Rapid economic growth, population growth and climate change have led to the growth of problems related to resource scarcity. For this reason, comprehensive global cooperation between nations continues to develop to support low-carbon sustainable development. However, a study in which the construction sector is not included as a polluting sector will render carbon emission reduction strategies insufficient to achieve their goals (Yazdani et al. 2021a). The share of the construction industry in the increase in global carbon emissions is quite high (Yazdani et al. 2021b). This share is increasing day by day in the face of unprecedented demand for construction materials in developed and developing countries. (Ghafourian et al. 2021). Demand for cement is directly related to the condition of the construction sector, and as investments increase, the demand for cement also increases. Production and consumption of construction materials, especially cement, emits enormous amount of carbon dioxide (CO2) (Mirmozaffari et al. 2020). On a global scale, 8% (approximately 2.2 billion tons) of all CO2 emissions caused by humans originates from cement production. 60–80% of the CO2 emissions associated with concrete come from cement production (Ariöz et al.). For this reason, in order to significantly reduce greenhouse gas emissions, it would be beneficial to prefer environmentally friendly production processes using recycled (Adesina and Das 2021; Alyousef et al. 2021; Gao et al. 2022; Valente et al. 2022; Signorini and Nobili 2022) or natural resources (Davraz et al. 2018; Prošek et al. 2020; Quedou et al. 2021; Thapa and Waldmann 2021; Danish et al. 2021) instead of traditional material production in the construction sector, which is known to have harmful effects on the environment (Shahmansouri et al. 2022). Considering the number of studies on reducing the effects of global warming caused by the construction sector, it can be said that the number of studies on the use of environmentally friendly and sustainable materials in this field is limited.
While the calcination process, called as the formation of CaO through the production of CO2 from CaCO3, accounts for about half of the CO2 emissions from cement production, the other half is indirect emissions from the energy used during the production process (Yang et al. 2013; Ávalos-Rendón et al. 2018). Since approximately 4 billion tons of Portland cement is produced every year in the World is the main reason for the fact that Portland cement is the most widely used binder in the construction industry (Pacheco-Torgal et al. 2012). According to the estimations for the next 40 years, it is estimated that this amount will reach 6 billion tons (Samimi et al. 2017). Considering all these consumption amounts, it is expected that reducing the clinker content in the cement composition by using qualified alternative raw materials will help to reduce environmental problems and make the construction sector more sustainable.
To reduce the proportion of Portland cement clinker, it is possible to reduce CO2 content of concrete using supplementary materials with pozzolanic properties such as fly ash (FA) (Deng et al. 2022), silica fume (SF) (Ghavami et al. 2021), ground granulated blast furnace slag (GGBS) (Amani et al. 2021), rice husk ash (Zhang et al. 2020). In the last few years, natural pozzolans have been widely used as a type of supplementary cementitious materials in concrete production (Grist et al. 2015; Tchamdjou et al. 2017b, a; Raggiotti et al. 2018; Robayo-Salazar et al. 2018; Ahmad et al. 2021; Deng et al. 2022; Santana-Carrillo et al. 2022). In the studies, it has been observed that the use of natural pozzolan in concrete improves the properties such as ultimate strength, durability, impermeability and thermal cracking (Rodríguez-Camacho and Uribe-Afif 2002; Shahmansouri et al. 2022). The use of natural pozzolans in concrete generally aims to reduce the production costs and compensate for the increase in the total cost of the building. However, the understanding of sustainability brought about by global climate change has changed this aim recently and reducing CO2 emissions with the use of natural pozzolans in concrete, and as a result, reducing the harmful effects of concrete production on the environment has become a priority. For this reason, it is of great importance to conduct studies on the use of natural pozzolans as partial replacement of cement as a solution to reduce the environmental harmful effects caused by cement production. While studies on producing more environmentally friendly materials by reducing the amount of cement in concrete types with the use of natural pozzolans continue, the effects of natural pozzolans on properties such as mechanical, physical and durability are questioned. Although the effects of natural pozzolans on these properties have not been fully clarified, it is obvious that they have significant effects on the properties of concrete and that they provide improvement in some properties (Michael Thomas 2013). The amount of andesite dust and mud produced during the cutting and polishing of andesite blocks and plates is quite high. If these wastes are not managed and recycled in harmony with the environment, most of them remain as waste materials and cause environmental pollution. The recycling and use of these wastes, which are not classified as hazardous waste, will contribute to the reduction of environmental pollution. Evaluation of these wastes in various sectors will provide significant benefits in terms of economy and prevention of environmental pollution.
In most of the design codes, laboratory investigation of other mechanical properties, especially compressive strength, requires time and cost. Although many studies have focused on developing prediction methods for concrete properties, using these methods for each concrete type does not always give accurate results. The use of estimation methods based on regression forms the basis of existing methods. Promising developments can be made in many fields, including engineering, by using methods such as neural networks to obtain more reliable and more accurate results (Jalal et al. 2020; Behnood and Golafshani 2021; Cao et al. 2021; Ceylan 2021). ANN can provide linear and nonlinear modeling without requiring any prior knowledge between input and output variables (Ali Shahmansouri et al. 2020). ANNs make generalizations from the examples given to them during their training and can generate information about new examples with these generalizations. Although many researchers use these methods extensively to model concrete properties, it would not be correct to talk about the existence of an optimum method for each concrete type. Consequently, efforts to develop more innovative and efficient techniques for predicting material properties will continue to exist in the future of science word.
The aim of this study is to estimate the compressive strength of WAD-replaced cementitious composites and to compare these results with real data obtained in the laboratory. For this purpose, an ANN model was developed to predict compressive strengths with high accuracy. The data obtained through experimental studies were used to develop a model. The compressive strength tests of the produced cement-based composite mixtures with different WAD replacement ratios and different weight component ratios were carried out at the end of 28 and 90-day curing periods. As a result, a database of the compressive strength values obtained from the experimental studies of the samples produced for six different mixtures was created. In addition, parametric and sensitivity analyzes were carried out to ensure the effectiveness of different input variables used to estimate compressive strengths. The findings of this study contribute to the development of an environmentally friendly type of concrete as an alternative to conventional concrete that can help reduce CO2 emissions in the construction industry.
1.1. Previous Studies
Unlike known calculation techniques, ANN is a calculation method that can make decisions by adapting to the environment, even if there is incomplete information, and can achieve successful results and tolerant against errors. Due to this feature, the use of ANNs has gained popularity in the field of civil engineering as well as in other fields in recent years. These methods are also used to predict the properties of concrete and cement (Feng et al. 2020). When the literature is examined, it is seen that ANNs, as a useful method for predicting the compressive strength of concrete and cement-based composite materials, were used in the studies. Marangu (Marangu 2020) conducted a study on the compressive strength prediction of clay-based cement mortars using ANN and Support Vector Machine (SVM). The results showed that the ANN was more successful in the prediction of the compressive strength of the samples obtained by mixing clay-based cement and Portland cement in certain proportions. Sevim et al., (Sevim et al. 2021) determined the compressive strength of the mortars that they produced by adding eight different fly ash to cement at rates of 10, 20, 30 and 40% by weight, at four different curing times. In their prediction model, which they also used ANNs, curing time, fly ash replacement ratios and SiO2 + Al2O3 + Fe2O3 content were used as independent values, while they developed prediction models using compressive strength as dependent variables. When the results were evaluated, it was found that the Genetic Algorithm (GA)-based Adaptive-Network Based Fuzzy Inference Systems (ANFIS) model was more effective in estimating the compressive strength. McElroy et al. (McElroy et al. 2021) used the ANN modeling approach to predict the compressive strength of oil well cement class “H”. 195 cement samples were embedded with three different strength-enhancing nanoparticle solutions at various simulated well temperatures until the experimental time. While the data was used to generate the ANN model, 70% was used to train the model, 15% to validate and 15% to test the model. As a result of the statistical studies, it has been determined that the developed ANN model can predict compressive strengths with high accuracy. Xue et al. (Xue et al. 2021) developed multi-scale models to predict mechanical properties for concrete materials. For this purpose, a numerical method based on ANN was used. An ANN model was developed using the data set of the volume fraction of aggregates, porosity, macroscopic uniaxial tensile and compressive strengths as input variables, and the friction coefficient and cohesion of cement particles as output unknowns. As a result, it has been found that the proposed ANN-based model can effectively predict the coefficient of friction and cohesion of porous cement paste with very good accuracy.