Predicting cement brick performance using Multilayer Perceptron Neural Networks

Abstract

Employing recycled solid waste as alternative construction materials presents considerable opportunities for diminishing waste disposal expenses and preserving natural resources. This study investigates the partial substitution of cement with Egg Shell Powder (ESP), Sea Shell Powder (SSP), and Recycled Mortar Powder (RMP) in cement mortar bricks. The study involved making cement mortar with a 1:2 ratio and testing mixtures with ESP, SSP, and RMP amounts of 5%, 10%, and 15%, respectively. The results show bricks with 10% ESP have better properties: they absorb 1.4% more water, are sound enough, have hardness of 21, and have compressive strength of 29.68 N/mm² after 28 d, and which is closely matched with predicted value 30.12 N/mm² from model. It is better at absorbing water by 7.53% when 5% SSP is added. It sound, has a hardness of 31, and a compressive strength of 28.63 N/mm², and predicted value of 28.56 N/mm². It absorbs 3.61% water, is sound enough, has a hardness of 23.75, and has a compressive strength of 29.62 N/mm² and predicted as 30 N/mm² when 15% RMP is used. A Multilayer Perceptron (MLP) Neural Network was utilized to evaluate brick performance, with predicted values aligning closely with the observed data. Results indicate the application of 10% ESP, 5% SSP, and 15% RMP enhances the characteristics of cement mortar bricks, fostering cost-effectiveness and sustainability in buildings.

Graphical abstract

Keywords

MLP Neural Network / cement mortar bricks / ESP / SSP / RMP

1 Introduction

Sustainable construction materials have been the subject of extensive study in an effort to identify alternatives to cement that are less harmful to the environment without sacrificing performance. This is mostly due to the scarcity of natural resources and the high cost of typical building materials. Bricks are one of the most basic and long-lasting building materials available. Various materials, including clay, shale, concrete, and cement mortar, are used to make these rectangular blocks. Walls, pavements, foundation works, fireplaces and chimneys, and retaining walls are often constructed by laying bricks in courses and using mortar to connect them together [1]. Bricks have been an essential component of building construction for thousands of years because they are strong, long-lasting, and insulating. Bricks also come in a wide range of sizes, shapes, and colors, which allows for more creative freedom in building designs. These days, we use modern machinery and materials to make bricks [2]. These days, high-tech materials and processes like concrete, cement mortar, and burned clay go into making bricks. Their longevity, fire resistance, and eco-friendliness have made them famous. Bricks are still widely used in construction because they are dependable, beautiful, and good for the environment [3]. In cement mortar bricks, cement stands in for Egg Shell Powder (ESP), Sea Shell Powder (SSP), or Recycled Mortar Powder (RMP). Because of its high calcium concentration, it is suitable for manufacturing purposes [4,5]. Compressive strength is a crucial property of a brick, as it determines its ability to withstand a maximum load under compressive forces before it falls or fractures. Bricks are subjected to various loads during their lifetime, such as wind loads, earthquake loads, heat loads, moisture loads, dead loads, and living loads. These loads cause the brick to deforming, which can lead to cracks and ultimately failure if the brick does not have sufficient compressive strength. Balouch et al. [6] investigated, compressive cement mortar bricks can enhance brick strength by adding more steel or fibers. For many years, people have used cement mortar bricks to enhance the strength of compressed bricks. However, the use of cement in bricks can be expensive. Paruthi et al. [7] introduced a technique to overcome this problem, which involves replacing some of the cement with recycled materials. Utilizing waste materials in cement mortar bricks is a type of brick with a different composition of materials to meet the required performance. On the other hand, few studies, Tayeh et al. [8] examined the applications of recycled seashells as a partial substitute for cement and noted their potential to improve sustainability. Ali Said Al Abri et al. [9] illustrated the inclusion of ESP as a cement substitute and find the optimum parentage of that. Further, Olivia and Oktaviani [10] observed that, ground cockle and calm seashells improved workability and strength at lower replacement levels and diminish at higher levels. Sangeetha et al. [11] examined seashell trash as a cement substitute and enhanced the mechanical strength with optimum mix ratio. Moreover, Lakshmi and Nivedhitha [12] found that the partial replacement of aggregates with recycled concrete debris preserved the strength. Rajesh et al. [13] found that, use of ESP in natural fiber reinforced composites, revealing enhanced bonding and mechanical capabilities, thereby establishing it as a potential bio-concrete material. On the other hand, few studies [14,15] focused on machine learning approaches to efficient identity and precise damage detection in concrete structures for infrastructural maintenance. In the current study, we prepare cement mortar bricks by partially replacing cement with 90 microns of waste material. Few studies demonstrate that using waste materials such as ESP, SSP, and recycled aggregates partially in place of cement leads to a significant improvement in compressive strength. For example, we form ESP and SSP by cleaning the raw materials, powdering them using a mixer, and then sieving them through a 90-micron sieve. Compressive Testing Machine hammers the recycled mortar, compresses it, and then sieves it through 90 microns. Researchers carried out several investigations to assess the potential of cement mortar bricks made with ESP, SSP, and RMP. Yerramala [4] investigated compressive strength increased with 5% replacement of ESP at 7 and 28 d. However, because ESP bricks absorb more water, compressive strength fell above 10%. Rajesh et al. [13] said that ESP was used instead of cement in bacterial concrete at weights of 0%, 5%, 10%, and 15%, with a constant 5% jute fiber dosage. The findings suggest that, in order to attain better values in these attributes, ESP replacement should not go above 5%. We found that the surface fracture healing rate in bacterial ESP specimens was 94.7%, which is 14.46% greater than the internal crack healing rate. Nandhini and Karthikeyan [16] were utilized assess if ground ESP was suitable to substitute some of the cement. We substituted 10%, 15%, and 5% of the cement in the mortars with ground ESP, resulting in cement mortars with a 1:3:0.5 ratio. Here, the ratio 1:3:0.5 refers to the mix proportion of cement: fine aggregate: water use in the preparation of mortar. In which, 1 part cement. 3 parts fine aggregate and 0.5 parts of water-cement ratio. We conducted compressive strength tests after 7, 14, and 28 d, and evaluated the fresh properties using the flow table test. Following microstructural investigations, the results indicated that 10% ground ESP offered the best compressive strength. With ash or SSP for cement, which can save a substantial amount of energy and have positive environmental effects. Studies on mechanical and chemical properties, such as tensile, flexural, and compressive strengths, as well as specific gravity and chemical composition, show that seashell ash works best when added to concrete at a rate of 4% to 5% replacement [17]. Tiong et al. [18] reported that ESP increased water absorption, initial surface absorption, and sorptivity overall, although it decreased spread diameter, according to the results. Moreover, compressive strength and foam stability increased to a level of 7.5% replacement. Consequently, it is possible to substitute some of the cement with ESP, up to 7.5%. Gabol et al. [19], Zaid et al. [20], Chong et al. [21] used ESP at weight percentages of 5%, 10%, 15%, and 20%. They evaluated the workability of freshly mixed concrete and compared the compressive strength with regular concrete. We evaluated the crushing value of each mixture after 7, 14, 28, and 63 d of curing. The compressive strength of ESP specimens was comparable to normal concrete after 63 d; however, an increase in ESP content delayed the strength improvement. Mansoor et al. [22] reviewed a study that examined the possible advantages of making concrete with ESP instead of some cement. In ordinary concrete of grade M25, ESP is used in place of typical Portland cement in varying quantities: 10%, 15%, and 20%. This implies that the powder’s replacement is only partial. The concrete’s compressive and flexural strengths had greatly increased, according to the findings. Additionally, 10% of oven-dried ESP is the ideal amount to use as a partial cement substitute, based on the data analysis from this experiment. Paruthi et al. [7] noted that in order to cut personnel costs and experimentation time, using an artificial neural network technique arose. The microstructure of the ESP, which can replace up to 20% of the cement in the mixture, ultimately increases the strength. The data also clearly demonstrates the Artificial Neural Network model’s predictions of ESP concrete’s strength, with an R-value of 0.96. That amply demonstrates that the ESP was used in lieu of cement to enhance concrete’s qualities and lower the proportion of cement used in the mix. The study concluded and discussed the environmentally friendly use of ESP in construction applications. Uygunoğlu [23] looked into the dynamic modulus of elasticity, water absorption, alkali-silica response, apparent porosity, crushing, and flexural value. The results when hydrated mortar waste (HMW) was used instead of 10% to 20% cement, on the other hand, were within acceptable limits when compared to the controls. Ordinary Portland cement cannot be substituted with HMW at a replacement ratio more than about 40% because the 28-d crushing values decrease by 29% when HMW replaces about 40% of the cement. Li et al. [24] added 10% Recycled concrete fine powder (RFP) to the cement mortar; 7.62% more compressive strength was achieved early on. In comparison to the pure cement (162.5 mm), the blended slurry’s flow ability (175.5 mm) rose by 8.0% at a 10% RFP concentration. The RFP content of the blended cement also increased. Zaharie et al. [25], Pachaiappan [26] reported recycling waste has a favorable impact on both the restoration of the environment and the enhancement of financial resources. This study uses brick powder waste, which falls within the category of recyclable trash generated during the construction, renovation, modernization, and demolition of various buildings. This article examines the mechanical properties of several sustainable recipes containing brick powder waste after 3, 7, and 28 d. It is possible using brick powder added to construction material without reducing the number of raw resources utilized or harming the environment. Aghili Lotf et al. [27], Yaswanth et al. [28] used three distinct types of mortar mixtures. In the first group of mixes, recycled concrete powder was substituted for 25%, 50%, and 75% of the regular Portland cement. The second group used normal Portland cement in place of 10% and 15% silica fume, while the third group utilized a combination of 10% and 15% silica fume and 25% recycled concrete powder in their mixes. According to the findings of mechanical and physical tests conducted on both fresh and hardened concrete, adding 10% or 15% silica fume to mortar that already contains 25% recycled concrete powder can achieve comparable or even better mechanical and physical results compared to the reference. Zheng et al. [29] considered three different cement replacement ratios: 10%, 20%, and 30%. Its replacement quantity and average particle size determined how clay-brick powder affected the flow and compressive strength. Average particle size and replacement level were observed to reduce compressive strength with increasing values. According to Ref. [30], bibio-concrete was made using ESP and POFA as two potential materials to partially replace cement. When eggshells are ground into ESP, they release a high concentration of CaO, which can supply the necessary CaH₂ and improve the process of pozzolanic. Ofuyatan et al. [31] used granulated ground blast furnace slag (GGBFS) and ESP to partially substitute cement. The fine aggregates absorbed 24 weight percent of the water, whereas a 1.6% impact value was associated with coarse aggregate. For flow ability, a partial replacement of 10% weight was ideal. Twenty-eight percent of partial replacements had the highest flexural strength of self-compacting concrete for both GGBFS and ESP after 28 d of healing, measuring 3.2 kN/mm². The Multilayer Perceptron (MLP) Neural Network is a machine-learning tool that is used in this study to evaluate the performance of the brick. In past, few studies [32–34] suggested, MLP model predictions were better in comparing to other machine-learning algorithms. The predicted values are found to be very close to the observed data. To manufacture cement bricks, we use ESP, SSP, and RMP in varying percentages as partial replacements for cement. ESP and SSP achieved good strength at 5%, while RMP achieved good strength at 15%.

Unlike prior works that primarily focused on single material substitutions, the present study uniquely combines three different waste derived material viz., ESP, SSP, and RMP to optimize and predict cement mortar brick performance. Hence, the objective of the present study is to find the best amount of ESP, SSP, and RMP that can be mixed into cement mortar bricks while still keeping good qualities like strength, durability, and ability to hold small amounts of water. By identifying these optimal percentages, the research aims to promote cost-effective and sustainable construction practices through waste utilization. Moreover, the application of an MLP Neural Network to forecast key brick properties based on physical, chemical, and mix details offers a new way to evaluate brick quality before construction in eco-friendly building projects. These factors represent notable progress in material design and performance forecasting.

2 Materials and methods

The primary constituents employed in the production of cement mortar bricks are ESP, SSP, RMP, and fine aggregates. The M-Sand that passes through a 4.75 mm filter and is retained on a 75-micron screen is classified as fine aggregate, utilized in cement mortar bricks. Figure 1 illustrates the method of making the efficient and sustainable brick.

Grade 53 cement was chosen as the primary binder in this study. The cement was meticulously cured and subsequently filtered through a 90 µm sieve to eliminate any lumps. To guarantee enough hydration. The tested properties of the cement are presented in Tables 1 and 2.

Eggshell waste was used to produce ESP. We manufacture ESP by collecting eggshell from poultry, noodle stores, and restaurants. Poultry preserves eggshells for several days; therefore, immerse them in water for a period to facilitate the removal of protein before cleaning. If these are standard eggshells, then clean them easily without submerging them in water. Currently Dehydrate until completely dry, then promptly crush by hand, followed by grinding the eggshell in a coffee grinder, food processor, or mortar and pestle. Subsequently, sift through a 90-micron mesh to achieve a uniform particle size comparable to cement and then utilize it (Fig. 2).

Seashells were utilized to produce the SSP gathered from nearby coastal area. Seashells are washed and dried until dry, then initially crushed with a hammer, followed by grinding into powder using a grinder or mixer. The powder is then sieved through a 90-micron mesh to get a uniform particle size comparable to cement, as illustrated in Fig. 3.

To make RMP, old mortar bricks that are about a year old are gathered, quickly crushed, and then put into molds where they are compressed using compression testing equipment. The material is then sieved through a 90-micron sieve to get cement particles that are all the same size, as shown in Fig. 4.

River sand served as the fine aggregate in this investigation. The aggregates were meticulously cleaned to ensure the complete removal of dust and contaminants. The specific gravity is 4.66, absorption is 1.87%, fineness modulus is 2.65, and density is 1565 kg/m³. The substance retained on a 75-micron sieve subsequent to traversing a 4.75mm filter is designated as fine aggregate, utilized in cement mortar bricks.

The methods described in IS 2250 1981 were used to figure out the mix percentage needed to get the necessary ingredients for cement mortar bricks. We used the ESP, SSP, and RMP as partial substitutes for cement in brickwork. The employed ESP ratios were 0%, 5%, 10%, and 15% per volume of cement. We also utilized SSP and RMP at the same percentages. ESP, SSP, and RMP were incorporated into the binder’s mass. The water-to-cement ratio generally specifies it’s mass. Table 3 presents the mixtures and their constituent components. MM represents mortar mix, while MM7.5 refers to the mortar mix with a compressive strength of 7.5 N/mm² for regular bricks. MWM denotes mortar with waste material mix, and MWM 5%, 10%, and 15% mortar containing 5%, 10%, and 15% waste material mix, respectively.

2.1 Mixing and preparation of specimens

The constituents utilized in the production of mortar bricks include cement, water, and fine aggregates. The chosen materials must adhere to the applicable Indian Standards codes [35] to guarantee their quality and appropriateness. Batching of materials involves evaluating and amalgamating the different constituents of mortar mix in precise amounts to attain the requisite quality and strength. The constituents of mortar comprise cement, water, fine aggregates, and recycled materials. The regular MM 7.5 mix requires hand mixing. Initially, dry mixing involves the amalgamation of cement and sand. Next, we gradually add water to the dry mixed mortar to achieve the desired consistency. Prior to casting, the molds measuring 190 mm × 90 mm × 90 mm must be coated with oil to provide the easy extraction of specimens, as illustrated in Fig. 5(a). The traditional technique for casting and compacting a standard mortar brick sample. Initially, specimen molds are uniformly cleaned and oiled, and the whole mold thickness is stratified into three levels, with each layer filled with mortar and compressed with a tamping rod for 25 blows. After all three layers have been cast and compacted, the entire mold is filled with mortar and subjected to vibration using a vibrating machine.

The vibration facilitates the elimination of residual air gaps and guarantees that the mortar is compacted thickly. After vibrating the mortar, a trowel refines the surface to achieve a smooth and uniform finish. After those specimens were permitted to harden for 24 h. The curing of mortar bricks is a procedure that entails sustaining optimal circumstances for the hardening of the mortar, which involves regulating the moisture content and temperature of the concrete during the curing process (Fig. 5(b)). Effective curing is essential for the long-term durability and strength of the mortar, and it can greatly influence the performance of the finished product (Fig. 5(c)). A total number of bricks are created for the purpose of this investigation. We produced 12 bricks by substituting 5%, 10%, and 15% ESP for cement, respectively.

Eight-six bricks were produced utilizing 10%, 15%, and 10% ESP as a substitute for cement. A total of 36 bricks are produced by partially substituting cement with ESP, 36 bricks are created using a partial replacement of cement with SSP, and 36 bricks are manufactured by partially replacing cement with RMP, as previously indicated. A total of 108 bricks were produced in the current investigation. To ascertain the compression of the brick, a loading application is utilized at a central point, following the procedure illustrated in Fig. 5(d). The brick samples were cast and cured for 28 d in water, subsequently dried, and tested using a compressive strength machine using universal testing machine with a capacity of 2000 kN. The testing method adhered to IS: 2250: 1981 standards [35], the compressive strength of bricks must exceed 10 N/mm². Compressive strength was measured at 7, 14, and 28 d in accordance with IS 456-2000 [36]. Specimens prepared for a 28-d compressive strength test have been inspected. The allowable water absorption rate for face bricks ranges from 4.5% to a maximum of 12%. Plaster may struggle to attach effectively to plaster bricks if they exhibit a water absorption rate over 12% in favor of All bricks underwent water absorption testing after 7, 14, 21, and 28 d of curing. We extracted the bricks from the curing tank 2 d before the test and dried them overnight at 100 °C in the oven. We then determined the mass of each specimen. The mass of each specimen was then recorded under W1. We then immersed the samples in trays filled with water. The commencement time was recorded immediately. After a complete day, we extracted the samples from the tank and promptly dried them with a cloth to eliminate any residual free water. We then weighed them again. The measured amount of water that each sample absorbed, called W2, was written down as a percentage of the dry mass gain in the oven (Fig. 6(a)). Water absorption = [(W2 – W1) / W1] × 100, where W1 represents the oven-dried mass of a brick and W2 denotes the mass of the brick after 24 h of curing (Fig. 6(b)). According to IS: 13311(2)-1992, a layer of cement mortar brick undergoes a hardness test. After curing for 7, 14, 21, and 28 d, we clean and check the brick before placing it down. While securely grasping the tool, ensure that the plunger is perpendicular to the surface of the brick. As the hammer strikes the brick, gradually advance the device in that direction. To secure the plunger in the retracted position, apply pressure to the device and depress the button located on the side. Round to the nearest integer, determine the rebound value on the scale, and record it (Fig. 6(c)). Further, refer the quality of specimen derived by rebound hammer [37].

2.2 Multilayer Perceptron model

A MLP Neural Network is a good machine-learning model that can see these nonlinear interactions and make accurate predictions. ESP, SSP, and RMP are input by the input layer of an MLP, a form of Artificial Neural Network. The hidden layers learn complex correlations between input and output. Forecasts brick characteristics. Neurons in every layer use activation functions like linear activation for regression outputs and Rectified Linear Unit (ReLU) for hidden layers to make changes to math. The detailed methodology of MLP model can be seen in the flow chart Fig. 7.

2.2.1 Input and output parameters

The list of input parameters of MLP model shown in Tables 1 and 2. Wherein, the input parameters includes physical and chemical characteristics of cement, ESP, SSP and RMP. Physical characteristics viz., fineness of cement (f), specific gravity (g), initial setting time (t_i), final setting time (t_f), normal consistency (Nc) and compressive strength (Cs). On other hand, chemical characteristics CaO, SiO₂, Al₂O₃, Fe₂O₃, MgO, SO₃, Na₂O. In addition to said characteristic, the properties of mix proportions (M1, M2, M3, M4) for all four groups viz., weight of materials (W), percentage of cement replacement (%C), fine aggregates weight (Wf), and amount of water added in liters (L) are hidden parameters. Further, the output parameters are Compressive strength, water absorption and hardness values for ESP, SSP, and RMP, respectively. Wherein, compressive strength of ESP, SSP, and RMP are represents CsE, CsS, and CsR, respectively. Water absorption of ESP, SSP, and RMP are WE, WS, and WR, respectively. Similarly, hardness test of ESP, SSP, and RMP are HE, HS, and HR, respectively. Overall, the total input parameters are 14 and output parameters are 9. Figure 8 represents the MLP architecture, in which Input layers are 14 nodes, Hidden Layers are 4 nodes and Output layers are 9 nodes. MLP presents a potential configuration of a MCP, in accordance with the mathematical framework outlined below. To train MCP networks, it is necessary to implement a backpropagation algorithm. A MCP is a type of supervised feedforward neural network that comprises a minimum of three layers: an input layer, at least one hidden layer, and an output layer. The hidden layer and the output layer employ a nonlinear activation function. The total input x_j^{k+1} received by a neuron j in layer k+1can be expressed as follows in Eq. (1):

(1)x_j^{k+1}=\sum _i{y}_i^k{\mathit{\text{z}}}_{i,j}^k-{\alpha }_j^{k+1}.

In this equation, y_i represents the state of the ith neurone within the kth layer, while z_ij denotes the weight connecting the ith neurone in layer k to the jth neurone in layer k + 1. θ represents the threshold of the jth neurone located in the k + 1 hidden layer. Additionally, the output of a neurone in any layer, excluding the input layer, can be expressed according to Eq. (2):

(2)y_j^k=\frac{1}{1+e^{-x_j^k}}.

These nine output parameters are CsE, CsS, CsR, WE, WS, WR, HE, HS, and HR. The MLP performance (network MLP 14-4-9) shows that four neurones are the best number for the hidden layer. Throughout the training period, the network demonstrated high Coefficient of determination (R²) values for compressive strength, water absorption, and hardness, with values of 97.22%, 95%, and 98.20%, respectively. Conversely, the compressive strength, water absorption, and hardness exhibit low error values of 0.72, 0.42, and 0.58, respectively. The MLP’s performance is shown in Table 4. It is the sum of the differences in measured and calculated output parameters (CsE, CsS, CsR, WE, WS, WR, HE, HS, and HR) during the training, testing, and validation phases. This performance is indicative of the goodness of fit between experimental measurements and model-calculated outputs. The random function was employed to select samples for these stages. We divided the collected database into three groups: 70% for training, 15% for testing, and 15% for validation.

The current study employed the MLP model to train the data set with variable proportions of sustainable waste materials (ESP, SSP, RMP) at 5%, 10%, and 15%. Further, to compute varied configuration in the data, the model used material type, percentage replacement, and physical/mechanical test characteristics such as water absorption, hardness, and compressive strength. After training, the previously mentioned model can predict new material configuration performance in the same feature space. Even without testing, the MLP can estimate the compressive strength, water absorption, and hardness of new mixes, such as a 7.5% ESP-SSP mix, if the inputs fall within the training data distribution. The model can also be retrained with data from new material kinds or combinations to improve prediction. This adaptability lets researchers and engineers electronically test different compositions before the actual prototype, saving time and resources and helping sustainable building.

3 Results and discussion

3.1 Compressive strength

Tests were done on all mixture ratios that had cement and similar materials like ESP, SSP, RMP, and fine aggregate to see how strong they were when compressed. The test showed that samples with 10% ESP, 5% SSP, and 15% RMP had compressive strengths of 29.68%, 28.63%, and 29.62% after 28 d, which was higher than the design strength of the reference mix. The application of ESP, SSP, and RMP at various ages demonstrated significant improvements in strength. It provides hydration and also serves as a pack. The results of the compressive strength tests align with prior research that utilized sustainable waste materials. The results show that, especially for ESP, a rise in the ESP percentage is linked to a fall in compressive strength, as shown in Fig. 9. The increasing demand for water is the underlying factor contributing to this decline in strength. Compressive strength exhibits a tendency to decrease with an increase in the SSP percentage. A higher RMP percentage typically correlates with an increase in compressive strength. If 10% and 0.08% of the cement in cement mortar bricks is replaced with ESP, the compressive strength goes up by 0.5% and 0.08%, respectively, when compared to standard mortar bricks after 28 d. The compressive strength of bricks made with 15% ESP was 11.39% lower after 28 d than that of bricks made with regular mortar. When 5% of the cement was replaced with powdered seashell, the compressive strength of bricks made with that mortar went up by 0.03% after 28 d of curing compared to standard mortar bricks (Table 5). When compared to bricks made with regular mortar after 28 d, those that had 10% and 15% SSP in them had a 3.01% and 7.63% lower compressive strength. The compressive strength of bricks made with cement mortar that had 5%, 10%, or 15% RMP added instead of cement showed an increase of 0.56%, 0.09%, or 0.02% when compared to standard mortar bricks after 28 d. The predicted values from the MLP model align closely with the experimental data, as illustrated in Fig. 10. The model demonstrated the best fit across all three cases, achieving an average of 96%. The corresponding uncertainties are detailed in Table 6. Compared to ESP and SSP, RMP shows the least error.

The experimental findings demonstrate that partial replacement of cement with ESP, SSP, and RMP has significant implications for sustainable construction materials. From the results of compressive strength, it can be observed that, the brick with 10% ESP replacement achieved the highest compressive strength of 29.68 N/mm² at 28 d, which closely match with the model prediction (30.12 N/mm²). Further, the results shows, the optimum replacement levels are varied among the materials. It is observed that, at 28 d, the highest strengths are seen at 10% ESP, 5% SSP, and 15% RMP. This clearly indicates, fine partial size of ESP, SSP, and RMP contributes as filler effects and given better hydration to gain the strength. This indicates, ESP enhances the pozzolanic activity and contributes to gain strength and showing potential as a viable cement substitute. These findings imply that ESP can be effectively utilized to reduce cement consumption, thereby lowering CO₂ emissions associated with cement production. From overall observations, ESP, SSP, and RMP can be used effectively as partial replacements with the optimum levels to improve the compressive strengths.

3.2 Water absorption

We quantified the water absorption of each sample by expressing the mass increase relative to the oven-dry mass as a percentage. The allowable water absorption rate for face bricks ranges from 4.5% to a maximum of 12%. Plaster may struggle to cling effectively to plaster bricks if their water absorption rate is insufficient, specifically below 12%. There were also connections between the mortar’s pore structure and the curing system, which had to do with how the mortar absorbed water. The pore structure got worse because the cement mortar had more holes in it, which led to more harmful pores appearing during the steam-curing process. A high rate of water absorption may adversely impact the hydration of mortar, leading to inadequate bonding between bricks and mortar. An adequate brick possesses sufficient strength to withstand a drop onto another brick from a height of approximately 1.25 m. Water absorption must not surpass 20% of dry weight during a 24-h soak in cold water for a high-quality brick.

The water absorption of bricks made with cement mortar was measured at 7, 14, and 28 d of age (Table 7). Our research shows that bricks with ESP, SSP, and RMP as partial replacements absorb water at a rate of 1% to 7%, which is within acceptable limits (Fig. 11). According to code, no more than 20% of water should be absorbed. The predicted values from the MLP model align well with the experimental data, as illustrated in Fig. 12. The model provided an optimal fit across all three cases, achieving an average of 92%. The associated uncertainties are detailed in Table 8. Compared to ESP and RMP, SSP shows the least error.

The results revealed, the hydration is improved at lower levels of ESP. Further, SSP exhibits, higher absorption at some replacement levels, in comparison with ESP and RMP due to its porous structure. Additionally, MLP model predictions are in line with experimental data, with R² of 92% for ESP and SSP. However, there is slightly higher error observed in RMP, which indicates the increase variability in the grain size distribution. The findings of water absorption shown, all brick samples with ESP, SSP, and RMP replacement exhibited water absorption within acceptable limits. The bricks with 10% ESP showed 1.35%–7.12% that are well below the permissible range, implying good durable and resistance to moisture penetration. The reduction in harmful pores due to ESP incorporation contributes to better hydration and bonding, ensuring improved service life. The overall observations depict, all three materials help to maintain acceptable limits to ensure adequate durability.

3.3 Hardness test

The hardness tests of cement mortar bricks were conducted at 7, 14, and 28 d of age, with the results summarized and presented in Table 9. According to the results, bricks with some ESP, SSP, or RMP added to them have a moderately hard surface (Fig. 13). The predicted values from the MLP model align well with the experimental data, as illustrated in Fig. 14. The model provided an optimal fit across all three cases, achieving an average of 89%. The associated uncertainties are detailed in Table 10. When compared to SSP and RMP, ESP shows the least amount of error.

The hardness behavior is shown similar trend as seen in compressive strength. Wherein, ESP contributed the better densification at optimum levels, results in stable surface hardness. Moreover, SSP shown increase hardness at certain replacements, due to the presence of CaCO₃ particles. RMP yielded moderate hardness due to the filler effects with compaction. On other hand, the MLP model predictions are closely matched with experimental data with R² ranging from 85% to 98%. However, ESP shown lowest error, indicates more consistent behavior in compared with SSP and RMP. ESP incorporated bricks consistently showed adequate hardness, with values ranging from 21 to 24 at 28 d, confirming their ability to resist surface wear. Further, when compared with SSP and RMP, and ESP yielded lower uncertainty in hardness prediction, highlighting its reliability as a partial replacement. Moreover, it suggests that brick with ESP can be considered for application where surface durability is essential. From the overall observations, the hardness test results indicated, the brick manufacturing with ESP, SSP, an RMP materials, maintains the adequate surface durability.

4 Conclusions

The current study looked at the mechanical properties and durability of cement mortar bricks that contained ESP, SSP, and RMP. These powders can be used in different amounts to replace cement. The compressive strength of cement mortar bricks that had up to 5% waste material powder added to them went up at 7, 14, and 28 d of curing. We used a machine-learning algorithm, the MLP Neural Network, to validate the experimental data and evaluate the bricks’ performance. We have drawn the following conclusions from the current investigation.

1) We observed a gradual decrease in compressive strength when seashell replacement with cement exceeds 5%.

2) Bricks incorporating up to 10% of ESP as a partial replacement serve as a viable substitute. When the addition exceeds 10%, there is a notable decrease of 40% in the compressive strength of ESP from its original strength.

3) ESP can incorporate RMP as a partial replacement, with a maximum limit of 15%. Exceeding 15% results in a notable reduction in compressive strength.

4) At 28 d, we assessed the water absorption of cement mortar bricks and found that the results were acceptable, as they were below 20%.

5) The surface hardness of the cement mortar bricks was adequate, making them suitable for the construction of load-bearing walls.

6) The calculated values from the MLP model closely match the observed data. The findings show that adding 10% ESP, 5% SSP, and 15% RMP together improves the properties of cement mortar bricks, which helps the construction industry be more cost-effective and last longer.

This study contributes to the field by integrating sustainable waste materials with advanced prediction models to enhance the performance and sustainability of cement bricks. The combine use of ESP, SSP, and RMP, along with accurate MLP based forecasting, provides a practical and cost effective frameworks for material selection and brick design in the construction industry. However, this study has certain limitations. Microstructural analyses (such as Scanning Electron Microscopy, X-ray Diffraction, Fourier Transform Infrared Spectroscopy) were not performed which restricts deeper understanding of hydration products and bonding mechanisms. Additionally, only water curing conditions were investigated, and the influence of alternative curing regimes (viz., steam and accelerated curing) remains unexplored.

References

[1] Sathiparan N . Utilization prospects of ESP in sustainable construction material––A review. Construction and Building Materials, 2021, 293: 123465

[2] Jonnala S N , Gogoi D , Devi S , Kumar M , Kumar C . A comprehensive study of building materials and bricks for residential construction. Construction and Building Materials, 2024, 425: 135931

[3] Hasan K , Karim A , Islam M T , Ali M B , Rahman M T , Rahman M M , Yahaya F M . A review of the mechanical properties and durability of concrete containing recycled seashells as a partial cement replacement. Journal of Material Cycles and Waste Management, 2023, 25(6): 3192–3216

[4] Yerramala A . Properties of concrete with ESP as cement replacement. Indian Concrete Journal, 2014, 88(10): 94–105

[5] Shadhar M H , Salih B M M , Khaleel O R , Kadhim Y M , Mohammed M M , Alalwan H A . The influence of eggshell nanoparticles as a partial replacement of cement in concrete. Innovative Infrastructural Solutions, 2024, 9(12): 465

[6] Balouch N , Rashid K , Javed S , Ahmad T . Experimental study on compressive strength of concrete by partial replacement of cement with ESP. Technical Journal of University of Engineering & Technology Taxila, 2017, 22(3): 91–94

[7] Paruthi S , Khan A H , Kumar A , Kumar F , Hasan M A , Magbool H M , Manzar M S . Sustainable cement replacement using waste eggshells: A review on mechanical properties of eggshell concrete and strength prediction using artificial neural network. Case Studies in Construction Materials, 2023, 18: e02160

[8] Tayeh B A , Hasaniyah M W , Zeyad A M , Yusuf M O . Properties of concrete containing recycled seashells as cement partial replacement: A review. Journal of Cleaner Production, 2019, 237: 117723

[9] Ali Said Al Abri S , Rahul Rollakanti C , Kumar Poloju K , Joe A . Experimental study on mechanical properties of concrete by partial replacement of cement with ESP for sustainable construction. Materials Today: Proceedings, 2022, 65: 1660–1665

[10] Olivia M , Oktaviani R . Properties of concrete containing ground waste cockle and clam seashells. Procedia Engineering, 2017, 171: 658–663

[11] Sangeetha P , Shanmugapriya M , Santhosh Saravanan K , Prabhakaran P , Shashankar V . Mechanical properties of concrete with seashell waste as partial replacement of cement and aggregate. Materials Today: Proceedings, 2022, 61: 320–326

[12] Lakshmi S M L S M , Nivedhitha R N R . Effect of partial replacement of aggregates by recycled concrete debris on strength of concrete. Malaysian Journal of Civil Engineering, 2015, 27(2): 250–259

[13] Rajesh A , Hari P S , Sumathi A . Assessment of ESP in natural fiber composite: A sustainable bio-concrete. Biomass Conversion and Biorefinery, 2023, 15(19): 26355–26370

[14] Bandi M R , Pasupuleti L N , Das T , Guchhait S . Deep learning based damage detection of concrete structures. Asian Journal of Civil Engineering, 2024, 25(7): 5197–5204

[15] Bandi M R , Pasupuleti L N , Sah A K , Jyothula H . Crack detection and categorisation on steel surfaces using machine learning techniques. Asian Journal of Civil Engineering, 2025, 26(9): 3901–3914

[16] Nandhini K , Karthikeyan J . Effective utilization of waste ESP in cement mortar. Materials Today: Proceedings, 2022, 61: 428–432

[17] Wan Mohammad W A S B , Othman N H , Ibrahim M H , Rahim M A , Shahidan S , Rahman R A . A review on seashells ash as partial cement replacement. IOP Conference Series. Materials Science and Engineering, 2017, 271(1): 012059

[18] Tiong H Y , Lim S K , Lee Y L , Yew M K , Lim J H . Absorption and strength properties of lightweight foamed concrete with egg shell powder as partial replacement material of cement. IOP Conference Series. Earth and Environmental Science, 2020, 476(1): 012021

[19] Gabol N A , Memon F A , Jawaduddin M , Zardari Z H . Analysis of ESP as a partial replacing material in concrete. International Journal of Modern Research in Engineering & Management, 2019, 2(9): 22–31

[20] Zaid O , Zamir Hashmi S R , El Ouni M H , Martínez-García R , de Prado-Gil J , Yousef S E A S . Experimental and analytical study of ultra-high-performance fiber-reinforced concrete modified with egg shell powder and nano-silica. Journal of Materials Research and Technology, 2023, 24: 7162–7188

[21] Chong B W , Gujar P , Shi X , Suraneni P . Assessment of waste ESP as a limestone alternative in portland cement. Materials and Structures, 2024, 57(10): 219

[22] Mansoor S S , Hama S M , Hamdullah D N . Effectiveness of replacing cement partially with waste brick powder in mortar. Journal of King Saud University. Engineering Sciences, 2024, 36(7): 524–532

[23] Uygunoğlu T . Reuse of hydrated mortar as partial replacement material in cement. Advances in Cement Research, 2011, 23(3): 113–122

[24] Li Z , Bian Y , Zhao J , Wang Y , Yuan Z . Recycled concrete fine powder (RFP) as cement partial replacement: Influences on the physical properties, hydration characteristics, and microstructure of blended cement. Journal of Building Engineering, 2022, 62: 105326

[25] Zaharie A , Tintisan M L , Siomin A C , Manea D L , Pleşa M L . Cement materials obtained by partial replacement of cement powder with brick powder. IOP Conference Series. Materials Science and Engineering, 2022, 1242(1): 012043

[26] Pachaiappan S . Free vibration analysis of FGM beam by numerical analysis. Innovative Infrastructural Solutions, 2024, 9(8): 333

[27] Aghili Lotf M , Ramezanianpour A M , Habibi A . Utilization of recycled concrete powder as cement replacement or filler in Masonry Mortar. Concrete Research, 1970, 11(4): 133–143

[28] Yaswanth K K , Sathish Kumar V , Revathy J , Murali G , Pavithra C . Compressive strength prediction of ternary blended geopolymer concrete using artificial neural networks and support vector regression. Innovative Infrastructure Solutions, 2024, 9(2): 32

[29] Zheng LGe ZYao ZGao Z. Mechanical properties of mortar with recycled clay-brick-powder. In: 11th International Conference of Chinese Transportation Professionals: Towards sustainable transportation systems. Reston: American Society of Civil Engineers, 2011, 3379–3388

[30] Jhatial A A , Goh W I , Mohamad N , Sohu S , Lakhiar M T . Utilization of palm oil fuel ash and ESP as partial cement replacement—A review. Civil Engineering Journal, 2018, 4(8): 1977–1984

[31] Ofuyatan O M , Adeniyi A G , Ijie D , Ighalo J O , Oluwafemi J . Development of high-performance self compacting concrete using ESP and blast furnace slag as partial cement replacement. Construction and Building Materials, 2020, 256: 119403

[32] Fesghandis G S , Pooya A , Kazemi M , Azimi Z N . Comparison of and radial basis function neural networks in predicting the success of new product development. Engineering, Technology & Applied Scientific Research, 2017, 7(1): 1425–1428

[33] Vidak-Vasic M , Pezo L , Gupta V , Chaudhary S , Radojević Z . An artificial neural network-based prediction model for utilization of coal ash in production of fired clay bricks: A review. Science of Sintering, 2021, 53(1): 37–53

[34] Sinkhonde D , Mashava D , Bezabih T , Mirindi D . Predicting the compressive strength of rubberized concrete incorporating brick powder based on MLP and RBF neural networks. Waste Management Bulletin, 2025, 3(1): 219–233

[35] Srinivas V , Saptarshi S . Experimental and numerical studies on ultimate load behaviour of brick masonry. Journal of the Institution of Engineers: Series A, 2016, 97(2): 93–104

[36] Ramamrutham SNarayan R. Design of Reinforced Concrete Structures (conforming to IS 456): Limit State Method and Working Stress Method. New Delhi: Dhanpat Rai Publishing Company, 2011

[37] Usman F , Khalid N H N . Predicting remaining life of lattice steel transmission tower based on condition assessment and corrosion hazard data. In: E3S Web of Conferences. Les Ulis Cedex A: EDP Sciences, 2021, 325: 01011

RIGHTS & PERMISSIONS

Higher Education Press