Predicting the mechanical properties of sustainable concrete containing waste foundry sand using multi-objective ANN approach
Introduction
Nowadays, moving towards a sustainable environment is one of the main goals of the construction industry and numerous attempts have been carried out in this regard. The use of waste materials in concrete mixtures is an efficient approach to mitigate the landfilling of hazardous materials in the environment and to reduce the consumption of raw materials in concrete. Several types of waste and by-product materials have been widely used in concrete industry as partial or full replacement for the main components of the concrete to produce more environmentally friendly concretes. For example, supplementary cementitious materials such as fly ash [1], [2], [3], silica fume [4], [5], blast furnace slag [3], [6], [7], and rice husk ash [8] are the by-product pozzolans containing silica-rich elements. Concretes containing recycled aggregate [9], [10], [11], [12], [13], waste foundry sand [14], ceramic [15], tire rubber [16], [17], and glass sands [9] are other types of sustainable concretes. Moreover, waste steel [18], plastic [19], cotton [20], and carpet [21] fibers have been used in the concrete mixtures to improve the performance of concrete against cracking induced by tensile stress.
Waste foundry sand (WFS) is a waste of the foundry industry where foundry sand consisting of high-quality silica sand with specific size is used in the molding and casting process. Foundry sand is usually recycled and reused several times until depreciation in its properties makes it unsuitable for casting process. About 6 to 10 million tons of foundry sand is annually discarded from foundries in the U.S. The removed WFS has a large amount of black fines and makes a major concern from the waste management viewpoint [22], [23], [24]. The application of WFS as a partial replacement for natural sand in concrete mixtures is relatively a new approach that dates back to about a decade ago and it has been found to be helpful towards the sustainability of the construction industry.
Prior to extensive consumption of a new material in concrete industry, its mechanical properties must be scrutinized. In addition to the properties of the concrete’s conventional constituents, the physical and chemical properties of the WFS can affect the mechanical properties of the concrete, which are influenced by the type of the poured metal, the employed technology, and the casting and finishing processes [23]. Therefore, introducing reliable models to estimate the mechanical properties of concrete containing WFS with considering all potentially effective variables seems to be necessary.
Over the last two decades, many attempts have been carried out to predict the mechanical properties of concretes containing waste materials using artificial neural network (ANN) [21], [25], [26], [27], [28], [29]. Siddique et al. [30] presented two ANN models to predict the compressive strength of self-compacting concrete containing bottom ash as a partial replacement for fine aggregates at different ages. Xu et al. [31] simulated the mechanical properties of recycled aggregate concretes under triaxial load using hybrid genetic algorithm and ANN and demonstrated the model’s reliability in the design of structural members made by recycled aggregate under complex loading conditions. Golafshani et al. [32] introduced an ANN model assisted by grey wolf optimizer algorithm in predicting the compressive strength of concrete made with fly ash and blast furnace slag and showed improved accuracy of the developed model compared to the traditional ANN. Hammoudi et al. [33] revealed that ANN is a powerful tool in estimating the compressive strength of concrete containing recycled aggregate. Sadowski et al. [27] developed an ANN model to predict the compressive strength of environmentally friendly concrete with high volume of waste quartz mineral dust. Gupta et al. [34] used ANN approach to formulate the mechanical properties of concrete with rubberized sand exposed to elevated temperature. Behnood and Golafshani [35] presented a reliable ANN model hybridized with multi-objective grey wolf optimizer to estimate the compressive strength of concrete with silica fume. Kandiri et al. [36] served salp swarm algorithm to discover the structure of the ANN model for predicting the compressive strength of concrete containing ground granulated blast furnace slag. Czarnecki et al. [37] used ANN to estimate the compressive strength of cementitious composite containing ground granulated blast furnace slag based on their experimental results. Golafshani and Behnood [38] compared the performance of ANN with three artificial intelligence-based methods in predicting the elastic modulus of recycled aggregate concrete. Chithra et al. [39] showed the high correlation of ANN in estimating the compressive strength of concrete containing nanosilica and copper slag.
Many researchers have focused on using the ANN technique in predicting the mechanical properties of concrete containing waste materials without considering the complexity of their developed models [31], [33], [39], [40], [41]. In this study, the network error and the complexity were considered simultaneously using a new multi-objective optimization algorithm based on multi-verse optimizer to achieve a set of optimal ANN models. The next sections in this paper are prepared as follows. In section 2, a brief description about the collected data is presented. All necessary explanations about the subjects related to the proposed model are given in section 3. The detailed results and discussions of the developed models are explained in section 4, followed by conclusion in section 5.
Section snippets
Data collection
In this study, an artificial neural network was used as a multi-objective optimization problem to estimate the mechanical properties of concrete containing waste foundry sand (WFS). The mechanical properties of concrete evaluated in this study included compressive strength (CS), splitting tensile strength (TS), modulus of elasticity (ME), and flexural strength (FS). In this regard, a comprehensive database was gathered from literature [22], [23], [42], [43], [44], [45], [46], [47], [48], [49],
Developed soft computing model
In this section, the basic concepts of artificial neural network, multi-verse optimizer algorithm, and its multi-objective version are introduced at first, which are followed by the fundamental notions of the proposed model. Then, the proposed multi-objective neural network hybridized with multi-verse optimizer algorithm is explained in detail.
Results and discussion
The MOANN-MVO algorithm was coded and implemented in MATLAB environment. Before running the proposed algorithm, the adjustment parameters should be specified which were considered as shown Table 1. The first three parameters related to the MOMVO algorithm were chosen based on try and error method, while the rest in this category were selected from original work on MVO algorithm [73]. In terms of the parameters related to the ANN model, partitioning of data into training, validating, and testing
Conclusions
In this paper, a comprehensive study was carried out to estimate the mechanical properties of concretes containing waste foundry sand. In this regard, seven input variables affecting the mechanical properties of concrete were identified and the compressive strength (CS), splitting tensile strength (TS), modulus of elasticity (ME), and flexural strength (FS) were modeled using the artificial neural network (ANN) assisted by the multi-objective multi-verse optimizer algorithm (MOMVO). The network
CRediT authorship contribution statement
Emadaldin Mohammadi Golafshani: Conceptualization, Data curation, Methodology, Supervision, Writing - original draft, Writing - review & editing. Ali Behnood: Conceptualization, Data curation, Methodology, Supervision, Writing - original draft, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (75)
- et al.
Prediction of the compressive strength of normal and high-performance concretes using M5P model tree algorithm
Constr. Build. Mater.
(2017) - et al.
Estimating the optimal mix design of silica fume concrete using biogeography-based programming
Cem. Concr. Compos.
(2019) Utilization of silica fume in concrete: Review of hardened properties
Resour. Conserv. Recycl.
(2011)- et al.
Effects of deicers on the performance of concrete pavements containing air-cooled blast furnace slag and supplementary cementitious materials
Cem. Concr. Compos.
(2018) - et al.
Predicting modulus elasticity of recycled aggregate concrete using M5′ model tree algorithm
Constr. Build. Mater.
(2015) - et al.
Prediction of compressive strength of recycled aggregate concrete using artificial neural networks
Constr. Build. Mater.
(2013) - et al.
Using artificial neural networks for predicting the elastic modulus of recycled aggregate concrete
Constr. Build. Mater.
(2013) - et al.
Properties of recycled aggregate concrete made with recycled aggregates with different amounts of old adhered mortars
Mater. Des.
(2014) - et al.
Evaluation of the splitting tensile strength in plain and steel fiber-reinforced concrete based on the compressive strength
Constr. Build. Mater.
(2015) - et al.
Compressive strength prediction of environmentally friendly concrete using artificial neural networks
Journal of Building Engineering.
(2018)