. Computers & Structures. 1998, 30, No. 4, 1009 1011. Lasdon, L. S. et al. Design and testing of a generalized reduced gradient code for nonlinear programming. ACM Transactions on Mathematical Software, 1978, 4, No. 1, 34. rio de Fomento. EF-96. InstrucEspanha, Ministe cion para el proyecto y la ejecucion de forjados unidireccionales de hormigon armado o preten rio de Fomento, Espanha, 1997. sado. Ministe cnicas. o Brasileira de Normas Te Associac a ABNT NBR 6118 Projeto de estruturas de concreto Procedimento, Rio de Janeiro, 2003 (Brazilian structural concrete code: Design of Structural Concrete Procedure). Campos, D. T. A. Estudo do processo de fabrica o e execuc o das lajes trelic c a a adas, analisando em termos de custos com a laje macic a. (In English: Costs comparison between latticereinforced joists slabs and solid slabs, considering the fabrication process and assembly). Graduate thesis, Universidade Estadual do Oeste do , Parana , 2002. Parana lise de seguranc Gaspar, R. Ana a estrutural das -fabricadas na fase de construc o. lajes pre a (In English: Safety analysis on precast slabs during construction). REIBRAC - 42 Ibracon, 2000, 1 15. rios de projeto de lajes nervuradas Borges, J. Crite -moldadas. (In English: Design of com vigotas pre precast concrete joists slabs). Masters thesis, cnica da Universidade de Sa o Escola Polite o Paulo, 1997. Paulo, Sa Franca, A. B. and Fusco, P. B. As lajes nervuradas cios. (In English: o de edif na moderna construc a Joists slabs on actual building construction). o Paulo, 2000. AFALA/ABRAPEX, Sa o em trelic Dinaz, H. Lajes com armac a a. (In English: Lattice-reinforced joists slabs). Vieira o Paulo, 1988. Campos, Sa
Structural Concrete
2007 8 No 1
26. Lima, J. C. O. Sistema trelic ado global. (In English: Global lattice-reinforced system). nea (Boletim te cnico), Campinas, Mediterra 1993. -molda27. Pereira, V. Manual de projeto de lajes pre das trelic adas. (In English: Design requirements o dos of lattice-reinforced joist slabs). Associac a o Paulo, Sa o Paulo, fabricantes de lajes de Sa 2000.
lise estrutural de lajes formadas 28. Droppa, Jr A. Ana -moldadas com armac o trelipor vigotas pre a c ada. (In English: Structural analysis of latticereinforced joists slabs). Masters thesis, Escola de o Carlos, Universidade de Sa o Engenharia de Sa o Carlos, 1999. Paulo, Sa o de componentes de 29. Castilho V. C. Otimizac a -moldado protendidos mediante concreto pre Algoritmos. (In English: Optimization of precast
prestressed elements using genetic algorithms). o Paulo, Sa o PhD Thesis, Universidade de Sa Carlos, 2003. 30. Castilho, V. C. et al. Application of Genetic Algorithm for optimization slabs of prestressed concrete joists. In 22nd Iberian Latin-American Congress on Computational Methods in Engineering, Campinas, Anais (CD-ROM), Campinas, 2001.
Structural Concrete
2007 8 No 1