The most abundantly used material for construction next to water is concrete. The expanding rate of urbanization and industrialization has resulted in over-utilisation of nonrenewable resources such as traditional sand and gravel, resulting in issues of sustainability. It is now imperative to look for alternatives to representative building products. A byproduct of casting industries – Waste Foundry Sand (WFS) which is obtained from both ferrous and non-ferrous yards, is one of those interesting products that can be used as a replacement to traditional sand in construction. Many studies have been carried out over the past few decades to examine the effect of incorporating WFS to traditional sand concrete as a limited and total substitute (de Barros Martins et al., 2019). The growing infrastructural industry needs concrete as a fundamental building material for its development. Consequently, the demand for originally occurring aggregates is increasing everyday due to the higher rate of consumption of natural aggregates as an ingredient part of concrete, which has added to an increase in its commercial value. Such difficulties make it possible for everybody to find a substitute to regular reserves by using some excess products which, if discarded, would otherwise pose an ecological hazard. Increasing population and technological advances results in waste production. As a consequence, numerous researchers around the globe are innovating new ways to minimize such left-overs or as an alternate to use it as a tool with beneficiary value. Since few decades, various wastes from many industries have been extensively studied as an alternate source for fine aggregates. Alternate material replacement in concrete has proved to show progress in both mechanical and durability properties, and this exercise may contribute to feasible growth in construction industry (Gurumoorthy & Arunachalam, 2019). The increasing worry about the over-utilisation of common sand and gravel, the elements of concrete has to be reduced. The extensive practice of concrete as a result of a boom in development and progress has negatively resulted in lack of availability of river sand from the river bed. It has its negative impact and damaging consequences, with depth of river beds being increased, sinking of water table, degradation of bridge structures, significant impacts on streams, estuaries and seaside and marine habitats, due to erosion larger area of land is lost, and a reduction in sediment supply. However, the existence of the casting industry has been harshly affected by the limits on the drawing of sand from the river ensuing in an increase in the worth of sand. It has therefore become authoritative to look for a substitute to the natural fine aggregate (Tittarelli, 2018).

1.1 Foundry Sand

Foundry industries use silica sand of high quality for their method of moulding and casting. These industries produce significant amounts of by-products of which sand contributes for more than 75–80%. When the foundry manufacturers try to utilize this sand after the amount of reprocessing, they are withdrawn from industry and this sand is referred to as spent foundry sand (SFS) (Siddique et al., 2009). Replacement of river sand in concrete by used foundry sand is up to 20 percent by weight. M20 grade concrete has been designed and mechanical, durability and micro structural properties were closely examined. Only splitting tensile strength increased after 20 percent replacement but all other properties were negatively affected (Manoharan et al., 2018). Foundry sand treated with resin has been used to replace sand by 100 percent. Tests on water absorption, carbonation, compression, splitting tensile and flexure were carried out and the results were positive. Workability and durability were good enough (Mavroulidou & Lawrence, 2019). Fresh properties of concrete were determined by replacing fine aggregate with WFS in various amounts (de Matos et al., 2019). Flowability and compressive strength were reduced in concrete manufactured using WFS but calcined WFS showed better results in case of compressive strength as well as in flowability of mortar. CFS replicated the properties of natural sand up to 20 Mpa. It has been attempted to produce concrete with cent percent replacement of natural sand by waste foundry sand.

1.2 Glass Fibres

Recycled aggregates replaced natural aggregates in predetermined fractions and glass fibres were added in increments of 0.25 percent up to 1 percent by volume. Mechanical properties like split tensile strength, tensile and flexural strength escalated gradually with elevated glass fibre content and porosity had been badly affected with increase in recycled aggregate and fibre content (Dehghan et al., 2017). Workability of the concrete containing glass fibres, marble and granite dust has been less compared to concrete containing marble and granite dust. Durability of concrete with glass fibres, marble and granite dust has been high when compared to control concrete (Deshmukh et al., 2012). To enhance the properties of natural fibres like sisal and jute for usage has been treated mechanically with high temperature conditioning. The results displayed that confinement strength of sisal FRP confined cylinders has been nearest to glass FRP confined strength. But confinement modulus of sisal has been greater than all the other FRP confinements and carbon FRP composites showed best mechanical properties. Ultimate axial load of carbon fibre showed more than all other fibres, and ultimate axial load of sisal has been near to glass fibre (Simões et al., 2018). It is discovered that augmenting concrete with glass fiber, not only stepped up the fiber dependent properties, like bending strength, but also improved compressive strength. It has been unveiled that in the existence of glass fiber, compressive strength enhanced about 20% (Dong et al., 2019). Using glass and basalt fiber more than 0.25%, would lead to an almost 5% increase in compressive strength, while the bending strength, tensile strength and fracture energy increased in a noteworthy manner (Katkhuda & Shatarat, 2017). It assessed the effect of disparate percentage of glass fiber on the mechanical properties of high strength concrete and spotted that an enhancement in glass fiber content within 0 – 1.2% correspondingly raised the compressive strength, tensile strength and bending strength by 13%, 63% and 52% respectively (Yugandhar et al., 2017).

Various materials used for the study are listed in Table 1.

Table 1. Various Materials used in the Present Study

2.1 Coarse Aggregate

The nominal size of coarse aggregate is 10 mm to 12.5 mm. Coarse aggregate should enter through 12.5 mm sieve and should be retained in 10 mm sieve. Physical properties of coarse aggregates are tested as per Bureau of Indian Standard (1963a) and the same are listed in Table 2.

Table 2. Physical Properties of Coarse Aggregate

2.2 M-Sand

Crushed or natural normal sand or M-sand is suitable for SCC. Since SCC contains more fines and is well graded, it plays major role in determining fresh properties of SCC. As the field of civil engineering is continuously expanding, the availability of river sand is scarce and hence M-sand is being used in its place. M-sand particle size should be less than 4.75 mm and the physical properties of M-sand should be tested according to Bureau of Indian Standard (1963b, 1970). Table 3 shows the physical properties of Msand and Figure 1 shows the grain size distribution of the same.

Figure 1. Grain Size Distribution of M-Sand

Table 3. Physical Properties of M-sand

2.3 Cement

Cement used in SCC is same as cement used in normal concrete. The present study used Ordinary Portland Cement of 53 grade (Birla Super) as per Bureau of Indian Standard (2003). Testing has been conducted according to the Bureau of Indian Standard (1988). The properties of cement are tabulated in Table 4.

Table 4. Physical Properties of Cement

2.4 Foundry Sand

Foundry sand used in this study has been collected from Bhuwalka Castings and Forging Private Limited, Kolar. They are one of the leading steel manufacturers in South India and Sri Lanka. They also manufacture castings and refractories in this plant at Kolar. The properties of foundry sand are listed in Table 5 and Figure 2 shows the grain size distribution of the same.

Table 5. Physical Properties of Foundry Sand

Figure 2. Grain Size Distribution of M-Sand

2.5 Glass Fibres

Utilizing glass fibres in concrete increases ductility of concrete. As normal concrete has less resistance to the growth of cracks, glass fibres are added to avoid the formation or growth of cracks. Table 6 briefs the physical properties.

Table 6. Physical Properties of Glass Fiber

2.6 Methodology

Figure 3. Bhuwalka Castings and Forging Plant, Kolar

Figure 4. Discarded Foundry Sand from Bhuwalka Castings and Forging Plant, Kolar

• Detailed methodology includes several steps such as material selection, basic test of materials, design mix, preparation of specimens, casting and curing of specimens and testing of specimens.
• Foundry sand has been procured from Bhuwalka Castings and Forging Private Limited, Kolar as shown in Figure 3 and different lengths of alkali resistant glass fibres are used. The discarded foundry sand image is shown in Figure 4.
• Basic material tests were conducted to gauge the physical properties of materials used for the study.
• Emprical equations were developed using multi linear regression analysis (MLRA) to obtain the strength parameter of specimen for any input percentage of foundry sand and glass fiber.
• Artificial Neural Network (ANN) for computing non-linear problems has been implemented to predict the strength properties of concrete.

2.7 Multi Linear Regression Analysis (MLRA)

• Multiple linear regression (MLR) is made to deduce a mathematical interdependence among diverse random variables.
• MLR scrutinizes the affiliation of multiple independent variables on single dependent variable.
• Once each of the individualistic factors has been governed to foresee the dependent variable, the details on the multiple variables can be used to devise an unerring prediction on the level of effect they have on the outcome variable (Huang et al., 2014).

The flow of work is explained by the flowchart in Figure 5. The multiple regression template is based on the following postulations.

Figure 5. Flowchart Depicting the Flow of Work

• There is linear association between the dependent variables and the independent variables.
• The unconstrained variables are not too immensely correlated with each other (Huang et al., 2014).

2.8 Artificial Neural Network (ANN)

ANN is a computing representation whose stacked structure mimics the networked structure of neurons in the brain, with layers of tethered nodes. A neural network can assimilate from data—so it can be tutored to perceive patterns, consign data and forecast succeeding events (Sarbayev et al., 2019).

• ANN is a general tool for designing nonlinear functions, so that it can approximate any convoluted behavior with any desired level of accuracy.
• It is used to perform time-series prediction and modelling (Liu et al., 2016).
• It is comprised of a feed in layer, one or more concealed layers and an output layer. The layers are cascaded via nodes or neurons, with each layer using the outcome of the preceding layer as its input (Pakgohar et al., 2011).

3.1 MLRA

Equations were generated by using Multi Linear Regression Analysis for two different categories. The first one being either substituting Foundry Sand for fine aggregate or inclusion of Glass Fibres in the mix. The second one being combination of Foundry Sand and Glass Fibres in the same mix. Parameters which were significant in impacting the strength properties were considered. Individual strength properties with their significant factors along with equations are listed below.

3.1.1 Compressive Strength of M20 Grade Concrete

Foundry Sand

y_c = 34.1909 + (-19.2517*(w/c)) + (0.0146*CA) + (-0.0185*FA) +(0.0178*FS)

Foundry Sand and Glass Fibres

y_c = 60.7704 + (-59.6646*(w/c)) + (0.0106*CA) + (-0.0138*FA) + (0.0045*FS) + (-1.7061*GF)

3.1.2 Compressive Strength of M25 Grade Concrete

Foundry Sand

y_c = 30.6472 + (8.1585*(w/c)) + (-0.0122*CA) + (0.0191*FA) + (-0.0373*FS)

Foundry Sand and Glass Fibres

y_c = 48.0727 + (-8.3745*(w/c)) + (-0.0192*CA) + (0.017*FA) + (-0.0184*FS) + (4.416*GF)

3.1.3 Compressive Strength of M30 Grade Concrete

Foundry Sand

y_c = 73.2373 + (-55.5179*(w/c)) + (-0.0038*CA) + -(0.0024*FA) + (-0.0059*FS)

Foundry Sand and Glass Fibres

y_c = 71.2857 + (-52.2749*(w/c)) + (-0.0028*FA) + (-0.0272*FS) + (0.5359*GF) (-0.0039*CA) +

3.1.4 Tensile Strength of M20 Grade Concrete

Glass Fibres

y_t = 1.045 + (-5.14*(w/c)) + (0.546*GF) + (0.277*Length of fibre)

Foundry Sand and Glass Fibres

y_t = 0.855 + (1.1563*(w/c)) + (0.0009*FA) + (0.0002*FS) + (1.0309*GF)

3.1.5 Tensile Strength of M25 Grade Concrete

Glass Fibres

y_t = 5.5694 + (-3.5324*(w/c)) + (-0.3652*GF) + (-0.0111*Length of fibres)

Foundry Sand and Glass Fibres

y_t = 5.5726 + (-3.3832*(w/c)) + (-0.0003*FA) + (-0.0014*FS) +(-0.3531*GF)

3.1.6 Tensile Strength of M30 Grade Concrete

Glass Fibres

y_t = 5.5726 + (-3.3832*(w/c)) + (-0.0003*FA) + (-0.0014*FS) +(-0.3531*GF)

Foundry Sand and Glass Fibres

y_t = 7.1227 + (-6.2298*(w/c)) + (0.0003*FA) + (-0.003*FS) + (0.352*GF)

3.1.7 Flexural Strength of M20 Grade Concrete

Glass Fibres

y_f = (8.9914 + (-11.881*(w/c)) + (1.7877*GF) + (0.0756*Length of fibre)

Foundry Sand and Glass Fibres

y_f = 9.492 + (-11.2905*(w/c)) + (0.0007*FA) + (-0.0088*FS) + (1.9843*GF)

3.1.8 Flexural Strength of M25 Grade Concrete

Glass Fibres

y_f = 4.2029 + (1.681*(w/c)) + (0.0141*GF) + (0.0131*Length of fibre)

Foundry Sand and Glass Fibres

y_f = 4.1042 + (1.6565*(w/c)) + (0.0001*FA) + (-0.002*FS) + (0.0105*GF)

3.1.9 Flexural Strength of M30 Grade Concrete

Glass Fibres

y_f = 9.9218 + (-13.3734*(w/c) + (3.3103*GF) + (-0.025*Length of fibre)

Foundry Sand and Glass Fibres

y_f = 6.9398 + (-10.3475*(w/c)) + (0.0026*FA) + (0.0054*FS) + (3.1925*GF)

where,

(w/c) – water to cement ratio

GF – Glass Fibres in % of volume of concrete

FA – Fine aggregate in kg

CA – Coarse aggregate in kg

FS – Foundry Sand in % weight of FA

y_c – Compressive strength

y_t – Tensile strength

y_t – Flexural strength

3.2 Combined Effect of Foundry Sand and Glass Fibres using ANN

Using ANN, strength property values were predicted with higher values of coefficient of regression. This meant that the strength properties were highly dependent on the selected parameters.

3.2.1 Compressive Strength

The regression plots for combined effect of Foundry Sand and Glass Fibres on strength properties respectively are represented in Figure 6(a), 6(b) and 6(c).

Figure 6. Combined Effect of Compressive Strength (a) M20 Grade Concrete, Regression Plot (R=0.92172) (b) M25 Grade Concrete, Regression Plot (R=0.96303) (c) M30 Grade Concrete, Regression Plot (R=0.9747)

3.2.2 Tensile Strength

The regression plots for tensile strength of Foundry Sand and Glass Fibres and individual effects of Foundry Sand and Glass Fibres on strength properties respectively are represented in Figure 7(a), 7(b) and 7(c).

Figure 7. Combined Effect of Tensile Strength (a) M20 Grade Concrete, Regression Plot (R=0.89518) (b) M25 Grade Concrete, Regression Plot (R=0.90999) (c) M30 Grade Concrete, Regression Plot (R=0.9554)

3.2.3 Flexural Strength

The regression plots for flexural strength of Foundry Sand and Glass Fibres and individual effects of Foundry Sand and Glass Fibres on strength properties respectively are represented in Figure 8(a), 8(b) and 8(c).

Figure 8. Combined Effect of Flexural Strength (a) M20 Grade Concrete, Regression Plot (R=0.95086) (b) M25 Grade Concrete, Regression Plot (R=0.93769) (c) M30 Grade Concrete, Regression Plot (R=0.96624)

3.3 Comparison of Results: Combined Effect of Foundry Sand and Glass Fibres

Table 7 shows the results of the compressive strength on the combined effect of Foundry Sand and Glass Fibres.

Table 7. Combined Effect of Foundry Sand and Glass Fibres: Compressive Strength

Table 8 shows the results of the tensile strength on the combined effect of Foundry Sand and Glass Fibres.

Table 8. Combined Effect of Foundry Sand and Glass Fibres: Tensile Strength

Table 9 shows the results of the flexural strength on the combined effect of Foundry Sand and Glass Fibres.

Table 9. Combined Effect of Foundry Sand and Glass Fibres: Flexural Strength

3.4 Individual Effects of Glass Fibres

3.4.1 Compressive Strength

The regression plots for individual effect of Foundry Sand on compressive strength properties respectively for different grade of cement are represented in Figure 9(a), 9(b) and 9(c).

Figure 9. Individual Effect of Glass Fibres on Compressive Strength (a) M20 Grade Concrete, Regression Plot (R=0.97669) (b) M25 Grade Concrete, Regression Plot (R=0.89714) (c) M30 Grade Concrete, Regression Plot (R=0.97389)

3.4.2 Tensile Strength

The regression plots for individual effect of Glass Fibres on tensile strength properties respectively for different grade of cement are represented in Figure 10(a), 10(b) and 10(c).

Figure 10. Individual Effect of Glass Fibres on Tensile Strength (a) M20 Grade Concrete, Regression Plot (R=0.93955) (b) M25 Grade Concrete, Regression Plot (R=0.93185) (c) M30 Grade Concrete, Regression Plot (R=0.94989)

3.4.3 Flexural Strength

The regression plots for individual effect of Glass Fibres on flexural strength properties respectively for different grade of cement are represented in Figure 11(a), 11(b) and 11(c).

Figure 11. Individual Effect of Glass Fibres on Flexural Strength (a) M20 Grade Concrete, Regression Plot (R=0.92225) (b) M25 Grade Concrete, Regression Plot (R=0.9195) (c) M30 Grade Concrete, Regression Plot (R=0.94859)

3.5 Comparison of Individual Results

Table 10 shows the comparison of the compressive strength on the individual effect of Foundry Sand.

Table 10. Comparison of the Comparative Strength of Foundry Sand

Table 11 shows the comparison of tensile strength on the individual effect of Glass Fibres.

Table 11. Comparison of the Tensile Strength of Glass Fibres

Table 12 shows the comparison of flexural strength on the individual effect of Glass Fibres.

Table 12. Comparison of the Flexural Strength of Glass Fibres

From literature survey as well as from MLRA and ANN, the maximum strength gain in all strength properties considered in this research study has been not more than 30% filling in of fine aggregate by foundry sand and adding 1% of glass fibres of total volume of concrete in the mix. The positive impact of these materials can be due to filler effect of the rounder, uniform and finer FS reducing the void ratios. Higher percentages of glass fibres induce balling effect and filler effect is lost in higher foundry sand percentage replacements.

• This study evaluates the consequences of foundry sand on concrete when it is being replaced with percentage of weight of fine aggregate and enhancement of tensile and flexural strength with the influence of glass fibres.
• Multi Linear Regression Analysis (MLRA) with the application of MATLAB is used for conducting linear analysis and Artificial Neural Network (ANN).
• Mathematical models generated from MLRA gives the strength parameters of concrete in terms of significant parameters while ANN predicts strength properties.
• It is inferred that strength properties of concrete mix increase with substituting waste foundry sand by M sand up to 30% replacement and decrease if the replacement increases more than 30%.
• The use of mathematical and prediction models for analysis avoids the tediousness, cost and time spent on performing experimental investigations and can be used as an effective alternative when experimental works cannot be conducted.
• Analytical results of concrete of foundry sand and glass fibres completely corresponds with experimental results obtained from literature review. When analytical results of this concrete were compared to ordinary concrete, significant improvement has been observed in compressive, tensile and flexural strength.