Wheel-tracking performance of recycled concrete aggregate with recycled glass and brick in unbound pavements under elevated loads

As natural quarry materials become increasingly scarce and uneconomical, the construction industry has turned to sustainable alternatives such as construction and demolition (C&D) wastes and recycled glass for road construction. The aim of this study was to evaluate the performance of mixtures consisting of recycled glass (RG), crushed brick (CB) and recycled concrete aggregate (RCA) under varied traffic conditions. This assessment was conducted through wheel-tracking (WT) tests under simulated high-traffic conditions, which involved subjecting the mixtures to elevated vertical loads and an increased number of load cycles compared to previous studies. The study revealed that both RCA + 20%RG and RCA + 20%CB blends displayed comparable or slightly greater mean surface deformations than natural crushed rock under default conditions. The default conditions specified by the local road authority include an 8 kN wheel load and 40,000 loading cycles. The study further revealed that both blends displayed a consistent increase in rut depth as the number of cycles increased up to 100,000 while being subjected to a 20 kN wheel load. The maximum rut depth of RCA + 20%RG was close to the lower end of the maximum allowable rut depth range specified by road authorities. This suggests that these blends are at the limits of carrying heavier loads on highly trafficked roads.


Introduction
The construction industry is experiencing high demand for natural quarry materials as the infrastructure continues to grow worldwide.In certain cases, using natural quarry materials in road construction can be uneconomical and unsustainable.Concurrently, solid waste disposal has become a significant problem due to rapid urban development and increased economic activities.To address these issues, the use of construction and demolition (C&D) waste in road construction is a sustainable and cost-effective solution.In current practice, greater quality virgin quarry material is used for roads carrying heavy traffic loads and roads which expect better performance.This study evaluates the usability of blends of recycled materials that include recycled glass (RG), crushed brick (CB), and recycled concrete aggregate (RCA) in high load environments with heavy traffic.
In recent times, international research has focused on reusing C&D waste materials in pavement engineering applications to conserve virgin quarry reserves and reduce the carbon footprint of projects.Several studies have been carried out to evaluate the performance of unbound RCA + RG and RCA + CB blends in pavements using either wheel-tracking (WT) tests or repeated load triaxial (RLT) tests.In a research project, laboratory tests and field trials confirmed that a blend of RCA and 20% RG has optimum level strength and workability in pavement subbases [1].A research demonstrated that up to 25% CB could be added to RCA and crushed rock (CR) blends to work satisfactorily in unbound sub-base pavement applications.The RLT testing revealed that CB would perform adequately at a 65% moisture ratio level.Shear strength of CB was reduced from the acceptable limits at higher moisture ratios [2].Another research showed that up to 30% of RG fines can be added to lower subbase CR mixes without detrimental effect on the performance [3].In a research, direct shear test results revealed that shear strength of RG is similar to the shear strength of angular natural sands and RG has similar geotechnical engineering behavior to a well graded sand mix [4].A study showed that fine and medium grained recycled glass had a generally lower Maximum Dry Density (MDD) by 10% to 15% but had a higher Optimum Moisture Content (OMC) in comparison to natural aggregates with the same soil classification [5].The specific gravity of RG is roughly 10% lower than that of natural sand mixes [4].The lightweight nature of RG indicates that it is unsuited for higher loading pavement bases/ subbases which need higher CBR values [6].A previous research found that RCA + CB blends had better results in repeated load triaxial (RLT) tests compared to crushed rock and CB blends and performed satisfactorily as a subbase material [7].
Rutting is regarded as the main cause of flexible pavement structural failures and is considered a crucial metric for evaluating a pavement's structural performance.Assessing the resistance to rutting of unbound granular base materials is crucial in Australia, where over 80% of roads consist of such pavements, covered only by a sprayed seal or a thin asphalt layer [8].Various laboratory WT devices are utilized to conduct simulated tests and make predictions regarding rutting, fatigue, moisture susceptibility, and stripping.Among the commonly employed laboratory wheel-tracking devices, the Asphalt Pavement Analyzer (APA), Hamburg Wheel Tracking (HWT) device, and French Rutting Tester (FRT) are particularly popular.APA has the capability to simulate field traffic and temperature conditions.It offers the flexibility to conduct tests on both beam and cylindrical specimens, allowing for controlled variations in temperature and the option to test under dry or soaked conditions [9].The HWT device is utilized to assess the combined impacts of rutting and stripping on highly trafficked pavements.This is achieved by immersing a test specimen in a water bath that is controlled for temperature [10].The HWT test is a commonly employed laboratory test that has consistently demonstrated its effectiveness in identifying and evaluating hot-mix asphalt mixes that are susceptible to rutting and/or vulnerable to moisture damage [11].
The HWT test shows great potential in evaluating the impact of repeated loading and moisture-induced damage on asphalt mixtures under high temperature conditions [12].The FRT is specifically designed to assess the ability of bituminous mixtures to withstand rutting.Initially intended for qualifying hot mix asphalts composed solely of pure bitumen, this test has evolved to evaluate the rutting resistance of warm, half-warm, and cold mix asphalts [13].Most of the WT devices commonly have smaller slab sizes and wheels, primarily used for assessing the rutting resistance of bituminous mixtures.The Austrack WT device, which was used for this study, stands out due to its larger dimensions.It utilizes a real-sized wheel and the slab employed in the device has dimensions of 500 mm (width) × 700 mm (length) × 300 mm (depth), which is significantly thicker than the slabs used in other WT devices.The local road authority has made modifications to the WT device in order to evaluate the rutting resistance of both bituminous mixtures and granular pavement materials.The Austrack WT device conducts tests under controlled conditions, ensuring consistent factors such as test temperature, moisture content, tire pressure, rolling load, steady wheel load, and frequency of wheel travel [14].
Accelerated pavement testing (APT) is an exceptional tool for efficiently testing and validating the adoption of novel and enhanced pavement solutions.It can be defined as the deliberate application of wheel loading on pavement structures, simulating the long-term effects of actual in-service loading conditions within a condensed timeframe.APT facilities utilize linear or circular loading systems to apply loads on relatively short sections of pavement, typically constructed within a test pit of various designs such as Accelerated Loading Facility (ALF), Heavy Vehicle Simulator and Canterbury Accelerated Pavement Testing Facility.The concept of a 'test road' involves subjecting a specific pavement section to controlled traffic, often utilizing standard trucks that adhere to closely regulated load and speed conditions.The examples for the 'test road' are National Center for Asphalt Technology at Auburn University and MnRoad test tracks located in Monticello, USA [15].In a research study, the test results of HWT device, FRT, and APA for hot mix asphalt were compared against the performance of an ALF known as WesTrack.The findings of the study revealed that the performance outcomes of the three laboratory devices aligned reasonably well with the permanent deformation observed in WesTrack [16].Compared to WT tests, APT produces exceedingly reliable results, but it comes with a significant cost.
The rut resistance of the granular base materials was assessed through accelerated loading tests and compared with results from several laboratory characterization tests conducted by the local road authority, including WT, RLT, and shear strength tests.The WT test simulates the impact of a rolling wheel on a material specimen in a laboratory setting, providing a more realistic representation of in-service loading conditions.The earlier findings indicated that the WT test was the best available and economical test method for evaluating the rut resistance of granular bases and determining their ranking.The comparison between laboratory and field results showed that the WT test more accurately reflected the impact of moisture sensitivity on performance, compared to the RLT test [17].
Both gross vehicle mass and axle group load overloading can cause significant damage to pavements.To avoid excessive road deterioration caused by overloaded vehicles, many countries have established limits for both axle load and gross mass.According to a research, the rate of road wear is found to increase in accordance with a power law as vehicle loading increases [18].A study carried out on Taiwan's freeway systems revealed that the average load factor for combined heavy vehicles, as calculated from the data collected by weigh-in-motion equipment, was 2.7 times greater than the original design value [19].
This study aims to compare and evaluate the performance of RCA + 20%RG and RCA + 20%CB blends under heavy traffic and high load environments.To assess the performance of these blends under more demanding conditions, the study used WT tests with a higher vertical wheel load and increased number of load cycles compared to the default wheel load used by local road authorities.The results of these tests will help in determining the effectiveness of these blends in handling heavy traffic and high loads.

Materials and methods
Based on the earlier research outlined in the introduction, it was determined that the blends containing 20%RG and 20%CB with RCA performed satisfactorily in upper subbase pavement layers under default loading conditions.As a result, this study utilized four blends of RCA + 20%RG and RCA + 20%CB, sourced from a local recycling facility.Figure 1 shows photos of the RG, CB and RCA used in the current study.The individual components of the blends were crushed separately, and then uniformly mixed using a pugmill [20].The mix ratios of each material were calculated based on their dry masses.To assess the presence of foreign materials in the RCA, testing was conducted on each crushed material using the test method specified by the local road authority [21], which determined the content of high and low density materials, as well as wood and vegetable matter.
Particle size distribution (PSD) testing was performed for all mixes prior to and after WT testing [22].Using modified compaction tests, the maximum dry density (MDD) and optimum moisture content (OMC) of the blends were determined.The samples were compacted in a 105 mm diameter metal mold, in 5 layers with an average height of 120 mm, at varying moisture contents.The weight of the samples was recorded before and after drying to determine their moisture content.The resulting dry density and moisture content data for each sample were graphed to obtain the OMC and MDD for each mixture [23,24].Atterberg Limits tests were carried out to determine the liquid limit (LL), plasticity index (PI) and linear shrinkage (LS) of the finer fraction of the mixes [25][26][27][28].To assess the resistance of materials to abrasion, Los Angeles (LA) abrasion tests were conducted in accordance with ASTM test method [29].In order to evaluate the soil strength capabilities, the mixtures were subjected to California Bearing Ratio (CBR) tests.These tests were conducted at 98% MDD and 100% OMC, which were determined in the modified compaction testing [30].All blends were compacted in molds measuring 152 mm in diameter and 117 mm in height at 100% MDD and 98% OMC, as specified by the Australian standards.The compaction process for each mixture involved 56 weighted blows, with 5 separate layers being compacted individually.Following compaction, the samples were weighed and subsequently submerged in water for a duration of four days prior to testing.
The testing device utilized in the study was the Austrack WT machine, which was developed by the local road authority.The test method employed in the study was also formulated by the same local road authority [14].During the test, a slab of material is subjected to multiple cycles of rolling passes from a loaded wheel, causing rutting of the material under the wheel.The depth of the rut is measured as a function of the number of cycles, and the resulting rut depth provides an indication of the material's resistance to rutting for a given number of cycles [31].The WT samples were compacted to reach 98% of its MDD and 80% of its OMC.Specimens of 300 mm thickness were formed by compacting the sample into six 50 mm thick layers using a segmented roller and load steps of 5 kN, 10 kN, 20 kN, and 30 kN.The layers were air dried at 38 °C until they reached approximately 70% of their OMC, and then cured for 4 h.Upon completion of the dry back process, an epoxy sealing product was applied to the specimen's surface to prevent further moisture loss during the WT test.The WT tests were conducted using a full-scale wheel with a tire pressure of 600 kPa, applying 8 kN and 20 kN axial loads on CC + 20%RG and CC + 20%CB samples, in accordance with the local road authority test method [14].The wheel was travelling at the frequency of 0.4 Hz and had a diameter of 550 mm and width of 110 mm.Both material blends underwent 40,000 and 100,000 loading cycles and no sample was failed during the tests.The local road authority has established the default conditions for WT testing as 8 kN vertical wheel load and 40,000 loading cycles [14].A wheel load of 20 kN was applied to assess the performance of both blends under more demanding loading conditions.To gain greater confidence in the blends and assess their resilience, the concept of implementing 100,000 loading cycles has been introduced.

Results and discussion
The particle size distributions of RCA + 20%RG and RCA + 20%CB blends prior to and after performing WT tests were plotted with the upper and lower envelope of local road authority's upper subbase material specification and all gradings were illustrated in Fig. 2(a) and (b).Upon conducting WT tests, it was noticed that all blends displayed a remarkably similar distribution of particles, indicating a negligible amount of particle breakdown.However, the RCA + 20%RG blends demonstrated some levels of breakdown in the finer sieve sizes (0.075 and 0.15 mm) after WT tests.Increasing the percentage of RG in the mixes may lead to a greater degree of particle breakdown and could potentially result in noncompliance with the PSD specification set by the local road authority.The PSD envelope specified for the local road authority's upper subbase material, composed of softer material with an LA value higher than 25, has a limited range at the finer end of the curve.This serves as a restriction on the possibility of further breakdown of the material during compaction [32].However, the particle size distributions of all blends in the study remained within the local road authority specified limits for upper subbase material, even after subjecting them to increased vertical load up to 20 kN and loading cycles up to 100,000.
The engineering characteristics of RCA + 20%RG and RCA + 20%CB blends are presented in Table 1, indicating their adherence to the local road authority specification requirements.Although both blends met the CBR requirements specified by local road authority for upper subbase material, the CBR results were lower for RCA + 20%RG compared to the other blend.LA test result revealed that RCA + 20%RG has greater hardness than RCA + 20%CB.The RCA + 20% CB blend was found to have 13.5% high density foreign material content, as the testing considered brick to be a foreign substance.However, the use of brick in subbase material is allowed within local road authority specified supplementary material limits [32].In general, blends of RG, CB and RCA exhibit non-plasticity or very low values of PI and LL.Typically, mixes incorporating RCA tend to exhibit a higher OMC in contrast to mixes containing crushed rock.This is mainly due to the superior water absorption capacity of RCA.The comparison between RG mixes and CB mixes reveals that CB mixes demonstrate a higher OMC, attributed to the higher water absorption capability of CB particles.
Figure 3(a) and (b) depict the stages of preparing the WT test samples for RCA + 20% RG, while Fig. 3(c) and (d) show the rut depth and condition of the test samples after testing.An automated laser scanning device records the information on the deformation caused by the subsequent load applied through the WT test on the samples.This data is captured utilizing 405 laser point readings at the end of each loading cycle across five predefined profiles to quantify the transverse deformation.The resulting rutting depths were measured at various specified cycles, providing a comprehensive evaluation of the material's ability to resist rutting.Figure 4 illustrates the rut depth measuring points and appearance of the wheel path of RCA + 20%RG sample after 100,000 loading cycles.[17].After 30,000 cycles, both RCA + 20%RG and RCA + 20%CB blends exhibited consistent deformation as the number of cycles increased under the default load of 8kN.As a result, the WT tests carried out under 8kN were concluded after reaching 40,000 cycles.Even with a 20 kN vertical load, MSD of both blends stayed within acceptable limits.Despite an increase in loading cycles to 100,000, there was no significant overall increase in MSD observed for RCA + 20%CB after 40,000 cycles.In contrast, the RCA + 20%RG blend exhibited a slight and gradual increase in MSD throughout the test until it was terminated at 100,000 cycles.Figure 6 illustrates the maximum rut depth (MRD) for RCA + 20%RG and RCA + 20%CB blends as a function of the number of loading cycles.When subjected to an 8 kN  vertical load, the MRD of both blends followed a similar pattern to MSD.At 40,000 cycles, RCA + 20%CB exhibited a MRD of 5.6 mm, which was higher compared to RCA + 20%RG.But with an increased vertical load of 20 kN and 100,000 cycles, RCA + 20%RG produced a higher MRD than RCA + 20%CB, reaching a value of 21.4 mm. Figure 7 illustrates the vertical deformations of the RCA + 20%RG and RCA + 20%CB blends across transversal positions of wheel path under 8 kN and 20 kN loads, over 40,000 and 100,000 cycles.The maximum surface deformations vary with the longitudinal locations (0 mm, ± 75 mm, ± 150 mm as shown in Fig. 4).The vertical surface deformations shown in Fig. 7 are the average of longitudinal values.It is important to note that deformation only represents the downward movement of the surface, resulting in values that are typically lower than the corresponding rut depth.The larger the load induced upward movement of the surface, the larger the discrepancy between deformation and rut depth values [17].Figure 7 illustrates the extent of deformation along the wheel path, measuring 400 mm.The figure prominently displays the maximum rut depth observed in the RCA + 20RG mix under a 20 kN load after 100,000 cycles.Figure 8 illustrates the average vertical surface deformation of cross sections across the wheel path.The deformations were measured under a 20 kN vertical load and recorded at various cycle intervals, ranging from 0 cycles to 100,000 cycles.Subfigure (a) represents the RCA + 20%RG mix, while subfigure (b) represents the RCA + 20%CB mix.For both mixes, the sequence of the cross sections revealed a notable pattern of deformation.Initially, from 0 to 5000 cycles, there was a rapid increase in deformation.Subsequently, the deformation continued to escalate steadily up to 40,000 cycles.However, beyond this point, the rate of deformation exhibited a lesser phase of increase, which persisted up to 100,000 cycles.
The RCA + 20%RG blend clearly exhibited higher heaving/rutting at 20 kN vertical loads under both 40,000 and 100,000 load cycles.Previous research showed a consistent rise in the deformation/rut depth as the moisture content of the materials increased [17].In this study, the dry back percentage of RCA + 20%RG was lower than that of RCA + 20%CB, and moisture was not the cause of its reduced performance.RG is a non-cohesive and unbound material, and its smooth and flat surfaces prevented robust bonding between RG and RCA aggregates [33].As the load and number of cycles increased, plastic movement became evident, primarily due to a lack of shear strength, particularly in the RCA + 20%RG blend.This deficiency in shear strength is the primary factor contributing to the escalation of permanent deformation.The RCA + 20%RG blend exhibited lower CBR compared with RCA + 20%CB and after WT test, some particle breakdown was observed at the finer end of the Under a 20 kN wheel load, the rut depth of both blends consistently increased with the number of cycles without showing a flattened curve.This implies that as the load increases, the rutting will also continue to increase with the number of cycles.In the context of flexible pavements, the local road authority specifications do not include a limit or threshold for the maximum allowable rut depth.However, AASHTO (1993) does provide a range of allowable rut depths for aggregate surfaced roads, which varies between 25 mm to 50 mm depending on the average daily traffic [34].
During the testing of base/subbase layers using a smallscale WT setup designed for asphalt testing, several issues were faced.These included the absence of overburden including asphalt sealing due to height restrictions in the testing chamber, difficulties in achieving the desired compaction and density and suboptimal compaction due to the use of a segmented roller that pushed the material towards the sides.Specially, the WT tests were carried out by enclosing the specimen in a steel mould.This enclosure effectively prevents any lateral deformation, which is known to be the main factor contributing to severe rutting.

Conclusions
The study found that as the vertical loads and cycle numbers were raised in WT testing, both the RCA + 20%RG and RCA + 20%CB blends exhibited comparable average surface deformations to those reported in earlier research conducted under standard conditions of local road authorities, with an 8 kN wheel load and 40,000 loading cycles.Although the RCA + 20%CB blends exhibited slightly higher overall mean surface deformations, they still fell within acceptable thresholds when compared to natural crushed rock performance in prior research.There was no significant increase in overall mean surface deformation observed for RCA + 20%CB after increasing vertical load from 8 to 20 kN and cycle numbers from 40,000 to 100,000, whereas the RCA + 20%RG blend demonstrated a gradual slight increase in mean surface deformation throughout the test until it was completed.In RCA + 20%RG blends, the RG fraction continues to break down into smaller particles under higher vertical loads and increased cycle numbers.However, it seems that after 40,000 cycles, most of the CB particles in RCA + 20%CB blends cease to further break down into smaller particles.As a result, RCA + 20%CB blends demonstrate better performance confidence than RCA + 20%RG blends.The study employed -5 mm RG in the blends, but incorporating finer fractions of RG, smaller than 5 mm, would yield better resilience.
The study also revealed that both blends consistently exhibited an increase in rut depth with an increase in the number of cycles when subjected to a 20 kN wheel load.At the end of the test, the maximum rut depth of RCA + 20%RG was close to the lower end of the maximum allowable rut depth range set by AASHTO (1993) for roads.This implies that the transportation of heavier loads on highly busy roads has become challenging for these blends, as they have reached their capacity limits.
As the research was limited in the number of tests conducted, further assessment of both blends, such as field trials, may be necessary to determine their suitability for use on highly trafficked roads with higher loads.

Fig. 3
Fig. 3 WT testing stages.(a) Surface of the slab after compaction; (b) Sealed surface of the slab prior to WT; (c) Rut depth indication after 100,000 cycles; (d) Side view of specimen when un-molding

Fig. 5
Fig. 5 Number of WT loading cycles vs mean surface deformation

Fig. 6
Fig. 6 Number of WT loading cycles vs maximum rut depth

Fig. 7 6 Fig. 8
Fig. 7 WT transversal position vs vertical surface deformation at 40,000 cycles and at 100,000 cycles

Table 1
Engineering properties of the blends

Table 2
displays the parameters of the WT test utilized in the study and the outcome of the test.Figure5provides mean surface deformation (MSD) of RCA + 20%RG and RCA + 20%CB blends under 8 and 20 kN against the number of loading cycles.Under an 8 kN vertical load, both blends showed similar MSD up to 20,000 cycles.Subsequently, the RCA + 20%CB blend showed increased deformation, reaching 2.8 mm at 40,000 cycles.When increasing the vertical load to 20 kN, at 40,000 loading cycles, MSD of RCA + 20%CB was 6.4 mm and it was higher than RCA + 20%RG.A rise in load from 8 to 20 kN led to a greater than two-fold increase in surface deformation.When comparing the results of WT tests conducted by local road authority on hornfels and granite unbound materials under 8 kN vertical load and 40,000 cycles, with similar moisture and density conditions, both RCA + 20%RG and RCA + 20%CB blends exhibited a similar or lower MSD

Table 2
WT test parameters and results a Overall mean surface deformation b Overall maximum rut depth c Test not performed