1. Introduction
Optimization is the art of finding the best-suited candidate from a set of available options, without explicitly enumerating and evaluating all the possible alternatives [
1]. The use of the optimization concept in concrete structures is for obtaining an economical cross-section of a reinforced concrete (RC) member that has sufficient capacity to counter the applied loading. The structural design engineers typically follow the iterative trial and error method to reach a feasible design option according to the applied loads. All this consumes a significant amount of time due to complicated calculations and does not always result in the most optimal and economical RC section [
2]. For the design of RC members, the cost is not dependent on a single material, as is the case for steel construction; rather it, depends on the volumes of both concrete (cross-sectional dimensions) and reinforcing steel, making the optimization problem complex in nature.
Different optimization methods have been used to solve different design problems in structural engineering. Some of these include the Generalized Reduced Gradient (GRG) [
1] technique, Genetic Algorithms (GA) [
3,
4,
5], Simulated Annealing (SA) [
6], Sequential Quadratic Programming (SQP), Non-Linear Programming (NLP) [
7], Harmony Search (HS) [
1], and Particle Swarm Optimization (PSO) [
8]. The optimization studies conducted using these methods have shown their effectiveness; however, their application in practical design is limited [
3,
9,
10]. This is mainly because these techniques require the design engineer to possess in-depth knowledge of the optimization algorithms, the formulation of the optimization problem, and the coding of the design problems using programming languages. Hence, for the economical design of RC members, not only a proper understanding of optimization is necessary but also the right tool is a need of the hour due to the complex and lengthy nature of the problem.
The RC beam is one of the most common structural members that a design practitioner has to design, and it is encountered in the design of a wide range of structures, from a single-story masonry building to an RC skyscraper. Under ideal considerations, the whole structure should be considered in the optimization process, and its capital, operational, and maintenance costs must be taken into consideration. However, in most of the designs, this approach is not of practical use due to its complicated nature. Therefore, the optimization of individual structural members is generally preferred and adopted.
A significant amount of research has been published on the optimization of the design of an RC beam using different optimization approaches and algorithms. A state-of-the-art review on the optimal design of RC beams is presented by Rahmanian et al. [
9] summarizing the design variables considered and the methods used by different researchers for the optimization of RC beams. The study also presents a spreadsheet implementation for the cost optimization of RC beams and performs a sensitivity analysis to investigate the effect of different design parameters such as the cost and strength of steel and concrete, the diameter of reinforcing bars, and the moment demand on the overall cost of RC beam. The study recommends an exhaustive enumeration method using the VBA code and finds that the higher strength of steel and the larger diameter of the steel bar greatly reduce the overall cost of an RC beam subjected to higher moment demand. Nigdeli et al. [
11] used the Random Search Technique (RST) to find optimum cross-sectional dimensions and reinforcements of continuous RC beams. The internal forces of the RC beam were solved by using the three-moment equation for all iterations of RST. The design constraints given in ACI-318 [
12] were satisfied, and the detailing of reinforcement bars was done in such a way as to promote maximum adherence. The presented approach was demonstrated with two numerical problems. The optimum results were obtained for different cross-section dimension ranges. Compared to the practice followed by designers, the presented approach was approximately 9% more economical for the two-span example. Thomas and Arulraj [
13] worked on the optimization of RC beams using the Sequential Quadratic Programming (SQP) algorithm. They took steel and concrete strengths as design variables and found the efficiency of the SQP algorithm optimization to be good. Hisham Ajmal [
14] studied the cost optimization of the singly reinforced RC beam designed as per ACI 318-08. The Genetic Algorithm (GA) optimization technique programmed in MATLAB was adopted. The different spans and imposed loads were considered, and the performance of the genetic algorithm was found to be satisfactory. Raouache et al. [
15] worked on the optimization of three parameters of the RC beam, which are the strength of concrete, the spacing of stirrups, and the inclination angle of stirrups by using the Response Surface Methodology (RSM). They used the concept of the influence of parameters on the strength of the RC beam and found that the inclination of stirrups has more effect on the strength of the beam, followed by the spacing of stirrups and the strength of concrete. Correia et al. [
16] investigated the effect of cross-sectional dimensions on the overall cost of the RC beam. They implemented the Evolutionary Algorithm (EA) in the Solver tool in MS Excel to optimize the beam design. Beams were designed according to the Brazilian standard [
16]. They found the possibility of a 35% reduction in the optimized cost of the beam. Ozimboski et al. [
17] optimized the cost of simply supported beams designed as per the Brazilian code using the Simulated Annealing (SA) optimization algorithm. They considered two different load levels and spans ranging from 1 to 25 m. It was observed that the difference in the optimal cost of RC beams regarding the minimum and maximum loads is lower than 15% and that the optimized dimensioning had the displacements limitation as to the active constraint for all spans and loads. Chutani and Singh [
18] optimized the design of the RC beam using the Constriction Factor Particle Swarm Optimization (CFPSO) technique. Beam depth and percentage of reinforcement were considered as variables, and the optimization algorithm was coded in C++, which resulted in a 20% reduction in the cost of the beam.
The studies mentioned above have demonstrated the effectiveness of different optimization techniques to obtain an economical design for RC beams. However, the use of these optimization techniques is not so common in RC design because of the difficulty level associated with these techniques. There is a need for a simple tool that can be used to obtain optimal solutions without having in-depth knowledge of complex optimization algorithms, and the ability to code optimization problems in programming languages. In addition, there is a lack of a research study investigating the effect of the depth of the beam and the reinforcement ratio on the cost of the beam for different values of important parameters such as the commercially available grades of steel, concrete strength, and the required moment capacity for the beam. Hence, in the current study, a simple and user-friendly tool is developed in the familiar MS Excel, in which input parameters from a typical design as per ACI 318-19 [
12] can be entered to obtain the most economical design. Further, as a novel objective, the developed tool is used to perform a parametric study to investigate the effect of the parameters such as the grades of steel and concrete, and the required moment capacity on the cost of the beam. The developed tool will be helpful in promoting design optimization among practising engineers, and the results of the parametric study will help them select such material grades that can further reduce the cost of the beam.
The paper first presents the methodology followed for the optimization of a beam designed as per ACI 318-19 using EA and its spreadsheet implementation. The results of three different design examples are presented next to prove the effectiveness of the implementation. Lastly, the results of a detailed parametric study conducted using the developed spreadsheet tool are presented.
2. Methodology for Spreadsheet Optimization
In this study, a spreadsheet-based tool is developed for the optimal design of RC beams. The spreadsheet-based optimization is comparatively easier to use for civil engineers as compared to script-based optimization using a programming language. The spreadsheet in this study follows the design procedure as per ACI 318-19. Initially, the objective function for the optimal design of a simply supported rectangular beam was formulated as discussed in the subsequent section. Afterwards, the design variables were set for this study. This also included the strength and serviceability constraints for those variables as per ACI-318 provisions. Once the formulation of the problem was done, an algorithm for economical design was prepared in MS Excel. All the possible design trials/iterations are run in the Excel Sheet using a built-in add-in, the “Solver” tool. The Evolutionary Algorithm (EA), which is a better-suited technique for beam design optimization-type non-smooth and non-convex problems [
16], is used to achieve the most economical solution while fulfilling all the design requirements.
2.1. Objective Function
The cost-based function provides the best results for concrete [
19]. Four factors
(cost of concrete),
(volume of concrete),
(cost of steel) and
(weight of steel)—have been considered in the cost objective function. The material costs (in PKR) are taken from local contractors as PKR 9167 per cubic meter (cu.m) for 20 MPa concrete, and PKR 120 and 135 per kg for reinforcing bars of 275 MPa and 414 MPa, respectively. The cost function
is defined in Equation (1).
It is imperative to note that the volume of concrete () has been calculated by the product of the width, depth, and span of the beam. While computing this volume, an adjustment for the steel reinforcement was made by subtracting the reinforcement volume, i.e., , where represents the area of steel in the beam and L denotes the beam span. A concrete cover of 40 mm was used in the volume computations.
2.2. Design Variables
In total, there are three design variables, namely, the beam depth, width, and steel reinforcement. All these variables are provided with their upper and lower bounds to avoid infeasible sections. The bounds are collected through literature reviews and through the ACI 318 provisions for the minimum allowable dimensions. For instance, the lower bound for width (b), i.e., 228 mm, is based on local practices adopted in Pakistan, and the lower bound for beam depth (d) is in accordance with the ACI-318 provisions. However, the upper bounds for these variables are user-defined. The maximum and minimum steel reinforcement ratios ( and ) are as per ACI-318.
2.3. Constraints
Constraints are the specified conditions for parameters involved that must be obeyed to obtain values pertaining to structural requirements. They are defined by codes and ensure that the structure is within strength and serviceability limit states. For the optimal design of RC beams, the following constraints for design variables have been considered:
- -
The minimum permissible depth;
- -
The deflection control;
- -
The maximum and minimum area of steel;
- -
The failure type (under-reinforced section);
- -
The architectural constraints;
- -
The flexural capacity of the designed beam.
In order to optimize the design of an RC beam, several trials are to be run once the algorithm is ready. The design of the beam is selected based on the trial that provides the smallest value for the objective function. All the design variables are kept within some permissible upper and lower limiting values. Some such constraints are defined by the code provisions, e.g., the minimum and maximum area of steel, while some are defined by the user, e.g., the upper constraint for the depth of the beam, and the strengths of concrete and steel. The above-mentioned design constraints are further elaborated on in the sections below.
2.3.1. Minimum Permissible Depth
The effective depth (
) of a beam is an important parameter in order to keep the deflections within the required limits. The minimum permissible depth of a beam depends on the required flexural strength of the beam and the allowable deflections in the beam. A generalized relation for the flexural requirement of a beam is expressed in Equation (2) [
12].
where
is the Strength Reduction Factor for tension-controlled sections and
represents the Whitney’s Stress Block Parameter. Generally,
is taken as 0.9, while the value of
is affected by 28 days compressive strength of concrete as:
| |
| |
| |
| |
| |
For instance, as a part of this study, the minimum effective depth of a singly reinforced simply supported beam (
) for
is expressed in Equation (3).
Here, is the required factored moment, is the width of the beams, and is the 28-days compressive strength of concrete. It must be kept in mind that a concrete cover of 40 mm (as suggested by ACI 318) has been considered in the design exercise. The total volume of the concrete material is calculated for the overall depth () of the beam.
2.3.2. ACI Minimum Reinforcement
Lesser steel may lead to a brittle failure of the member without providing sufficient warning before the failure. This type of failure is undesirable. To ensure against this type of failure, a lower limit must be put in place by equating the cracking moment. ACI-318 recommends the minimum value of the steel reinforcement ratio
for the beam as given in Equation (4).
In Equation (4), is the yield strength of reinforcing steel and is the compressive strength of concrete.
Following the basic steel ratio definition, i.e.,
, the minimum area of steel (
) is as given in Equation (5). The second expression in Equation (5) governs when
where
is the width of the beam web and
is the depth of the beam.
For optimizing the quantity of reinforcement in the beam, the design constraint of the steel area (
) as given in Equation (6) was considered with respect to a minimum steel area (
).
2.3.3. Maximum Steel Reinforcement ()
To make sure the flexural member fails due to the yielding of the tension steel and provides sufficient warning before failure, ACI 9.3.1 requires that if the axial load on a non-prestressed member is less than
(
being the gross area of the structural element), then strain in steel is larger than 0.005, and it can be ensured when the maximum reinforcement ratio
is as given in Equation (7)
where
is a parameter that represents the percentage of distance to the neutral axis when stress-strain block for concrete is converted into a simple rectangular geometric shape (Whitney’s Stress Block) for the simplicity of analysis. Following the basic steel ratio definition, i.e.,
, the maximum area of steel (
) is obtained using Equation (8).
For optimizing the quantity of the reinforcement in the beam, the design constraint of steel area (
) as given in Equation (9) was considered with respect to a maximum steel area (
).
2.3.4. Architectural Constraints
Architectural constraints may or may not be ensured by looking at the scenario. Sometimes the architectural restrictions will not allow for the changing of one dimension, e.g., the width of the beam (
) is taken equal to the width of support, and sometimes the upper range of the beam depth (
) is already known. As an example, if the permissible width of the beam is at least 228 mm then the design constraint of Equation (10) is applied when the iterative process is run by the algorithm.
2.3.5. Flexural Constraint
The flexural check as given in Equation (11) is applied to check the capacity of the optimally designed section. The capacity must be greater than or equal to the applied moment.
In Equation (11), is the strength reduction factor, is the nominal capacity of the optimally designed section of the beam, and is the required flexural capacity of the beam.
No shear reinforcement constraint was considered in the optimization process because shear ties in the beam are mostly governed by the minimum requirements. However, when there is significant shear forces in the beam, then the shear design will not be carried out with minimum requirements. Additionally, the cost of shear stirrups depends on the cross-section of the beam, and the increase in the cross-sectional dimensions of the beam would affect the overall cost owing to an increase in the steel for stirrups. However, in this study, considering the dominant stresses to be the flexural stresses, no shear design constraints have been considered in the design optimization and could be considered for future study. A summary of the design constraints used in the spreadsheet is given in
Table 1.
2.4. Optimization Techniques
An Evolutionary Algorithm (EA) has been used in this study. Solver, a built-in MS Excel optimization tool, is used in the spreadsheet. Solver is easily available as an add-in tool for the familiar MS Excel. To ease its application, pre-set settings have been utilized and Macros have been applied to create a simple interface. The settings used for the heuristic technique of EA in the spreadsheet include a population size of 100, a cross-over rate of 0.5, and a mutation rate of 0.1. Due to a “Random Seed” providing different results for each trial, a strategy was developed to apply the Generalized Reduced Gradient (GRG) before running the Evolutionary Algorithm (EA), to bring the initial solution closer to the global optimal.
2.5. Spreadsheet Implementation
The optimization is applied on reinforced concrete beams designed in prominent literature using the objective function, variables, and constraints presented in the above sections. The parameters utilized and the optimization sheet are shown in
Figure 1, in which the simplicity of the developed sheet’s interface is evident.