Strength prediction of hybrid fiber reinforced self-compacting concrete using optimal recurrent neural network

ABSTRACT

Self-Compacting Concrete (SCC) is a special type of concrete which does not require external mechanical compaction and it compacts by its own self weight. Due to its high flow ability and high resistance property, it is being used in congested reinforcement areas and high rise structures. Inclusion of fiber in SCC is an effective solution to manage the heavy load conditions. SCC with fiber addition results in enhancing tensile and flexural strength properties. This present research aims to predict the strength behaviour of SCC using hybrid fibers. Furthermore, the automatic validation is achieved by Optimal Recurrent Neural Network (ORNN) technique which has been developed to analyze the compressive strength, tensile strength and flexural strength. The ORNN is a combination of Recurrent Neural Network (RNN) and Dingo Optimizer (DO). Moreover, in the aspect of durability behaviour SCC with hybrid fiber concrete is made and heated at a temperature of 210°C, 320°C, 530°C and 790°C after their curing period as per ISO 834 guidelines. Test results proved that when the concrete is subjected to high temperature, there was decrease in compressive strength properties. Moreover, the present research findings are implemented in MATLAB for computing the strength properties and the results are discussed elaborately. In order to validate the proposed method, 80% of data is used for training and remaining 20% of data is utilized for testing. This research study is highly valuable towards Civil and Structural Engineering, particularly in the design and analysis of fire-resistant infrastructure, high-performance concrete elements, and intelligent construction materials.

Keywords:
Self-compacting Concrete (SCC); hybrid fibers; recurrent neural network; dingo optimizer; MATLAB

1. INTRODUCTION

Self-compacting concrete or self- consolidating concrete is characterized by considerable fluid and thick properties, which can be put into shape without external compaction. The absence of vibration in the SCC provides high compressive strength due to the high aggregate adhesive interface against conventional vibrated concrete [¹]. However, the larger the glue size in the SCC, the higher the lower coefficient and the greater the risk of breakage. The presence of microcracks formed during the common way to solidify the substrate makes the mixture powerless. When exposed to stockings, these microcracks combine to form more brittle planes that become cement short [²]. In addition, miniature gaps can lead to enlarged pores and reduce the stability of the mixture. The idea of a fiber-matrix mix is not new and is gaining popularity in the current development area. Straw was used by the Egyptians to make mud blocks 5000 years ago. Reinforced cement with fiber was used by Porter in 1910 [³, ⁴]. It has been noticed that, the reduced dry compression properties are based on crossing and capturing the expanding gap by expansion of the fiber. The pressure strain relationship of the solidified substrates showed the best results when the fibers are joined. In addition, the flexibility expands extensively with the presence of fibers [⁵, ⁶]. Furthermore, cellulose or ordinary fibers have evolved in the field of fiber-structured composites due to their over-the-counter benefits as opposed to the embossed partner. Low cost, low thickness, apparent resistance, durability, mechanical legal relationships and ordinary fibers are better respected than airlines, car businesses and structural design endeavours [⁷, ⁸, ⁹]. Hybrid fiber reinforced self-compacting concrete exhibits flexible behavior with significant residual strength limitation over significant amounts of large stress strains. Fiber determination [¹⁰], structure selection, fiber dispersion, fiber orientation, interface strength, porosity and assembling process are the primary factors affecting the mechanical properties of the composite [¹¹]. Delayed examination has produced frantic growth with a hybrid of normal fibers for support. Effective results have been obtained when ordinary fibers are mixed with polymer fibers [¹²]. Significant improvement in the mechanical properties of the structure is observed with the expansion of the fibers [¹³]. The interchangeable polymer fibers provide a better uniform dispersion over a substantial network against the conventional partner. In addition, they can significantly extend the stability of the new condition by restricting drainage and isolation [¹⁴]. Improvement of ductile and flexibility is seen with the extension of the nylon fibers to separate the ductile with the durable file. In flexibility, the nylon fiber structured bar was in contact with the heaviest load and the break was almost short. The fracture tip pressure decrease with sluggish rotation. Improving joint stability by achieving optimal levels of fiber efficiency is important along with energy retention, fracture control and rigidity [¹⁵, ¹⁶]. HARI and MINI [¹⁷] have provided high strength SCC sisal-Nylon 6 mono and hybrid fiber composites in the ratio of 0/100, 25/75, 50/50, 75/25 and 100/0. Hybridization worked on flexible and tensile properties against nanofiber compounds [¹⁸]. In addition, stress strain and compliance behavior are increased by hybridization. The formed fiber-structured concrete went through a satisfactory water infiltration and prevents the chances of fiber decay affecting considerable stability [¹⁹]. Despite the fact that nylon enhances mechanical properties it starts to have solidity issues when thinking about openness [²⁰]. A scalable study using ANOVA was performed to demonstrate the adequacy of hybridization and fiber size in the mechanical properties of self-compressing half- and half-fiber concrete. TURK et al. [²¹] have introduced the impact of all-out volume area and large-scale and miniature steel fiber composites. SCC composites with 1 % and 1.5% complete block section fiber were designed as a composite structured with two single strands of large-sized steel fiber and two cross-sectional fibers and a fiber-free control compound. To understand the new features of the SCC, the troupe stream distance was made throughout the T500 and J-Ring test. Compressive strength, separation stiffness, flexural strength, load-carrying range, flexibility durability and flexibility were evaluated to evaluate the mechanical and flexible performance of SCC. The test results passed for the new properties showed that all steel fibers in SCC alloys yielded the benefit of EFNARC, despite the reduction in the activity of FR-SCC compounds with an increase in the absolute volume fractions of the fibers. Flexibility stability values ranged from 1% to 1.5% with the absolute volume area of the fibers. SIMALTI et al. [²²] have introduced new and hardening (e.g. mechanical and rigidity) behavior of recycled steel fiber (RSF) extracted from destroyed tire in SCC. To describe the above properties, seven compounds are ready, one control compounds for example SCC without fiber, SCC with manufactured steel fiber (MSF) and RSF with 0.5%, 1% and 1.5% components each separately. The new properties were illustrated by the Trope Stream and J-Ring test, capable of filling and passing separately. However, the mechanical properties were explored up to compressive, separable and flexible strength. Furthermore, rapid chloride penetration test (RCPT) and ultrasonic heart rate testing were operated for rigidity and individually non-destructive testing. Regulatory results show that SCC with RSF generally performs better with a 1.5% volume fraction of RSF when MSF contrasts with SCC. With R² = 0.7, a decent correlation between the new, mechanical and hardness properties was found. KUMAR et al. [²³] have introduced SCC tests. Tubular shaped SCC tests to determine the impact of SCC additives on steel support consumption up to mass distress then accelerated corrosion. The findings of this study show that SCC structured fiber can be used in the manufacture of concrete to support new and mechanical properties practically without any barriers. PETROVI et al. [²⁴] have presented investigations into the flexible behavior of reinforced Concrete (RC) Fixed Bars Made of SCC Reinforced with Fiber Reinforced Polymer (FRP) Materials (GFRP) and Carbon (CFRP) Bars. Using near surface mounted (NSM) and remotely reinforced (EB) techniques. Six two-range static light emitters with a full length of 3200 mm in the range between 1500 mm and 120/200 mm cross-sectional supports were exposed to the temporary load and attempted. Removal of bars and strains on concrete, steel support, FRP bars and tapes was recorded until disappointment under the monopoly expanding load. The firm load limits of the reinforced shafts were improved from 22% to 82% compared to the reinforced control pole. HARI and MINI [¹⁷] have provided high strength SCC sisal-Nylon 6 mono and hybrid fiber composites in the ratio of 0/100, 25/75, 50/50, 75/25 and 100/0. The hybridization technique is achieved efficient strength but consumes huge time. TURK et al. [²¹] have introduced the impact of all out volume area and large-scale and miniature steel fiber composites. However, five composite structures only analyzed. SIMALTI et al. [²²] had introduced new and hardening (e.g. mechanical and rigidity) behavior of RSF extracted from destroyed tire in SCC. However, it only analyzed the experimental analysis. KUMAR et al. [²³] have introduced SCC tests. Moreover, it is analyzed the tabular shape concrete. PETROVI et al. [²⁴] have presented investigations into the flexible behavior of reinforced Concrete RC Fixed Bars Made of SCC Reinforced with FRP GFRP and CFRP Bars. However, it does not analyze the temperature based concrete analysis. The literature on concrete mix design emphasizes optimizing materials to achieve desired properties such as strength, durability, and sustainability. PEREIRA et al. [²⁵] propose a method for designing lightweight concrete mixes, focusing on balancing reduced density while maintaining structural integrity. LEDEZMA and YAURI [²⁶] discuss the development of concrete mix designs for paving blocks using recycled tire materials, highlighting the potential for using sustainable resources while ensuring performance standards. SILVA et al. [²⁷] present a concrete mix design method based on the particle packing concept, which improves durability by optimizing particle arrangement to reduce voids and enhance the material’s mechanical properties. These studies illustrate the growing emphasis on innovative mix design techniques that meet functional requirements and address sustainability and environmental concerns. The compressive strength evaluation of hybrid fiber reinforced self-compacting concrete is an essential topic in recent years. The Artificial Intelligence (AI) techniques [²⁵, ²⁶] are utilized to analysis the compressive strength and flexural strength of the developed concrete [²⁷]. The main objective of this present study is to predict the strength behaviour of hybrid fiber reinforced self-compacting concrete and the findings are implemented in MATLAB and the results are discussed elaborately.

2. MATERIALS AND METHODS

The utilization of fibers as reinforcement is not a new notion. In the ancient times, fibers are utilized in the reinforcement purposes. The fiber reinforced building components are being used in special structures.

2.1. Materials used

2.1.1. Cement

OPC 53 grade, conforming to the parameters specified in IS 12269:2013 (Reaffirmed 2018) and tested as per IS 4031:1999 (Part II, IV, V) (Reaffirmed 2018) and IS 1727:1967 (Reaffirmed 2018), was used along with fly ash as the binder material for SCC preparation.

2.1.2. Water

Ordinary potable water, along with the superplasticizer (SP) and viscosity-modifying agent (VMA), was used to achieve the desired workability. In this study, Conplast SP 430, a superplasticizer conforming to IS: 9103-1999 with a specific gravity of 1.1 to 1.2, was utilized. Additionally, MasterMatrix 2 (Glenium Stream 2), a VMA with a specific gravity of 1.19, was incorporated.

2.1.3. Composite binder

The materials used in this study included limestone micro filler, ground quartz sand, and active silica. Limestone micro filler, with a specific gravity of 2.3, was sourced from the local area in Coimbatore. ground quartz sandwas derived from natural quartz deposits, silica sand deposits, or crushed quartz crystals, while active silica was collected from nearby local areas.

2.1.4. Fibers

The fibres were received from local suppliers (Go Green Products Pvt.Ltd Chennai). The SEM image of fibres utilized for the experimental investigation is shown in Figure 1a, 1b and 1c respectively and the properties of fibers listed in Table 1.

Figure 1
(a) Basalt fiber, (b) Steel fiber (c) Active Silica containing Additive Morphology Design.

Table 1
Properties of fiber.

S. NO	BEHAVIORS	BASALT FIBER	STEEL FIBER
1	Density kg/m3	2600	7800
2	Resistant to corrosion and alkalis	High	Medium
3	Melting temperature °C	1450	1550
4	Elongation ratio (%)	3.2	3.5
5	Elastic modulus (GPa)	75	190
6	Fiber length, mm	6 + 1.5	13
7	Fiber diameter, mm	13*10–3	1.2
8	Tensile strength MPa	3500	600–1500

2.2. Mix design

The development of active silica containing additive is a partial changeset for cement confirms its efficacy up to substitution in the range of 30% and it is a trend achieves for the fineness of grinding in the period of 500 to 900 m²/kg. In this scenario, the highest cause is attained with the consideration of surface location of 550m²/kg [²¹]. Additionally, the improvement in the fineness of grinding did not encourage to an activity improvement. The mix composition can compensate complete performance condition for the concrete in combined hardened states and fresh states. Normally, it is recommended to develop conservatively to empower that the concrete have the ability to manage the specified fresh properties. In this research, different mix compositions are considered and analyzed with laboratory trails. The complete concrete mix is designed related with the concrete mix design method of ENFARC [²⁸]. Below the optimized mix proportion for M70 (Mix 70) SCC, Cement 530 kg/m³, Fine aggregate 910 kg/m³, Coarse aggregate 810 kg/m³ water 170 kg/m³ (Table 2).

Table 2
Composite binder parameters.

MIX ID	SUPERPLASTICIZER % OF THE BINDER	BINDER BEHAVIOR (%)
		ACTIVE SILICA CONTAINING ADDITIVE	LIMESTONE	GROUND QUARTZ SAND	PORTLAND CEMENT
1–1	1.0	25	2.5	2.5	70
1–2	1.3	31	2.5	2.5	64
1–3	1.6	37	2.5	2.5	58
2–1	1.0	25	5	5	65
2–2	1.3	31	5	5	59
2–3	1.6	37	–	–	63
3–1	1.0	25	7.5	7.5	60
3–2	1.3	31	–	–	–
3–3	1.6	37	–	–	63
CM	–	–	–	–	100

Additionally, to develop this concrete, 0.5% volume of basalt fiber and 1% of steel fiber is added to the 1 cubic meter of concrete mix. The binder ratio of the water is 0.4.

2.3. Recurrent neural network

In the developed concrete, the strength is evaluated with the help of ORNN. The ORNN is a combination of RNN and DO. In the RNN, the optimal weighting parameters are selected by using DO. The detail description of the RNN [²⁹] is explained in this section. The RNN can be a sequential data networks and its main purpose can be identifying the subsequent phase in a sequence of computations related to before phase in the similar sequence. It consists of hidden layer distributed across time that permits them to save data achieved in before phases of sensing serial information. The concrete mix based on their materials and it considered as the long-term problems. The conventional RNN design is unable to manage the long period dependencies. In the RNN, the unfolding problem is increased which the gradient of different weights initiates to large or small if the network can be unfolded for different time. This operation can be defined as the gradient vanishing issue and it is saving the short-term memory only, due it consists the operations of starting the hidden layer of the before phase only and it creates the loss of data in long term. This type of error is solving by utilizing the LSTM network. In the operation, the hidden layer is changed with the consideration of blocks, it designed by trap gates in the error block, it is defined as error carrousel. The RNN structure is illustrated in Figure 2. The RNN model is designed with the basis of below formulation.

Figure 2
RNN structure.

(1) $$\{H_t=\tan h(W_{HX}{X}_T+W_{HY}{Y}_{T-1})Y_T=\sigma (W_{Ht})$$

Here, σ can be described as sigmoid function which is considered the activation function of the output layer, tan h can be described as hyperbolic tangent which is an activation function in the hidden layer, W_HY can be described as the weight of the delayed outcomes in the time period of t−1, W_HX can be described as the weight of the input layer, Y_T can be described as the parameter at the output layer in the time period t, h_T can be described as the state parameter of the hidden layer and X_T can be described as the input variable.

2.3.1. Long short-term memory system

Long short-term memory system is a kind of RNN developed to decrease the issue of long-term dependence here every neuron consists a memory cell able to save the previous data with the consideration of RNN or forgetting it if required. Normally, it can be utilized in the success time series detection issues. LSTM-RNN network is developed with the consideration of memory cell which save long term dependencies. Additionally, the memory cell, the LSTM cell consist of an forget gate, output gate and input gate. Every fate in the memory cell achieves the hidden state, current input at the previous stage in addition state data of the cell internal storage memory to present different functions in addition to compute if to operate with the utilization of logic function. The state of unit operation, input hidden state at time, output at time can be computed by non-linearly activation function in addition data of the output gate [³⁰, ³¹] (Figure 3).

Figure 3
LSTM structure.

The operations of time series inside a LSTM is explained as follows,

• The unnecessary data in the LSTM can be detected in addition omitted from the state cell with the consideration of sigmoid layer named as forget gate layer is described by,

(2) $$F_t=\sigma \left[w_f(h_{t-1},x_t)+b_f\right]$$

Where, x_t can be described as the new input information, h_t−1 can be described as the output from the previous time stamp, b_f can be described as the weight, w_f can be described as the bias function [³²].

• The novel data is saved in the cell state of LSTM and it computed and updated in the layer of sigmoid which named as the input gate layer. After that, the tanh layer generates a vector parameter of novel candidate solutions which can be added to the state variable.

(3) $$i_t=\sigma \left[w_i(h_{t-1},x_t)+b_i\right]$$

(4) $$C_t^{*}=tanh\left[w_c(h_{t-1},x_t)+b_i\right]$$

• The old cell state function can be upgraded into the novel cell state. The updating procedure of the cell state is explained below.

(5) $$C_t=F_t*C_{t-1},+i_t*C_t*$$

Where, C_t−1 can be described as old cell state, C_t can be described as new cell state.

• A sigmoid layer can be operated toward decide which cell parameters of the cell state can be generating the output. It can be formulated as follows,

(6) $$o_t=\sigma \left[w_o(h_{t-1},x_t)+b_i\right]$$

(7) $$h_t=o_t*tanh(C_t)$$

The compressive strength of the proposed concrete is assessed and analyzed based on the given information.

2.4. Dingo optimizer

In the projected technique, DO is utilized to choose optimal weighting parameters of RNN. The dingoes have the ability to generate the optimal sense of communication. In the DO, its communication with different dingoes with the consideration of sensing behaviour with various acoustic levels in the atmosphere. In dingos, it generates acoustic response in such a way which dingoes transfer their data in addition to generate general community information. The vibration and amplitude us changed with the consideration of person strength in the dingoes enter a novel location from the last one. Group hunting is an efficient social characteristic of dingoes that generate its more extension to the social characteristics of dingoes. The hunting method can be divided in their stages such as attack, encircling and harassing, chasing and approaching. The mathematical model of the dingos is presented in this section.The management of the attacking prey, encircling and searching can be designed mathematically to functioning the dingo optimization [³³, ³⁴].The management of the attacking prey, encircling and searching can be designed mathematically to functioning the dingo optimization.

2.4.1. Encircling

Dingoes have the ability to determine the coordinates of their prey. Once the location is computed, the dingo pack follows as the alpha encircles the prey. When designing the social hierarchy of dingoes, it can be considered similar to the optimal state before the search location is predefined. At a specific time, the remaining search agents continue to refine their methods for the upcoming required approach. The characteristics of dingoes are presented as follows:

(8) $${\overrightarrow{d}}_D=|\overrightarrow{a}.{\overrightarrow{P}}_p(x)-\overrightarrow{p}(i)|$$

(9) $$\overrightarrow{p}(i-1)={\overrightarrow{P}}_p(i)-\overrightarrow{b}.\overrightarrow{d}$$

(10) $$\overrightarrow{a}=2.{\overrightarrow{a}}_1$$

(11) $$\overrightarrow{b}=2\overrightarrow{b}.{\overrightarrow{a}}_2-\overrightarrow{b}$$

(12) $$\overrightarrow{b}=3-\left(i*\left(\frac{3}{i_{max}}\right)\right)$$

Position of the neighbourhood dingoes is described with the consideration of two-dimensional position vector. Based on the prey position, the dingo can be updated their position. Complete the possible locations can be marked in the figure considering the optimal agent, considering the present location through variation of the parameter vectors [³⁵].

2.4.2. Hunting behaviour

Moreover, the search space related to theory, search agent cannot normally relate with the computation of prey position. Developing the dingoes hunting behaviour mathematically, here assumption which complete the pack members contains alpha, beta and others contain the best knowledge related with the location of prey. The alpha dingo gives the rules for hunting. Moreover, the beta and remaining dingoes may participate in the hunting. Moreover, consider the initial two optimal parameters attained so far. Based on the location of optimal search agent and remaining dingoes also required to position updating. This hunting behaviour is mathematically formulated and presented as follows,

(13) $$\overrightarrow{d}a=\left|{\overrightarrow{a}}_1.{\overrightarrow{P}}_{\alpha }-\overrightarrow{p}\right|$$

(14) $${\overrightarrow{d}}_{\beta }=\left|{\overrightarrow{a}}_1.{\overrightarrow{P}}_{\beta }-\overrightarrow{p}\right|$$

(15) $${\overrightarrow{d}}_0=\left|{\overrightarrow{a}}_2.{\overrightarrow{P}}_0-\overrightarrow{p}\right|$$

(16) $${\overrightarrow{P}}_1=\left|{\overrightarrow{P}}_{\alpha }-\overrightarrow{b}.{\overrightarrow{d}}_{\alpha }\right|$$

(17) $${\overrightarrow{P}}_2=\left|{\overrightarrow{P}}_{\beta }-\overrightarrow{b}.{\overrightarrow{d}}_{\beta }\right|$$

(18) $${\overrightarrow{P}}_0=\left|{\overrightarrow{P}}_0-\overrightarrow{b}.{\overrightarrow{d}}_0\right|$$

Where, i_max can be described as a maximum number of iterations, i can be described as the count of parameters, | | can be described as absolute parameter and multiplication with vectors, $\overrightarrow{b}$ can be described as linearly decreased from 3 to 0 each iteration, ${\overrightarrow{a}}_1,{\overrightarrow{a}}_2$ can be described as the random vector, ${\overrightarrow{P}}_p$ can be described as coefficient vector, $\overrightarrow{b}$ and $\overrightarrow{a}$ can be described as dingo position vector, $\overrightarrow{p}$ can be described as position vector prey and $\overrightarrow{d}$ can be described as distance among the dingo and prey.

To compute the intensity of every dingo, below equations are considered.

(19) $${\overrightarrow{l}}_{\alpha }=log\left(\frac{1}{f_{\alpha }-(le-100)}+1\right)$$

(20) $${\overrightarrow{l}}_{\beta }=log\left(\frac{1}{f_{\beta }-(le-100)}+1\right)$$

(21) $${\overrightarrow{l}}_0=log\left(\frac{1}{f_0-(le-100)}+1\right)$$

The positions of the alpha, beta and different dingoes are updated. Initially, the positions are randomly selected and compute the prey position in the search space. The pseudocode of the DO is presented in algorithm 1.

Algorithm 1: Dingo Optimizer Input: The random weight parameters of RNNOutput: Optimal weight parameter of RNNInitialization of search agentInitialization of parametersWhile termination condition not achieved doCompute dingo intensity cost and dingo fitnessDingo with the optimal search (alpha)Dingo with second optimal search (beta)Dingo of remaining search resultsIteration 1RepeatFor i = 1; Din doRenew the new search agent statusEnd forCompute the intensity cost and fitness of dingoesSave the valuesiteration = iteration + 1Maximum iteration > Stopping conditionOutputEnd while

Algorithm 1: Dingo Optimizer

Input: The random weight parameters of RNN Output: Optimal weight parameter of RNN Initialization of search agent Initialization of parameters While termination condition not achieved do Compute dingo intensity cost and dingo fitness Dingo with the optimal search (alpha) Dingo with second optimal search (beta) Dingo of remaining search results Iteration 1 Repeat For i = 1; Din do Renew the new search agent status End for Compute the intensity cost and fitness of dingoes Save the values iteration = iteration + 1 Maximum iteration > Stopping condition Output End while

2.4.3. Prey attacking

If there is no position upgrade, it describes complete the hunt through prey attacking. To mathematically arranged the method, the value of b is reduced linearly. The projected encircling technique does indeed reveal exploration in few extents. Moreover, the accentuate exploration, the DO is necessary more operators [³⁶, ³⁷].

2.4.4. Searching

Dingoes hunt for the prey mostly related to the location of packs. The dingoes travel forward to hunt for strike predators. Based on the DO, the optimal weight parameters are selected which sent to the RNN classifier to analysis the concrete behaviour.

3. RESULTS AND DISCUSSION

The performance of the projected technique is analyzed and validated in this portion. The SCC mixture presentation is analyzed with four test specimens such as slump, v-funnel, L-box and U-box.The V-funnel test is analyzed with the V-funnel device which is a 10L stainless steel with waterproof design inside area. Additionally, the top edge can be reinforcing and smooth and output area is connected with sealing valve. The L-box test also considered to analysis the flow rates and the flow of the mixture. The L-box is containing a test basin, three obstacles, sliding interior surfaces and concrete tank. Additionally, the U-box also utilized to compute the flowability of concrete mix in communicating vessels. The flow diagram of testing is presented in Figure 4. Temperature significantly affects the properties of concrete. At higher temperatures, concrete experiences reduced strength due to the thermal degradation of cement paste, which can lead to cracking and a decrease in durability. Conversely, low temperatures can delay the hydration process, leading to slower strength development. Proper temperature control during mixing and curing is essential to maintain optimal concrete performance. In this study, the model was developed using a single LSTM layer with 100 hidden units and a dropout rate of 0.2 to prevent overfitting, while the DO was implemented with a population size of 30, a maximum of 100 iterations, and default control parameters for convergence. These hyper parameter choices were empirically tuned for optimal performance based on the characteristics of the concrete mix dataset.

Figure 4
(a) flow diagram of testing and (b) graphical representation of the heating cooling cycles and (c) training of ANN.

The concrete mix general properties are illustrated in Figure 5 and Table 3. Additionally, the slump test results also presented in Figure 6. To validate the projected concrete, it is analyzed with different temperatures such as 210°C , 320°C, 530°C and 790°C. The compressive strength of the projected concrete is presented in Figure 7 and Table 4. Based on the analysis, the projected concrete is achieved 70.36GPa compressive strength at low temperature. When temperature increase, the compressive strength also decreases. For example, during high temperature, the compressive strength is 61.25 GPa. From the analysis, the projected concrete is attained the best compressive strength with the temperature analysis. To validate the projected concrete, it is analyzed with different temperatures such as 210°C, 320°C, 530°C and 790°C. The split tensile strength of the projected concrete is presented in Figure 8 and Table 5. Based on the analysis, the projected concrete is achieved 0.5 N/mm² split tensile strength at low temperature. When temperature increase, the split tensile strength also decreases. For example, during high temperature, the split tensile strength is 0.19 N/mm². From the analysis, the projected concrete is attained the best split tensile strength with the temperature analysis. To validate the projected concrete, it is analyzed with different temperatures such as 210°C, 320°C, 530°C and 790°C. The flexural strength of the projected concrete is presented in Figure 9 and Table 6. Based on the analysis, the projected concrete is achieved 3.15 N/mm² split tensile strength at low temperature. When temperature increase, the split tensile strength also decreases. For example, during high temperature, the flexural strength is 1.58 N/mm². From the analysis, the projected concrete is attained the best flexural strength with the temperature analysis. For Self-Compacting Concrete (SCC) specimens, curing conditions typically involve maintaining adequate humidity and temperature to ensure proper hydration and strength development. The specimens are usually cured under 95% relative humidity and at temperatures ranging from 20°C to 25°C for a minimum duration of 28 days. However, some studies may also involve accelerated curing at elevated temperatures (e.g., 40°C to 60°C) for specific testing requirements, such as early strength or exposure to high-temperature conditions. These conditions can vary depending on the research objectives and the specific mix design used.

Figure 5
General properties of concrete mix.

Table 3
Workability property of SCC with hybrid fibers.

IDENTITY OF MIX	SLUMP TEST (mm)	VISCOSITY T50 U-BOX (s)	V FUNNEL (s)	L BOX (h2/h1)
1–1	710	2.18	11.58	0.905
1–2	710	2.22	9.34	0.924
1–3	700	3.12	10.03	0.908
2–1	700	2.02	14.32	0.932
2–2	680	2.10	12.00	0.980
2–3	710	3.09	14.12	0.880
3–1	720	4.15	15.57	0.840
3–2	700	4.56	13.91	0.905
3–3	710	4.03	12.75	0.932
CM	710	3.81	8.58	0.912

Figure 6
Slump test variable.

Figure 7
Compressive strength analysis.

Table 4
Compressive strength of projected mixture at 28 days.

MIX ID	210°C (Mpa)	320°C (Mpa)	530°C (Mpa)	790°C (Mpa)
1–1	70.36	68.15	65.87	61.25
1–2	71.96	70.56	68.54	64.21
1–3	72.21	71.54	69.12	64.45
2–1	71.36	70.36	68.12	61.25
2–2	70.21	67.15	63.98	60.95
2–3	74.39	70.14	69.54	67.12
3–1	71.36	70.15	68.45	65.14
3–2	72.65	70.58	67.54	64.45
3–3	70.11	68.57	65.45	61.54
CM	70.42	68.95	65.45	61.54
SD	1.17	1.52	1.59	2.21

Figure 8
Split tensile strength analysis.

Table 5
Split tensile strength of projected mixture.

MIX ID	210°C (Mpa)	320°C (Mpa)	530°C (Mpa)	790°C (Mpa)
1–1	0.5	0.4	0.2	0.19
1–2	0.54	0.35	0.28	0.25
1–3	0.6	0.54	0.34	0.30
2–1	0.15	0.10	0.09	0.07
2–2	2.15	1.98	1.78	1.25
2–3	1.54	1.78	1.15	1.12
3–1	0.34	0.32	0.21	0.15
3–2	0.99	0.78	0.65	0.55
3–3	1.12	1.02	0.97	0.54
CM	2.14	1.89	1.56	1.34
SD	1.17	1.52	1.59	2.21

Figure 9
Flexural strength analysis.

Table 6
Flexural strength of projected mixture.

MIX ID	210°C	320°C	530°C	790°C
1–1	3.15	2.10	1.98	1.58
1–2	5.75	4.35	3.87	2.54
1–3	2.58	2.25	2.10	1.58
2–1	6.75	5.13	4.89	4.78
2–2	4.58	4.32	4.02	3.88
2–3	8.15	7.15	6.98	5.18
3–1	6.54	5.15	4.98	4.32
3–2	7.54	7.02	6.54	5.18
3–3	5.12	4.85	4.32	3.89
CEM	6.48	5.48	5.02	4.01

4. CONCLUSION

1. The results demonstrated that incorporating hybrid fibers in self-compacting concrete (SCC) improves its mechanical properties but compromises workability, as indicated by a significant increase in the difference slump flow diameters.

2. The reduction in compressive strength is primarily attributed to the loss of free water in capillary pores and the initiation of microcracking. The decomposition of Calcium Hydroxide (Ca (OH)₂) into CaO and H₂O causes volumetric shrinkage, which leads to an increase in microcracks.

3. The experimental results arrived were implemented in MATLAB for computing the compressive strength, tensile strength and flexural strength. Based on the analysis, the projected concrete is achieved 70.36 GPa compressive strength at low temperature. When temperature increase, the compressive strength also decreases. For example, during high temperature, the compressive strength is 61.25 GPa.

4.1. Limitations and future scope

The study on hybrid fiber reinforced SCC has limitations, including a narrow scope of fibers and experimental conditions that may not reflect real-world applications. Future research should explore a broader range of fibers, optimize fiber hybridization, and investigate long-term durability and environmental impact. Additionally, large-scale applications and integration with smart materials for real-time monitoring could enhance SCC’s performance. Exploring these areas will improve the sustainability, cost-effectiveness, and durability of SCC in practical use.

5. BIBLOGRAPHY

• ^[1] REVILLA-CUESTA, V., SKAF, M., FALESCHINI, F., et al., “Self-compacting concrete manufactured with recycled concrete aggregate: an overview”, Journal of Cleaner Production, v. 262, pp. 121362, Jul. 2020. doi: http://doi.org/10.1016/j.jclepro.2020.121362.
  » https://doi.org/10.1016/j.jclepro.2020.121362
• ^[2] ASLANI, F., GUOWEI, M.A., YIM WAN, D., “Development of high-performance self-compacting concrete using waste recycled concrete aggregates and rubber granules”, Journal of Cleaner Production, v. 182, pp. 553–566, May. 2018. doi: http://doi.org/10.1016/j.jclepro.2018.02.074.
  » https://doi.org/10.1016/j.jclepro.2018.02.074
• ^[3] ABID, S., ABDUL-HUSSEIN, M.L., AYOOB, N.S., et al., “Repeated drop-weight impact tests on self-compacting concrete reinforced with micro-steel fiber”, Heliyon, v. 6, n. 1, e03198, Jan. 2020. doi: http://doi.org/10.1016/j.heliyon.2020.e03198.PubMed PMID: 31956719.
  » https://doi.org/10.1016/j.heliyon.2020.e03198
• ^[4] SHARIFI, E., SADJADI, S., ALIHA, M.R.M., et al., “Optimization of high-strength self-consolidating concrete mix design using an improved Taguchi optimization method”, Construction & Building Materials, v. 236, pp. 117547, 2020. doi: http://doi.org/10.1016/j.conbuildmat.2019.117547.
  » https://doi.org/10.1016/j.conbuildmat.2019.117547
• ^[5] REVILLA-CUESTA, V., ORTEGA-LÓPEZ, V., SKAF, M., et al., “Effect of fine recycled concrete aggregate on the mechanical behavior of self-compacting concrete”, Construction & Building Materials, v. 263, pp. 120671, Dec. 2020. doi: http://doi.org/10.1016/j.conbuildmat.2020.120671.
  » https://doi.org/10.1016/j.conbuildmat.2020.120671
• ^[6] HENDI, A., MOSTOFINEJAD, D., SEDAGHATDOOST, A., et al., “Mix design of the green self-consolidating concrete: Incorporating the waste glass powder”, Construction & Building Materials, v. 199, pp. 369–384, Feb. 2019. doi: http://doi.org/10.1016/j.conbuildmat.2018.12.020.
  » https://doi.org/10.1016/j.conbuildmat.2018.12.020
• ^[7] AWOYERA, P., KIRGIZ, M., VILORIA, A., et al., “Estimating strength properties of geopolymer self-compacting concrete using machine learning techniques”, Journal of Materials Research and Technology, v. 9, n. 4, pp. 9016–9028, Jul-Aug. 2020. doi: http://doi.org/10.1016/j.jmrt.2020.06.008.
  » https://doi.org/10.1016/j.jmrt.2020.06.008
• ^[8] SINGH, N., KUMAR, P., GOYAL, P., “Reviewing the behaviour of high-volume fly ash based self-compacting concrete”, Journal of Building Engineering, v. 26, pp. 100882, Nov. 2019. doi: http://doi.org/10.1016/j.jobe.2019.100882.
  » https://doi.org/10.1016/j.jobe.2019.100882
• ^[9] GARCIA-TAENGUA, E., “Fundamental fresh state properties of self-consolidating concrete: a meta-analysis of mix designs”, Advances in Civil Engineering, v. 2018, n. 1, pp. 5237230, 2018. doi: http://doi.org/10.1155/2018/5237230.
  » https://doi.org/10.1155/2018/5237230
• ^[10] SASANIPOUR, H., ASLANI, F., TAHERINEZHAD, J., “Effect of silica fume on durability of self-compacting concrete made with waste recycled concrete aggregates”, Construction and Building Materials, v. 227, pp. 116598 Dec. 2019. doi: http://doi.org/10.1016/j.conbuildmat.2019.07.324.
  » https://doi.org/10.1016/j.conbuildmat.2019.07.324
• ^[11] JAGADESH, P., DE PRADO-GIL, J., SILVA-MONTEIRO, N., et al., “Assessing the compressive strength of self-compacting concrete with recycled aggregates from mix ratio using machine learning approach.”, Journal of Materials Research and Technology, v. 24, pp. 1483–1498, 2023. doi: http://doi.org/10.1016/j.jmrt.2023.03.037.
  » https://doi.org/10.1016/j.jmrt.2023.03.037
• ^[12] JAGADESH, P., KHAN, A.H., PRIYA, B.S., et al., “Artificial neural network, machine learning modelling of compressive strength of recycled coarse aggregate based self-compacting concrete”, PLoS One, v. 19, n. 5, pp. e0303101, 2024. doi: http://doi.org/10.1371/journal.pone.0303101.PubMed PMID: 38739642.
  » https://doi.org/10.1371/journal.pone.0303101
• ^[13] ZHANG, J., MA, G., HUANG, Y., et al., “Modelling uniaxial compressive strength of lightweight self-compacting concrete using random forest regression”, Construction & Building Materials, v. 210, pp. 713–719, Jun. 2019. doi: http://doi.org/10.1016/j.conbuildmat.2019.03.189.
  » https://doi.org/10.1016/j.conbuildmat.2019.03.189
• ^[14] DE-PRADO-GIL, J., MARTÍNEZ-GARCÍA, R., JAGADESH, P., et al., “To determine the compressive strength of self-compacting recycled aggregate concrete using artificial neural network (ANN)”, Ain Shams Engineering Journal, v. 15, n. 2, pp. 102548, 2024. doi: http://doi.org/10.1016/j.asej.2023.102548.
  » https://doi.org/10.1016/j.asej.2023.102548
• ^[15] JAIN, A., GUPTA, R., CHAUDHARY, S., “Sustainable development of self-compacting concrete by using granite waste and fly ash”, Construction & Building Materials, v. 262, pp. 120516, Nov. 2020. doi: http://doi.org/10.1016/j.conbuildmat.2020.120516.
  » https://doi.org/10.1016/j.conbuildmat.2020.120516
• ^[16] SASANIPOUR, H., ASLANI, F., “Durability properties evaluation of self-compacting concrete prepared with waste fine and coarse recycled concrete aggregates”, Construction & Building Materials, v. 236, pp. 117540, Mar. 2020. doi: http://doi.org/10.1016/j.conbuildmat.2019.117540.
  » https://doi.org/10.1016/j.conbuildmat.2019.117540
• ^[17] HARI, R., MINI, K., “Mechanical and durability properties of sisal-Nylon 6 hybrid fibre reinforced high strength SCC”, Construction & Building Materials, v. 204, pp. 479–491, Apr. 2019. doi: http://doi.org/10.1016/j.conbuildmat.2019.01.217.
  » https://doi.org/10.1016/j.conbuildmat.2019.01.217
• ^[18] UYSAL, M., “Self-compacting concrete incorporating filler additives: performance at high temperatures”, Construction & Building Materials, v. 26, n. 1, pp. 701–706, 2012. doi: http://doi.org/10.1016/j.conbuildmat.2011.06.077.
  » https://doi.org/10.1016/j.conbuildmat.2011.06.077
• ^[19] ANDRUSHIA, D., ANAND, N., ARULRAJ, P., “Anisotropic diffusion based denoising on concrete images and surface crack segmentation”, International Journal of Structural Integrity, v. 11, n. 3, pp. 395–409, 2020. doi: http://doi.org/10.1108/IJSI-06-2019-0061.
  » https://doi.org/10.1108/IJSI-06-2019-0061
• ^[20] THANARAJ, D.P., ANAND, N., ARULRAJ, P., “Experimental investigation of mechanical properties and physical characteristics of concrete under standard fire exposure”, Journal of Engineering, Design and Technology, v. 17, n. 5, pp. 878–903, 2019. doi: http://doi.org/10.1108/JEDT-09-2018-0159.
  » https://doi.org/10.1108/JEDT-09-2018-0159
• ^[21] TURK, K., BASSURUCU, M., BITKIN, R.E., “Workability, strength and flexural toughness properties of hybrid steel fiber reinforced SCC with high-volume fiber”, Construction & Building Materials, v. 266, pp. 120944, Jan. 2021. doi: http://doi.org/10.1016/j.conbuildmat.2020.120944.
  » https://doi.org/10.1016/j.conbuildmat.2020.120944
• ^[22] SIMALTI, A., SINGH, A.P., “Comparative study on performance of manufactured steel fiber and shredded tire recycled steel fiber reinforced self-consolidating concrete”, Construction & Building Materials, v. 266, pp. 121102, Jan. 2021. doi: http://doi.org/10.1016/j.conbuildmat.2020.121102.
  » https://doi.org/10.1016/j.conbuildmat.2020.121102
• ^[23] KUMAR, K.R., SHYAMALA, G., ADESINA, A., “Structural performance of corroded reinforced concrete beams made with fiber-reinforced self-compacting concrete”, Structures, v. 32, pp. 1145–1155, Aug. 2021. doi: http://doi.org/10.1016/j.istruc.2021.03.079.
  » https://doi.org/10.1016/j.istruc.2021.03.079
• ^[24] PETROVIC, Z., MILOŠEVIĆ, B., RANKOVIĆ, S., et al., “Experimental analysis of continuous beams made of Self-Compacting Concrete (SCC) strengthened with Fiber Reinforced Polymer (FRP) materials”, Applied Sciences, v. 11, n. 9, pp. 4032, 2021. doi: http://doi.org/10.3390/app11094032.
  » https://doi.org/10.3390/app11094032
• ^[25] PEREIRA, W.B., GALVÃO, S.P., VERAS, J.C., et al., “Mix desing method lightweight concrete “, GeSec - Revista de Gestao e Secretariado, v. 15, n. 2, e3509, 2024. doi: http://doi.org/10.7769/gesec.v15i2.3509.
  » https://doi.org/10.7769/gesec.v15i2.3509
• ^[26] LEDEZMA, F., YAURI, W., “Diseño de mezcla del concreto para elaboración de adoquines con material reciclado de neumáticos en la provincia de Huancavelica”, Revista Científica Ciencias Ingenieriles, v. 3, n. 1, pp. 51–62, 2023. doi: http://doi.org/10.54943/ricci.v3i1.224.
  » https://doi.org/10.54943/ricci.v3i1.224
• ^[27] SILVA, L.A., SILVA, W.K.D., CABRAL, A.E.B., et al., “Concrete mix design method for durability based on particle packing concept.”, Revista IBRACON de Estruturas e Materiais, v. 17, n. 4, pp. e17404, 2023. doi: http://doi.org/10.1590/s1983-41952024000400004.
  » https://doi.org/10.1590/s1983-41952024000400004
• ^[28] EUROPEAN FEDERATION OF PRODUCERS AND APPLICATORS OF SPECIALIST PRODUCTS FOR STRUCTURES, Specification and Guidelines for Self-compacting Concrete, EFNARC, 2002.
• ^[29] JAVID, G., ABDESLAM, D., BASSET, M., “Adaptive online state of charge estimation of EVs lithium-ion batteries with deep recurrent neural networks”, Energies, v. 14, n. 3, pp. 758, Feb. 2021. doi: http://doi.org/10.3390/en14030758.
  » https://doi.org/10.3390/en14030758
• ^[30] HADDADPAJOUH, H., DEHGHANTANHA, A., KHAYAMI, R., et al., “A deep recurrent neural network-based approach for internet of things malware threat hunting”, Future Generation Computer Systems, v. 85, pp. 88–96, Aug. 2018. doi: http://doi.org/10.1016/j.future.2018.03.007.
  » https://doi.org/10.1016/j.future.2018.03.007
• ^[31] LIN, J.C., SHAO, Y., DJENOURI, Y., et al., “ASRNN: a recurrent neural network with an attention model for sequence labeling”, Knowledge-Based Systems, v. 212, pp. 106548, Jan. 2021. doi: http://doi.org/10.1016/j.knosys.2020.106548.
  » https://doi.org/10.1016/j.knosys.2020.106548
• ^[32] SHERSTINSKY, A., “Fundamentals of recurrent neural network (RNN) and long short-term memory (LSTM) network”, Physica D. Nonlinear Phenomena, v. 404, pp. 132306, Mar. 2020. doi: http://doi.org/10.1016/j.physd.2019.132306.
  » https://doi.org/10.1016/j.physd.2019.132306
• ^[33] BAIRWA, A.K., JOSHI, S., SINGH, D., “Dingo optimizer: a nature-inspired metaheuristic approach for engineering problems”, Mathematical Problems in Engineering, v. 1, pp. 1–12, 2021. doi: http://doi.org/10.1155/2021/2571863.
  » https://doi.org/10.1155/2021/2571863
• ^[34] ALMAZÁN-COVARRUBIAS, J.H., PERAZA-VÁZQUEZ, H., PEÑA-DELGADO, A.F., et al., “An improved dingo optimization algorithm applied to SHE-PWM modulation strategy”, Applied Sciences, v. 12, n. 3, pp. 992, Jan. 2022. doi: http://doi.org/10.3390/app12030992.
  » https://doi.org/10.3390/app12030992
• ^[35] BAIRWA, A., JOSHI, S., SINGH, D., “Dingo optimizer: a nature-inspired metaheuristic approach for engineering problems”, Mathematical Problems in Engineering, v. 2021, n. June, pp. 1–12, 2021. doi: http://doi.org/10.1155/2021/2571863.
  » https://doi.org/10.1155/2021/2571863
• ^[36] ALMAZÁN-COVARRUBIAS, J.H., PERAZA-VÁZQUEZ, H., PEÑA-DELGADO, A., “An improved dingo optimization algorithm applied to SHE-PWM modulation strategy”, Applied Sciences, v. 12, n. 3, pp. 992, Jan. 2022. doi: http://doi.org/10.3390/app12030992.
  » https://doi.org/10.3390/app12030992
• ^[37] VÁZQUEZ, A.F., “A bio-inspired method for engineering design optimization inspired by dingoes hunting strategies”, Mathematical Problems in Engineering, v. 2021, n. Sep, pp. 1–19, 2021. doi: http://doi.org/10.1155/2021/9107547.
  » https://doi.org/10.1155/2021/9107547

Publication Dates

• Publication in this collection
  25 July 2025
• Date of issue
  2025

History

• Received
  17 Feb 2025
• Accepted
  24 June 2025