Advances in the Study of the Behavior of Full-Depth Reclamation (FDR) with Cement

: Road maintenance and rehabilitation are expected to meet modern society’s demands for sustainable development. Full-depth reclamation with cement as a binder is closely linked to the concept of sustainability. In addition to the environmental beneﬁts of reusing the existing pavement as aggregate, this practice entails signiﬁcant technical and economic advantages. In Spain, in the absence of tests speciﬁcally designed to determine the behavior of recycled pavements stabilized with cement, these materials are treated as soil-cement or cement-bound granular material. This assumption is not entirely accurate, because this recycled pavement contains some bituminous elements that reduce its sti ﬀ ness. This study aimed to obtain the relationships between ﬂexural strength (FS) and the parameters that describe the pavement behavior (long-term unconﬁned compressive strength (UCS) and indirect tensile strength (ITS)) and compare the ﬁndings with the relationships between these parameters in soil-cement and cement-bound granular materials. The results showed that the similar behavior hypothesis is not entirely accurate for recycled pavements stabilized with cement, because they have lower strength values—although, this is not necessarily an indication of poorer performance.


Introduction
Pavement recycling is a road-rehabilitation technique in which a deteriorated pavement is transformed into a new course. Depending on the recycling processes and mixing temperature, pavement recycling technique can be classified as hot recycling (HR) and cold recycling (CR). HR methodology involves two techniques: Hot in-place recycling and hot central-plant recycling. On the other hand, there are three techniques for CR according to the processing place, the construction technology, and the reclamation depth: Cold in-place recycling, cold central-plant recycling, and full-depth reclamation [1,2]. Full depth reclamation (FDR) is a recycling technique in which all the asphalt pavement section and a previously quantified amount of underlying base material are treated. This mixture is pulverized, either mixed with a stabilizing agent or not, and compacted to produce a stabilized base course [3]. The usual depth that is reclaimed varies from 100 to 300 mm [4][5][6]. Sometimes, due to the structural capabilities of the mixture, it is not necessary to add any stabilizing additive, and therefore, the compacted material can be the base for a new surface layer. Nevertheless, if the obtained material does not provide enough structural strength, possible stabilizers are classified as strength and the modulus of elasticity declined with rising bituminous content. At the same time, they concluded that flexural and tensile strength did not fall at low proportions of bituminous mix, but did so very quickly at higher contents.
In 2008, Díaz et al. [26] published a preview of the unconfined compressive strength (UCS), indirect tensile strength (ITS), and flexural strength (FS) results that form part of the first phase of this study. Since then, the number of trials conducted has grown significantly.
This FS test is carried out using prismatic specimens and manufacturing them requires a high level of qualification and experience within the testing team [10,39]. This is the main reason for usually estimating their behavior from standardized tests, such as the unconfined compressive strength and the indirect tensile strength (ITS) tests [31,[40][41][42]. For this reason, the method used in this research is the one proposed by the University of Burgos [33].
This research aimed to fill the void in the understanding of the relationships among flexural strength, unconfined compressive strength, and indirect tensile strength based on the results of the tests conducted. To this end, the methods used for other materials mixed with hydraulic binders [33,34,40,43] and the tests described by Kolias et al. [25] were taken as a starting point. Here, however, the applicable European (EN) or Spanish (UNE or NLT, as appropriate) standards were used to characterize the behavior of an FDR with cement. The accuracy of the initial hypothesis of similarity with soil-cement and cement-bound granular material was also evaluated.

Material
While the number of possible granular material/bituminous material combinations is virtually countless, the proportion consisting of one-third mix asphalt and two-thirds granular material is the one most commonly used in roads [16], and was consequently chosen for this study (10 cm of mix asphalt and 20 cm of granular material), as can be seen in Figure 1. The bituminous layer has approximately 4.5% of bitumen.
Appl. Sci. 2019, 9, x FOR PEER REVIEW  3 of 14 bituminous content. At the same time, they concluded that flexural and tensile strength did not fall at low proportions of bituminous mix, but did so very quickly at higher contents. In 2008, Díaz et al. [26] published a preview of the unconfined compressive strength (UCS), indirect tensile strength (ITS), and flexural strength (FS) results that form part of the first phase of this study. Since then, the number of trials conducted has grown significantly.
This FS test is carried out using prismatic specimens and manufacturing them requires a high level of qualification and experience within the testing team [10,39]. This is the main reason for usually estimating their behavior from standardized tests, such as the unconfined compressive strength and the indirect tensile strength (ITS) tests [31,[40][41][42]. For this reason, the method used in this research is the one proposed by the University of Burgos [33].
This research aimed to fill the void in the understanding of the relationships among flexural strength, unconfined compressive strength, and indirect tensile strength based on the results of the tests conducted. To this end, the methods used for other materials mixed with hydraulic binders [33,34,40,43] and the tests described by Kolias et al. [25] were taken as a starting point. Here, however, the applicable European (EN) or Spanish (UNE or NLT, as appropriate) standards were used to characterize the behavior of an FDR with cement. The accuracy of the initial hypothesis of similarity with soil-cement and cement-bound granular material was also evaluated.

Material
While the number of possible granular material/bituminous material combinations is virtually countless, the proportion consisting of one-third mix asphalt and two-thirds granular material is the one most commonly used in roads [16], and was consequently chosen for this study (10 cm of mix asphalt and 20 cm of granular material), as can be seen in Figure 1. The bituminous layer has approximately 4.5% of bitumen. The granular material used in the laboratory trials was recycled pavement taken from road SA-801 (Peñaranda de Bracamonte to Campo de Peñaranda) from the west of Spain, with a maximum aggregate size of 40 mm. Figure 2 shows the granulometry of the material, which is inside the range of the SC40 (soil-cement with a maximum aggregate size of 40 mm) according to the Spanish standards [28]. It not was necessary to add any aggregate to improve the grading. The recycled material exhibited no plasticity and was free of organic matter and other substances that might prevent the cement setting. The granular material used in the laboratory trials was recycled pavement taken from road SA-801 (Peñaranda de Bracamonte to Campo de Peñaranda) from the west of Spain, with a maximum aggregate size of 40 mm. Figure 2 shows the granulometry of the material, which is inside the range of the SC40 (soil-cement with a maximum aggregate size of 40 mm) according to the Spanish standards [28]. It not was necessary to add any aggregate to improve the grading. The recycled material exhibited no plasticity and was free of organic matter and other substances that might prevent the cement setting. The cement used was ESP VI- 1 32.5 N [44]. This is a widely used cement type for recycled pavements stabilized with cement in roads, because of its low thermal shrinkage and long period workability due to the low quantity of clinker (<50%), high quantity of additives, and moderate strength, mainly short-term [45].
The characteristics of this type of cement are showed in Table 1.

Mix Design
The determination of maximum dry density and optimum moisture content was conducted following the UNE 103-501-94 [46] for cylindrical samples, whose prescriptions are analogous to the ASTM D1557-12 [47]. The density to be achieved in the test specimens was 2.10 g/cm 3 with an optimum modified Proctor moisture content of 7.61% [46] (Figure 3).
Further to the results of the proportioning study, 3.5% ESP VI-1 32.5 N cement [44] was used to ensure a 7-day compressive strength [48,49] of at least 2.5 MPa, the minimum value required by the Spanish Ministry of Public Works [29] and the Council of Castilla y León [50] (Table 2). The cement used was ESP VI- 1 32.5 N [44]. This is a widely used cement type for recycled pavements stabilized with cement in roads, because of its low thermal shrinkage and long period workability due to the low quantity of clinker (<50%), high quantity of additives, and moderate strength, mainly short-term [45].
The characteristics of this type of cement are showed in Table 1. Table 1. Cement ESP VI-1 32.5 N properties [44].

Mix Design
The determination of maximum dry density and optimum moisture content was conducted following the UNE 103-501-94 [46] for cylindrical samples, whose prescriptions are analogous to the ASTM D1557-12 [47]. The density to be achieved in the test specimens was 2.10 g/cm 3 with an optimum modified Proctor moisture content of 7.61% [46] (Figure 3).
Further to the results of the proportioning study, 3.5% ESP VI-1 32.5 N cement [44] was used to ensure a 7-day compressive strength [48,49] of at least 2.5 MPa, the minimum value required by the Spanish Ministry of Public Works [29] and the Council of Castilla y León [50] (Table 2). Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 14

Testing Program
Twenty-four prismatic specimens were prepared for flexural strength testing to characterize the recycled pavement in accordance with standard UNE-EN 12390-5, "Testing hardened concrete. Flexural strength of test specimens" [51], which is analogous to the ASTM D1635/D1635M-12 [52]. The mould dimensions where 15 cm × 15 cm × 60 cm. Samples were stored in a curing room at 20 ± 2 °C and 95% relative humidity [53]. At a curing age of at least 90 days, the four-point flexural beam test was conducted. This method ensures that the specimens break at the weakest section (uniformity of the bending moment between the two points where the load is applied).
The rollers over the specimen were placed at a distance of 15 cm (the height of the specimen), and the rollers bellow the specimen at a distance of 45 cm (three times the height of the specimen).
The applied load was transmitted by means of a plate between the specimen and the rollers over it. An increasing tension of 0.04 MPa was selected in the slowest way of the standard range of 0.04-0.06 MPa/s [51].
After each specimen of 15 × 15 × 60 cm was tested for flexural strength, specimens are broken approximately in the middle. The two resulting halves were also tested without being trimmed, one for the unconfined compressive strength (UCS) test and the other for the indirect tensile strength (ITS) test, to find the relationship between these values and the FS of the initial test specimen.
For simulating the behavior of a cubic sample in the UCS test, an auxiliary metal sheet (15 cm × 15 cm) was introduced between the lower plate and the lower side of the sample (a half from the prismatic sample), and between the top plate and the top side of the sample. This way, a uniform tensile distribution in a 15 cm cube is obtained (Figure 4a). In the case of the ITS test, the load was applied perpendicularly to the axle of the specimen with a modified metal sheet. Hence, the load was applied with a width of 15 cm (Figure 4b).

Testing Program
Twenty-four prismatic specimens were prepared for flexural strength testing to characterize the recycled pavement in accordance with standard UNE-EN 12390-5, "Testing hardened concrete. Flexural strength of test specimens" [51], which is analogous to the ASTM D1635/D1635M-12 [52]. The mould dimensions where 15 cm × 15 cm × 60 cm. Samples were stored in a curing room at 20 ± 2 • C and 95% relative humidity [53]. At a curing age of at least 90 days, the four-point flexural beam test was conducted. This method ensures that the specimens break at the weakest section (uniformity of the bending moment between the two points where the load is applied).
The rollers over the specimen were placed at a distance of 15 cm (the height of the specimen), and the rollers bellow the specimen at a distance of 45 cm (three times the height of the specimen).
The applied load was transmitted by means of a plate between the specimen and the rollers over it. An increasing tension of 0.04 MPa was selected in the slowest way of the standard range of 0.04-0.06 MPa/s [51].
After each specimen of 15 × 15 × 60 cm was tested for flexural strength, specimens are broken approximately in the middle. The two resulting halves were also tested without being trimmed, one for the unconfined compressive strength (UCS) test and the other for the indirect tensile strength (ITS) test, to find the relationship between these values and the FS of the initial test specimen.
For simulating the behavior of a cubic sample in the UCS test, an auxiliary metal sheet (15 cm × 15 cm) was introduced between the lower plate and the lower side of the sample (a half from the prismatic sample), and between the top plate and the top side of the sample. This way, a uniform tensile distribution in a 15 cm cube is obtained (Figure 4a). In the case of the ITS test, the load was applied perpendicularly to the axle of the specimen with a modified metal sheet. Hence, the load was applied with a width of 15 cm (Figure 4b).  UCS tests were conducted following the standard UNE-EN 13286-41 [49], with a load speed in the range interval of 0.1 ± 0.1 MPa/s [54]. ITS strength tests were performed in accordance with UNE-EN 12390-6, "Testing hardened concrete. Tensile splitting strength of test specimens" [55].

Results and Discussion
Obtained results from the 72 tests conducted on the 24 prismatic specimens are shown in Table 3.  UCS tests were conducted following the standard UNE-EN 13286-41 [49], with a load speed in the range interval of 0.1 ± 0.1 MPa/s [54]. ITS strength tests were performed in accordance with UNE-EN 12390-6, "Testing hardened concrete. Tensile splitting strength of test specimens" [55].

Results and Discussion
Obtained results from the 72 tests conducted on the 24 prismatic specimens are shown in Table 3.

Relationship Between Flexural and Unconfined Compressive Strength
The correlation between the values of the UCS at long-term and FS at long-term is shown in Figure 5.

Relationship Between Flexural and Unconfined Compressive Strength
The correlation between the values of the UCS at long-term and FS at long-term is shown in Figure 5. After examining various possible functions for correlating these two variables, the best correlation was obtained with an S shape function, with natural logarithm of the FS as dependent variable and 1/UCS as the independent variable. The developed relationship is shown in Equation (1).
Ln(FSLT-UCS)= 0.33 − 0.3225/UCSLT (1) where FSLT-UCS is the estimated value of the flexural strength at long-term by means of UCSLT, and UCSLT is the unconfined compressive strength at long-term, both expressed in MPa.
The coefficient of determination (R 2 ) has a value of 0.629, which indicates that the model can explain more than the 62% of the variability of the model.
The average values obtained in the tests for these two parameters for the recycled material were compared to the usual values for soil-cement and cement-bound granular material [26,42,45,[56][57][58] in Table 4. As seen in Table 4, while the values obtained in the analysis were lower than soil-cement strength due to the bituminous matrix, the relationship between the two parameters was closer to the cement-bound granular material.

Relationship Between Flexural Strength at Long-Term and Indirect Tensile Strength at Long-Term
The correlation between FS and ITS values at long-term is shown in Figure 6, indicating a linear relationship between these parameters. After examining various possible functions for correlating these two variables, the best correlation was obtained with an S shape function, with natural logarithm of the FS as dependent variable and 1/UCS as the independent variable. The developed relationship is shown in Equation (1).

Ln(FS LT-UCS
where FS LT-UCS is the estimated value of the flexural strength at long-term by means of UCS LT , and UCS LT is the unconfined compressive strength at long-term, both expressed in MPa. The coefficient of determination (R 2 ) has a value of 0.629, which indicates that the model can explain more than the 62% of the variability of the model.
The average values obtained in the tests for these two parameters for the recycled material were compared to the usual values for soil-cement and cement-bound granular material [26,42,45,[56][57][58] in Table 4. Table 4. Comparison between average UCS values at long-term and flexural strength (FS) values at long-term for soil-cement, cement-bound granular material, and obtained values for the full-depth reclamation (FDR) with cement of the study. As seen in Table 4, while the values obtained in the analysis were lower than soil-cement strength due to the bituminous matrix, the relationship between the two parameters was closer to the cement-bound granular material.

Relationship Between Flexural Strength at Long-Term and Indirect Tensile Strength at Long-Term
The correlation between FS and ITS values at long-term is shown in Figure 6, indicating a linear relationship between these parameters. Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 14 A statistical analysis was performed and it was observed that the best relationship was obtained by means of a simple linear regression, expressed in Equation (2).
FSLT-ITS = 0.187 + 1.063 ITSLT (2) where FSLT-ITS is the estimated value of the flexural strength at long-term obtained by means of ITSLT, and ITSLT is the indirect tensile strength at long-term, both in MPa.
The regression has a R 2 value of 0.65 and the F of Fisher-Snedecor test indicated that the relationship was true with a significance level over 99%. The Student's t-tests for the coefficients indicated that they were true, different from 0 with a significance level over 99%.
Once again, the average values obtained in the tests for these two parameters for the recycled material were compared to the usual values for soil-cement and cement-bound granular material [45] in Table 5. Although a direct relationship between ITS and FS is not established for cement-bound granular materials, it is indicated that the UCS value is approximately 10 times the ITS value [26,42,45,[56][57][58]. This assumption is adopted for the analysis in Table 5. Table 5. Comparison between averages of ITS values at long-term and FS values at long-term for soil-cement, cement-bound granular material, and obtained values for the FDR with cement of the study. It is observed that the ITS of the recycled material was similar to the value specified for soil-cement, while the relationship between ITS and FS was closer to a cement-bound granular material.

Relationship Between Indirect Tensile Strength and Unconfined Compressive Strength at Long-Term
The values of these two parameters (UCS and ITS) are compared in Figure 7. A statistical analysis was performed and it was observed that the best relationship was obtained by means of a simple linear regression, expressed in Equation (2).
where FS LT-ITS is the estimated value of the flexural strength at long-term obtained by means of ITS LT , and ITS LT is the indirect tensile strength at long-term, both in MPa. The regression has a R 2 value of 0.65 and the F of Fisher-Snedecor test indicated that the relationship was true with a significance level over 99%. The Student's t-tests for the coefficients indicated that they were true, different from 0 with a significance level over 99%.
Once again, the average values obtained in the tests for these two parameters for the recycled material were compared to the usual values for soil-cement and cement-bound granular material [45] in Table 5. Although a direct relationship between ITS and FS is not established for cement-bound granular materials, it is indicated that the UCS value is approximately 10 times the ITS value [26,42,45,[56][57][58]. This assumption is adopted for the analysis in Table 5. Table 5. Comparison between averages of ITS values at long-term and FS values at long-term for soil-cement, cement-bound granular material, and obtained values for the FDR with cement of the study. It is observed that the ITS of the recycled material was similar to the value specified for soil-cement, while the relationship between ITS and FS was closer to a cement-bound granular material.

Relationship Between Indirect Tensile Strength and Unconfined Compressive Strength at Long-Term
The values of these two parameters (UCS and ITS) are compared in Figure 7. The correlation between both parameters was statistically analyzed and a linear correlation was proposed, as shown in Equation (3).
where ITSLT and UCSLT are as defined in Equations (1) and (2), respectively, both in MPa.
Equation (3) omitted the intercept because the p-value of the Student's t-test was over 0.99, indicating that it was not significant. The relationship has an R 2 value of 0.49. The F test indicated that the relationship was true with a significance level over 99%.
The relationship between these two parameters at long-term obtained for the recycled material and the usual values for cement-treated base courses [45] were found to be similar (Table 6). Table 6. Comparison of the relationship between unconfined compressive strength and indirect tensile strength at long-term for soil-cement, cement-bound granular material, and obtained values for the FDR with cement of the study.

Estimation of Flexural Strength at Long-Term Using the UCS and ITS Values
An additional equation for estimating the flexural strength at long-term of the FDR with cement was developed as a function of the unconfined compressive strength and the indirect tensile strength by means of a multiple linear regression, as shown in Equation (4).
FSLT-2 = 0.074 UCSLT + 0.826 ITSLT (4) where FSLT-2 is the flexural strength at long-term by means of UCSLT and ITSLT simultaneously, and UCSLT and STSLT are as defined in Equations (1) and (2), respectively. Equation (4) has a coefficient of determination (R 2 ) of 0.684. Including an intercept in Equation (4) made the coefficients of the intercept and UCS not significant. Without the intercept, both coefficients are different from 0, with a significance level over 99% (p-value of the Student's t-test >0.99). Figure 8 shows the obtained values of FS and the values estimated by Equations (1), (2), and (4). The correlation between both parameters was statistically analyzed and a linear correlation was proposed, as shown in Equation (3).
where ITS LT and UCS LT are as defined in Equations (1) and (2), respectively, both in MPa. Equation (3) omitted the intercept because the p-value of the Student's t-test was over 0.99, indicating that it was not significant. The relationship has an R 2 value of 0.49. The F test indicated that the relationship was true with a significance level over 99%.
The relationship between these two parameters at long-term obtained for the recycled material and the usual values for cement-treated base courses [45] were found to be similar (Table 6). Table 6. Comparison of the relationship between unconfined compressive strength and indirect tensile strength at long-term for soil-cement, cement-bound granular material, and obtained values for the FDR with cement of the study.

UCS LT /ITS
where FS LT-2 is the flexural strength at long-term by means of UCS LT and ITS LT simultaneously, and UCS LT and STS LT are as defined in Equations (1) and (2), respectively. Equation (4) has a coefficient of determination (R 2 ) of 0.684. Including an intercept in Equation (4) made the coefficients of the intercept and UCS not significant. Without the intercept, both coefficients are different from 0, with a significance level over 99% (p-value of the Student's t-test >0.99). From the point of view of sustainability, the advantages of FDR when compared with soil-cement and cement-bound granular mixture are considerable. When manufacturing FDR, it is avoided to transport -emove material to landfills; there is no need to use quarries and the quantity of material that must be transported is lower and, hence, CO2 emissions are reduced. Moreover, the roads that are used for transporting the material are not so damaged.
With regard to the (expected) behavior, it can be said that the average UCS and ITS value at long-term are similar to soil-cement. In the case of the FS at long-term, the value is lower than usual for soil-cement. The fact that the FS values are lower could be regarded as a disadvantage, and perhaps the expected life of the pavement structure would not be as long as with soil-cement. However, if we compare the expected life of the new higher quality base that we are designing with the previous pavement structure, which was composed of unbound aggregates, an improvement is observed. The quality is not as high as with soil-cement, but it must be taken into account that there is a big increase in the quality of the new base compared to the previous one. With this technique, a material that is near to a standardized material is designed, which is cheaper and more sustainable. Consequently, the advantages overcome the disadvantages.

Conclusions
The study aimed to establish the long-term relationships among flexural, unconfined compressive, and indirect tensile strength in FDR with cement, and compared them to the strength relationships between soil-cement and cement-bound granular materials to verify the hypothesis that their behavior was similar. From the point of view of sustainability, the advantages of FDR when compared with soil-cement and cement-bound granular mixture are considerable. When manufacturing FDR, it is avoided to transport -emove material to landfills; there is no need to use quarries and the quantity of material that must be transported is lower and, hence, CO 2 emissions are reduced. Moreover, the roads that are used for transporting the material are not so damaged.
With regard to the (expected) behavior, it can be said that the average UCS and ITS value at long-term are similar to soil-cement. In the case of the FS at long-term, the value is lower than usual for soil-cement. The fact that the FS values are lower could be regarded as a disadvantage, and perhaps the expected life of the pavement structure would not be as long as with soil-cement. However, if we compare the expected life of the new higher quality base that we are designing with the previous pavement structure, which was composed of unbound aggregates, an improvement is observed. The quality is not as high as with soil-cement, but it must be taken into account that there is a big increase in the quality of the new base compared to the previous one. With this technique, a material that is near to a standardized material is designed, which is cheaper and more sustainable. Consequently, the advantages overcome the disadvantages.

Conclusions
The study aimed to establish the long-term relationships among flexural, unconfined compressive, and indirect tensile strength in FDR with cement, and compared them to the strength relationships between soil-cement and cement-bound granular materials to verify the hypothesis that their behavior was similar.
The statistical analysis proved the existence of fairly close relationships among these three strength tests in the FDR, but with different behaviors to what it was expected. Flexural strength exhibited lower values in the recycled pavements than in soil-cements, whereas the indirect tensile strength and unconfined compressive strength values were similar. The relationships between unconfined compressive strength and flexural strength, and between indirect tensile strength and flexural strength, were even closer than in cement-bound granular material. In the analyzed recycled material, only the relationship between unconfined compressive strength and indirect tensile strength was similar to the relationship in cement-bound granular material and soil-cement.
With the research, in the case that only the unconfined compressive strength value is available, Equation (1) is recommended to calculate the flexural strength at long-term of the FDR. If only the indirect tensile strength is known, Equation (2) is then recommended to calculate the flexural strength of the FDR. If we have both the unconfined compressive strength and indirect tensile strength at long-term, Equation (4) is proposed to estimate the flexural strength of the FDR.
It is important to know the flexural strength, because the fatigue strength of the FDR material is calculated using this value. The hypothesis that the FDR with cement, soil-cement, and cement-bound granular material exhibit similar behaviors is not accurate and, therefore, there is a need to undertake a fatigue behavior study on this type of recycled base course to ensure the optimum design of this type of pavements.