PET Granules Replacement for Fine Aggregate in Concrete and FRP Wrapping Effect: Overview of Experimental Data and Model Development

1. Introduction

Polyethylene terephthalate (PET) is a polymer of great importance among the world's plastics. Thermoplastic recyclability makes it the first choice for various applications [1]. PET, a polyester plastic, is one of the most used packaging materials in beverages. Due to its excellent transparency, light weight, gas and water barrier properties, impact resistance, UV resistance and unbreakability, the production and use of PET bottles for beverage packaging has continuously increased worldwide [2]. With this increase in production and use, recycling solutions for PET began to be produced. PET is seen as one of the plastic types with the highest recycling rate today [3,4,5,6,7].

PET can be recycled in the construction industry as well as in the plastics industry. Reusing plastics such as PET instead of aggregates in concrete is seen as an event that can contribute to environmental protection and sustainable development [8]. It is thought that the use of PET in concrete as a partial or complete replacement for natural aggregates can protect natural aggregate resources [9]. For this purpose, there are many studies examining various properties of concrete by using PET as a fine aggregate replacement in concrete [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34]. When the study results were examined, it was seen that the use of PET generally caused a decrease in the mechanical properties of concrete. It is thought that strengthening methods can be used to compensate for the possible decrease in mechanical properties due to the use of PET. In a study [35], it was stated that wrapping PET substituted concrete with carbon fiber reinforced polymer (CFRP) could be one of these methods.

A frequently used technique to strengthen existing reinforced concrete structures is the use of fiber reinforced polymer (FRP) systems. It is clearly understood that lateral confinement of concrete can significantly increase its strength and ductility. With the introduction of FRP composites into the construction industry, the use of FRP as a wrapping material has attracted great attention [36]. The main advantage of FRP systems is that FRP materials have low mass and high tensile capacity and can provide greater durability when installed properly [37]. FRP is widely used in the construction of new structures and rehabilitation of old structures due to its non-corrosive properties and high strength/weight ratios [38]. The majority of concrete wrapping studies with FRP to date have been done on CFRP and GFRP [39].

In this study, sand in concrete was partially replaced with PET in order to protect natural resources and recycle PET. The results and models of similar studies in literature were brought together and the connections between the experimental results were interpreted. The mechanical properties of concrete have been tried to be increased by using CFRP and GFRP. Similarly, an evaluation was made with studies in literature.

2. Experimental Study

2.1. Materials

CEM I 42.5 R Portland cement, produced in accordance with the EN 197-1 standard of the Marmara Cimento brand, was used in concrete production. Chemical and physical properties of cement are given in Table 1.

Sieve analysis was carried out separately for coarse aggregate (Gravel) and fine aggregate (Sand) in accordance with EN 933-1 standard. Grading curves obtained from sieve analysis are given in Figure 1.

PET granules used in the study were supplied from a plastic recycling company in Turkey. The image of PET granules is given in Figure 2. The properties of PET granules and aggregates are given in Table 2.

CFRP and GFRP were supplied for wrapping around concrete samples. FRP properties are given in Table 3 and visuals are given in Figure 3.

Epoxy was used to bond FRP and concrete together. Epoxy properties are given in Table 4.

$\ge 4MPa$$\ge 80MPa$$\ge 30MPa$$\ge 40MPa$

2.2. Concrete Samples Mix Design and Naming

Concrete production was carried out with the help of a concrete mixing machine. In all samples, the amounts of water, cement and coarse aggregate were kept constant. The water/cement ratio was determined as 0.5. In samples using PET material, the substitution process was carried out as 10%, 20% and 30% of the sand by volume. In the samples with FRP wrapping, after the 28-day curing pool process, epoxy was applied around the concrete and the FRP wrapping process was carried out.

The sample names for the tests followed specific conventions: "CS" for compressive strength, "STS" for splitting tensile, and "FS" for flexural strength. Samples containing PET granules included "PET" in their names, while reference samples without PET were labeled with "R." The substitution percentages were also included. Samples wrapped in carbon FRP were named with "CFRP," and those with glass FRP were labeled "GFRP." Table 5 lists the characteristics and naming details of the samples.

2.3. Methods

The concrete was mixed in a concrete mixing machine, then transferred to the test funnel for a slump test. The concrete was transferred in two stages. After filling half of the funnel, the filling process was paused, and the test rod was dipped into the concrete 25 times. Once this was done, the funnel was completely filled with concrete. After filling was completed, the test rod was dipped into the concrete 25 more times. The upper level of the funnel was leveled off, and the funnel was lifted to release the concrete. The released concrete collapsed at varying heights. The difference between the upper level of the funnel and the upper level of the concrete determined the slump value.

Cylindrical samples, each 10 cm in diameter and 20 cm in height, were created for the compressive strength and splitting tensile tests. Samples with 10x10 cm cross-sectional dimensions and 40 cm in length were produced for the flexural strength tests. While creating the reference concrete samples, 25 Mpa was determined as the target strength. All samples were kept in the curing pool for 28 days after they were produced. The samples coming out of the curing pool were dried and weighed and their densities were found. Compressive strength and splitting tensile strength tests were carried out using the laboratory facilities of Sakarya University, while flexural strength tests were carried out using the laboratory facilities of Sakarya University of Applied Sciences.

In the compressive strength test, cylindrical samples were placed in compressometers with potentiometers that can measure displacements in axial and horizontal directions. Strain gauges were placed on the samples in axial and horizontal directions. The displacement values ​​were found by taking the average of the values ​​in the strain gauges and potentiometers. The samples were taken to the hydraulic pressure machine with the compressometers and loaded in the axial direction at a speed of 0.5 MPa. In the splitting tensile test, a cage mechanism was created so that the cylindrical samples would be subjected to tensile force. The samples were placed horizontally in the pressure machine and loaded at a speed of 0.5 MPa. In the flexural strength test, prismatic samples were subjected to a 3-point bending test and the loading speed was determined as 5 mm/min. Visuals related to the experimental setups are given in Figure 4.

2.4. Experimental Results

2.4.1. Slump

The results of the slump tests are given in Figure 5. The slump value was measured as 48 mm in the reference concrete. Gradual decreases were observed in the slump value as the PET substitution rate increased. The slump value decreased by 6.25% to 45 mm in the case of 10% PET substitution, by 16.67% to 40 mm in the case of 20% PET substitution, and by 20.83% to 38 mm in the case of 30% PET substitution.

2.4.2. Density

The results of the density tests are given in Figure 6. The density value in the reference concrete was measured as 2307.88 kg/m³. As the PET substitution rate increased, gradual decreases were observed in the density value. The density value decreased by 1.57% to 2271.66 kg/m³ in the case of 10% PET substitution, decreased by 3.18% to 2234.49 kg/m³ in the case of 20% PET substitution, and decreased by 5.63% to 2177.94 kg/m³ in the case of 30% PET substitution.

2.4.3. Compressive Strength

The stress-strain graphs obtained as a result of the compressive strength tests performed on samples without FRP are given in Figure 7. The maximum compressive strengths of the samples ($f_{co}^{\prime }$), axial strain (${\epsilon }_{co}$) and lateral strain (${\epsilon }_{lo}$) values ​​at maximum compressive strength are given in Table 6. Compressive strength in the reference concrete was measured as 26.91 MPa. As the PET substitution ratio increased, the compressive strength decreased. Compressive strength decreased by 6.68% to 25.11 MPa in case of 10% PET substitution, by 13.36% to 23.31 MPa in case of 20% PET substitution, and by 18.99% to 21.80 MPa in case of 30% PET substitution. Axial and lateral strain values ​​increased with the increase of PET substitution ratio, contrary to the trend in compressive strength. Axial strain values ​​increased by 2.83%, 7.58% and 7.92% for 10%, 20% and 30% substitution, respectively. These rates were 2.95%, 14.71% and 21.50% for lateral strain values.

Stress-strain graphs of samples using FRP and the sample with 10% PET substitution are given in Figure 8. Maximum compressive strengths ($f_{cc}^{\prime }$), axial strain (${\epsilon }_{cu}$) and lateral strain (${\epsilon }_{lu}$) values ​​at maximum compressive strength of samples using FRP are given in Table 7. The FRP wrapped around the 10% PET substituted sample had a significant effect on the compressive strength and strain values. Although the strength value of the 10% PET substituted sample was lower than the reference sample, it exceeded the strength value of the reference sample thanks to the CFRP and GFRP wrapped around it. The strength value of the 10% PET substituted sample increased by 1.91 times from 25.11 to 47.86 when it was wrapped with CFRP. It increased by 1.54 times to 38.66 when it was wrapped with GFRP. The axial strain values ​​increased by 6.83 times in the CFRP wrapping case and 7.15 times in the GFRP wrapping case. The lateral strain values ​​increased by 11.10 and 13.04 times for CFRP and GFRP, respectively.

2.4.4. Splitting Tensile Strength

The results of the splitting tensile strength test are given in Table 8. The splitting tensile strength of the reference concrete was found to be 2.46 MPa. The increase in PET substitution rate caused decreases in splitting tensile strength values. The splitting tensile strength decreased by 4.47% to 2.35 MPa in the case of 10% PET substitution, by 8.93% to 2.24 MPa in the case of 20% PET substitution, and by 12.11% to 2.16 MPa in the case of 30% PET substitution. The positive effect of FRP was observed in the splitting tensile strength tests as well as in the compressive strength tests. The splitting tensile strengths of the samples with 10% PET substitution reached 3.45 MPa thanks to the use of CFRP and 3.04 MPa thanks to the use of GFRP.

2.4.5. Flexural Strength

Figure 9 shows the graphs obtained as a result of the flexural strength tests of samples without FRP. Table 9 shows the flexural strength and maximum strain values ​​of the samples. The flexural strength of the reference concrete was measured as 3.37 MPa. As the PET substitution rate increased, there were decreases in the flexural strength. The flexural strength decreased by 4.18% to 3.23 MPa in the case of 10% PET substitution, by 7.98% to 3.10 MPa in the case of 20% PET substitution, and by 11.17% to 3.00 MPa in the case of 30% PET substitution. The maximum strain values ​​increased with the increase in the PET substitution rate. The maximum strain increased by 9.36%, 16.16% and 21.90% for the 10%, 20% and 30% substitution cases, respectively.

In Figure 10, flexural stress-flexural strain graphs of samples using FRP and 10% PET substituted sample are given. Table 10 shows the maximum flexural strengths and maximum strain values ​​of samples using FRP. The 10% PET substituted sample managed to exceed the reference sample in terms of compressive and splitting tensile strength as well as flexural strength thanks to the CFRP and GFRP wrapped around it. The flexural strength value of the 10% PET substituted sample increased by 1.42 times when CFRP was wrapped around it, reaching 4.59 MPa. In the case of GFRP wrapping, it increased by 1.27 times and reached 4.09 MPa. The maximum strain values ​​increased by 5.40 times when CFRP was wrapped, and by 5.84 times when GFRP was wrapped.

2.4.6. Modulus of Elasticity

Table 11 shows the elasticity modulus of the samples obtained from the graphs of the compressive strength test. The elasticity modulus of the reference concrete was found to be 30574.17 MPa. Increasing the PET substitution rate decreased the elasticity modulus. The elasticity modulus decreased by 10.96% to 27221.85 MPa in the case of 10% PET substitution, decreased by 18.03% to 25061.49 MPa in the case of 20% PET substitution, and decreased by 30.27% to 21320.87 MPa in the case of 30% PET substitution. The use of FRP increased the elasticity modulus of the 10% PET substitution sample very little, but the value of the reference sample could not be achieved. The elasticity modulus of the 10% PET substitution sample reached 27512.26 MPa with the use of CFRP and 27465.83 MPa with the use of GFRP. The slopes in the second part of the graph were 1704.56 in case of using CFRP and 988.90 in case of using GFRP.

3. Experimental Database and Model Development

In this section, experimental databases were collected separately for PET-substituted concrete and FRP-confined concrete from the literature, and the results of this study were added to these databases. Model proposals were made for various properties of PET-substituted concrete based on the collected database. The applicability of the proposed models and previously proposed models was checked.

Table 12 provides statistical metrics and their explanations for calculating how the results suggested by the models align with the experimental results. In Table 12, while calculating R2 (Coefficient of determination), MABE (Mean absolute bias error) and MAPE (Mean absolute percentage error) values, ${exp}_i$ represents the experimental results, ${mod}_i$ represents the model prediction, ${\overline{exp}}_i$ represents the average of the experimental results, ${\overline{mod}}_i$ represents the average of the model predictions.

3.1. PET Substituted Concrete

The results obtained for PET substituted concrete in the study were combined with previous experimental data in the literature and are given in Table 13. While creating Table 13, the following details were considered:

• Only studies in which fine aggregate substitution was performed were considered, coarse aggregate substitution was not considered;
• Only granular substitution was used, fiber substitution was not used;
• Only concrete samples were added to the table, mortar samples and self-compacting concrete samples were not added;
• Only PET substituted samples were used and no other type of plastic was used;
• For the compressive strength (CS), tensile strength (TS) and flexural strength (FS) tests, only 28-day data were collected and added to the table;
• If cube samples were used for the compressive and tensile strengths, the cylinder strengths were found by multiplying by 0.8 and written in a separate column in the table;
• Modulus of elasticity (MoE) values of the samples are in the table;
• Substitutions made were divided into two as volume (Vol) and weight (Wei);
• Sample dimensions used during the experiments are given in the table. If sample dimensions are not mentioned and the standard complied with is mentioned, the standard complied with is added to the table;
• Specific gravity and bulk density of PET material are given in the table;
• Sizes of PET material are added to the table;
• PET substitution percentage is given in the table;
• Concrete density values are given in the table. If fresh and dry densities are given together in the study, only dry density is taken;
• Water/cement (W/C) ratio used in the formation of concrete is given in the table;
• The slump test results are in the table.

In the literature, some previous studies on PET substitution in concrete have proposed models for the mechanical properties of concrete. The values in the experimental database have been transformed into graphs based on the models proposed for various properties of concrete. Various equations have been obtained from the graphs through trend lines, and newly developed models have been proposed. The proposed models have been analyzed alongside those in the literature, and their applicability has been assessed.

3.1.1. Compressive Strength Model Development

Table 13 shows the compressive strength values obtained from experimental studies in the literature. To better illustrate the effect of PET substitution on compressive strength, the compressive strengths of PET-substituted samples were divided by the compressive strength of the reference sample. The compressive strength graph was created by using the compressive strength ratios as the vertical axis and the PET substitution percentages as the horizontal axis. This graph is shown in Figure 11. With the help of the trendline in the graph, model suggestions were made for volume substitution, weight substitution, the experimental data from this study, and all data. It was found that more accurate results could be obtained if the trendline were drawn exponentially. In the models proposed in previous studies, linear equations were mostly used. Given this, all results were calculated both linearly and exponentially, and the developed models were proposed. The equations obtained and the developed models are presented in Table 14. In Table 14, $f_{cpet}^{\prime }$ represents the compressive strength of PET-substituted concrete, $f_c^{\prime }$ represents the compressive strength of unsubstituted concrete, and $pet\%$ represents the PET substitution percentage.

The proposed improved models were subjected to applicability control, along with other models suggested in the literature for estimating the compressive strength of PET-substituted concrete. The control was carried out using statistical metrics with the data from the experimental database. The calculated values and models are provided in Table 15. In Table 15, ${\gamma }_{cpet}$ represents the density of PET-substituted concrete.

In some models proposed in the literature, PET substitution was carried out in fiber form instead of granules. In other models, coarse aggregate was substituted instead of fine aggregate. These models are marked and shown in Table 15. Some models used the density ${\gamma }_{cpet}$ value of PET-substituted concrete in their estimations. Since some studies in the dataset did not specify the ${\gamma }_{cpet}$ value, the data points in these models were fewer than in the others. Similarly, since Meena and Ramana [45] proposed a model valid only for samples with up to 15% substitution, the data point in this model was also lower than in the others.

As a result of the metric analyses performed, it was found that the exponential and linear models proposed in this study were the most successful, with R² values of 0.741 and 0.733, respectively. Following these two models, the model by Mohammed [43] came next, with an R² value of 0.716. It was observed that the Saikia and De Brito [17] model had an R² value of 0.256 and provided low accuracy results. Figure 12 shows the graph of the R² and MABE values for the proposed models.

3.1.2. Tensile Strength Model Development

While creating the tensile strength graph of the data, a process similar to that used for the compressive strength graph was followed. A graph was created with one axis representing the tensile strength ratio and the other axis representing the PET substitution percentages. The resulting graph is shown in Figure 13. With the help of the trendline in the graph, model suggestions were made for volume substitution, weight substitution, the experimental data from this study, and all data. The equations obtained and the developed models are provided in Table 16. In Table 16, $f_{tpet}^{\prime }$ represents the tensile strength of PET-substituted concrete, $f_t^{\prime }$ represents the tensile strength of unsubstituted concrete, and $pet\%$ represents the PET substitution percentage.

The proposed improved models were subjected to an applicability check, along with other models in the literature that estimate the connection between PET substitution rate and the tensile strength of concrete. The check was performed using statistical metrics with data from the experimental database. The calculated values and models are provided in Table 17.

As a result of the metric analyses, it was found that among the models that included all tensile strength data, Models I and II proposed by this study were the most successful, with R² values of 0.823 and 0.811, respectively. It was determined that the models of Bamigboye et al. [27] and Bamigboye et al. [44] I had an R² value of 0.255 and provided low accuracy results. Since Meena and Ramana [45] proposed a model valid only for samples with up to 15% substitution, substitutions greater than 15% were not evaluated. Therefore, the data point in this model is lower than the others. Although a successful model was proposed with an R² value of 0.809 for samples with a substitution rate of up to 15%, large errors were observed in the model's predictions as the substitution percentage increased. Figure 14 shows the graph of the R² and MABE values for the proposed models.

3.1.2.1. The Connection Between Tensile Strength and Compressive Strength

There are studies in the literature that examine the connection between the compressive strength and tensile strength of PET-substituted concrete samples and propose model suggestions in this regard. These suggestions were considered in this study, and a graph was created with tensile strength on one axis and compressive strength on the other. While creating this graph, the cube-shaped sample results were converted to the equivalent cylinder strength to avoid differences between compressive and tensile strengths. All results were processed in the graph to reflect the equivalent cylinder sample strength. The graph is shown in Figure 15. Model suggestions were generated using a trendline in the graph. The equations obtained and the models developed are given in Table 18.

The proposed developed models were subjected to an applicability check, along with other models in the literature that predict the connection between tensile strength and compressive strength of PET-substituted concrete. The checks were performed using statistical metrics based on data from the experimental database. The calculated values and models are given in Table 19.

As a result of the metric analyses, it was found that Models II and I proposed in this study were the most successful, with R² values of 0.545 and 0.540, respectively. It was determined that the Tayeh [29] model had an R² value of 0.038 and provided low accuracy results. Figure 16 shows the graph of the R² and MABE values of the proposed models.

3.1.3. Flexural Strength Model Development

When creating the flexural strength graph from the data, a graph was plotted with the flexural strength ratio on one axis and PET substitution percentages on the other. The resulting graph is shown in Figure 17. Model suggestions were made for volume substitution, weight substitution, experimental data from this study, and all data, with the help of a trendline in the graph. The equations obtained and the developed models are given in Table 20. In Table 20, $f_{rpet}^{\prime }$ represents the flexural strength of PET-substituted concrete, $f_r^{\prime }$ represents the flexural strength of unsubstituted concrete, and $pet\%$ represents the PET substitution percentage.

No other model has been found in the literature that estimates the connection between the PET substitution percentage and the flexural strength of concrete. The control was carried out using statistical metrics based on the experimental database. The calculated values are given in Table 20.

3.1.3.1. The Connection Between Flexural Strength and Compressive Strength

Some studies involving PET-substituted concrete samples have suggested a connection between compressive strength and flexural strength and have proposed models for this connection. In this study, these suggestions were considered, and a graph was created with flexural strength on one axis and compressive strength on the other. While creating this graph, the compressive strengths were converted to cylinder sample strength. The resulting graph is shown in Figure 18. Model suggestions were made using a trendline within the graph. The equations obtained and the models developed are given in Table 21.

The proposed developed models were subjected to an applicability check, along with other models in the literature that predict the relationship between the flexural strength and compressive strength of PET-substituted concrete. The checks were carried out using statistical metrics based on the experimental database. The calculated values and models are given in Table 22.

As a result of the metric analysis, it was found that Model I proposed in this study was the most successful, with an R² value of 0.480. It was determined that the Tayeh [29] model had an R² value of 0.280 and provided low accuracy results. Figure 19 shows the graph of the R² and MABE values of the proposed models.

3.1.4. Modulus of Elasticity Model Development

While creating the elasticity modulus graph from the data, a graph was plotted with the elasticity modulus ratio on one axis and PET substitution percentages on the other. The resulting graph is shown in Figure 20. Model suggestions were made for volume substitution, weight substitution, experimental data from this study, and all data, with the help of a trendline in the graph. The equations obtained and the developed models are given in Table 23. In Table 23, $E_{cpet}$ represents the elasticity modulus of PET-substituted concrete, $E_c$ represents the elasticity modulus of unsubstituted concrete, and $pet\%$ represents the PET substitution percentage.

Among the previous studies, only Aocharoen and Chotickai [46] proposed a model that estimates the relationship between the PET substitution rate and the elasticity modulus of concrete. This model was subjected to applicability control together with the model obtained from this study. Control was carried out through statistical metrics using the data in the experimental database. The calculated values ​​and models are given in Table 24.

3.1.4.1. The Connection Between Modulus of Elasticity and Compressive Strength

To examine the connection between compressive strength and modulus of elasticity, a graph was created with compressive strength on one axis and modulus of elasticity on the other. The resulting graph is shown in Figure 21. Model suggestions were made using a trendline in the graph. The equations obtained and the developed models are given in Table 25.

Models estimating the connection between the elasticity modulus and compressive strength of PET-substituted concrete were subjected to applicability checks, along with other models from the literature. The checks were performed using statistical metrics based on an experimental database. The calculated values and models are presented in Table 26.

As a result of the metric analysis, it was found that Model I, proposed in this study, was the most successful, with an R² value of 0.431. The Alqahtani [47] model was found to have an R² value of 0.297, indicating low accuracy. Figure 22 shows the graph of R² and MABE values for the proposed models.

3.2. FRP Wrapping Effect

Previous researchers have considered the ultimate lateral confining pressure ($f_{lu}$) of the FRP material, as defined in Equation 1. In Equation 1, $E_{frp}$ represents the elasticity modulus of the FRP material, $t_{frp}$ the thickness, and ${\epsilon }_{frp}$ the ultimate tensile strain value. D represents the diameter of the concrete sample.

$$f_{lu}={\displaystyle \frac{2E_{frp}{t}_{frp}{\epsilon }_{frp}}{D}}$$

(1)

In later studies, researchers found that the FRP material ruptured at a lower strain value than expected during the experiment. Therefore, they suggested that using the FRP hoop rupture strain (${\epsilon }_{h,rup}$), obtained by multiplying the FRP ultimate tensile strain (${\epsilon }_{frp}$) by the strain reduction factor ($k_{\epsilon }$), would be useful. This value is calculated as shown in Equation 2. The equation for the actual ultimate lateral confining pressure ($f_{lu,a}$), obtained using the FRP hoop rupture strain (${\epsilon }_{h,rup}$), is given in Equation 3.

$${\epsilon }_{h,rup}=k_{\epsilon }{\epsilon }_{frp}$$

(2)

$$f_{lu,a}={\displaystyle \frac{2E_{frp}{t}_{frp}{\epsilon }_{h,rup}}{D}}$$

(3)

In the study, samples with 10% PET substitution were wrapped in CFRP and GFRP, leading to increased compressive strength and strain value. The data from the experimental study, along with sample and FRP properties, are given in Table 27.

There are many model proposals in the literature that estimate the increase in compressive strength and strain values of concrete samples when FRP is used. The predictions of some old and new models were compared with experimental data. A graph was created to compare compressive strength, with one axis representing $f_{lu,a}/f_{co}^{\prime }$ and the other representing $f_{cc}^{\prime }/f_{co}^{\prime }$. The graph is shown in Figure 23.

Even though the experimental data are limited, a model was proposed for the compressive strength of PET-substituted concrete wrapped in FRP, using the trendline on the graph. The estimates from previous models were shown on the graph, and the MABE and MAPE values were calculated and presented in Table 28.

As a result of the metric analysis, it was found that the compressive strength model for PET-substituted concrete proposed in this study was the most successful, with a MABE value of 3.220. In general, all models produced consistent and similar results, and the more recent models were found to be more accurate. The graph showing the MABE and MAPE values of the proposed models is given in Figure 24.

To compare the ultimate axial strain (${\mathsf{\epsilon }}_{cu}$) values, a graph was created with one axis representing $f_{lu,a}/f_{co}^{\prime }$ and the other representing ${\mathsf{\epsilon }}_{cu}/{\epsilon }_{co}$. The graph is shown in Figure 25.

Even though the experimental data are limited, a model was proposed for the ultimate axial strain value of PET-substituted concrete wrapped in FRP, using the trendline on the graph. The estimates from previous models were shown on the graph, and the MABE and MAPE values were calculated and presented in Table 29.

As a result of the statistical analysis, it was found that the ultimate axial strain model for PET-substituted concrete proposed in this study was the most successful, with a MABE value of 0.493. It was observed that the predictions of the Ilki et al. [49] model overestimated the experimental data, while the predictions of the other models were lower. The predictions of the more recent models were found to be more accurate. The graph showing the MABE and MAPE values of the proposed models is given in Figure 26.

4. Conclusions

In this study, the effect of using PET as a fine aggregate substitute in concrete on various concrete properties was investigated. Additionally, the effects of using CFRP and GFRP in PET-substituted concrete were examined. A comprehensive experimental database was created for this study. Using this database, model proposals were made, and the prediction accuracy of previously proposed models was evaluated. As a result of the experiments and analyses, the following results emerged:

• For 30% PET substitution, the slump value decreased by approximately 20%. It was anticipated that further increasing the substitution rate could be disadvantageous for the workability of fresh concrete.
• Increasing the PET substitution rate decreased the density of concrete. It was concluded that PET substitution in concrete could contribute to reducing the weight of the structure or obtaining lightweight concrete if the concrete is used structurally.
• PET substitution in concrete decreased the elastic modulus, compressive strength, splitting tensile strength, and flexural strength, while increasing the ultimate strain values. PET substitution can be used in cases where displacement is required, but it cannot be used to increase strength.
• The decrease in compressive, splitting tensile, and flexural strengths due to PET substitution was compensated for by using FRP, which even exceeded the reference strength. Using PET inside the concrete and FRP outside is an effective method to obtain concrete that is both lighter and has more strength.
• The models proposed in this study were compatible with previous models and experimental data. They were found to provide the most accurate predictions and were applicable based on the experimental database.

The reliability of the proposed models is related to the quantity and quality of the data in the database. In this study, efforts were made to collect data from studies conducted on the subject so far. The presence of studies with vastly different results or insufficient experimental data may undermine the reliability of the models. It is important to note that the models may be revised as the content of the database evolves and the amount of data increases.