1. Introduction
In highway infrastructures, quality control of random embankments is carried out using tests that cannot correctly evaluate the compaction process [
1]. Sopeña [
2] indicated that topographic control has no reference values.
Zhong et al. [
3] developed an automatic process for the monitoring of compaction parameters. This avoids the influence of the operator or the limitations of conventional methods. The process was correctly applied to the Nuozhadu dam in China.
For Teijón el al. [
1], pit gradings weighing rock fractions were ineffective. Well-track tests ran under normal compaction conditions. The plate load test (PLT) requires the diameter of the load element to be five times the maximum size of the aggregate. Radioactive isotope density testing is conditioned by particle sizes and layer thicknesses less than 30 cm. The modified Proctor has the disadvantage of replacing sizes larger than 20 mm, with a minimum of 70% substituted material. The sand method is also useless as it is limited to a maximum size of 50 mm.
Mazari and Nazari [
4] consider that quality and density are not related. Density is a control element; quality must be based on determinations of the modulus of elasticity for quality acceptance. This can be estimated by means of formulas or usual values.
For Fernández et al. [
5], the limited development of compaction control tests justifies the execution of test sections.
A summary about the method of embankment quality control tests for a better comprehension, with the research deficiencies based on the literature review, is described in
Table 1. It shows the research gaps that the authors identified.
This study focuses on slate rocks. The reason for choosing the slate family is that these rocks are usually obtained from highway diggings, similar to those obtained from the demolition of buildings or pavements. Thus, the scientific knowledge was focused on slates because is a rock widely abundant and therefore the results on this material have wide applicability.
Possible correlations between the wheel-tracking test, topographic settlement, PLT and the on-site density test have been analyzed. The statistical analysis provided dependency relationships, allowing a new method of compaction control to be defined by applying only representative tests. It avoids unnecessary interruptions and the need for expensive equipment. It also proposes values that allow an efficient compaction control.
This research establishes the revision of certain methods, such as wheel-tracking or topographic settlement tests, in slate random embankments for highways, particularizing the proposed new compaction control method to be used on slate rocks in the crown zone with a maximum layer thickness of 600 millimeters.
Fernández et al. [
5] considered that it is possible to use rocks with low resistances (below 25 MPa) to obtain random fill. Such rocks are usually obtained from the demolition of structural and pavement concrete. Although the quality of slates and shales is lower than that of other materials such as greywackes, their strong anisotropic behavior associated with stratification and granulometric degradation after compaction is difficult to predict. Even so, they can be considered stable rocks and suitable for usage as random fill. In general, this can encourage high percentages of slab forms. Several experiments on test sections of stony materials have been conducted. Since the mechanical strength of grauwacke (a detrital rock formed from the consolidation of disintegrated granite minerals) is lower than that of granite, it usually ends up forming random fill.
The laboratory and field compaction data reported by Horpibulsuk et al. [
6] show that the relationship between relative density and the number of roller passes is represented by the logarithm function in laterite soils. Likewise, Oteo [
7] associated the requirements of materials to be used in fills with granulometry and density.
The effect of size ratio and air volume between the particles on the aggregate structure and packing density of binary mixtures was researched by Pouranian, M. R. and Haddock, J.E. [
8]. In addition, compaction parameters, including compaction slope, initial density, locking point and compaction energy index have been analyzed.
Onana et al. [
9] characterized the charnockite of Cameroon. The samples presented characterization tests with fine contents between 16 and 44%, high plasticity rates 26%–55%. In terms of its mechanical properties, it presents a high bearing capacity, with CBR 31-68 indexes, average RCS values 0.88–1.20 MPa and low tensile strength 0.07–0.15 MPa.
According to the Casagrande plasticity chart, the tested laterite gravels are clayey and highly plastic, which is due to their high kaolinite content. Southern Cameroon laterite materials are very low compressibility clayey gravel (GC) or silty gravel (GM) and can be used as sub-base layers for any volume of traffic.
Regarding embankment seats, Sagaseta [
10] indicated that associations could develop in random fills, which are made up of evolutionary materials, such as shales (fine grain detrital sedimentary rocks). In these cases, deferred settlements can become increased by the action of external agents (weathering, freezing cycles) that highly damage these rocks.
The Construction Embankment Technical Guide [
11] provides a classification of rocks. The R
6 group includes metamorphic rocks, such as slates and schists. The working method should be defined for the available machinery, earth moving methods, layer thickness, compaction procedures, number of roller passes, adjustment to optimum moisture and similar tests.
Oteo [
7] considered altered granite as a random fill. A specific study should be conducted before its excavation, transport and setting in place, and the appropriate control system must also be selected, since the classic Proctor test is hardly useful as a reference for such heterogeneous materials. For control of compacted random fill, the plastic density method, alongside geophysical methods, would be best. Radioactive isotope density can lead to specific problems in rock lacking fine fractions, since, because of their dimensions, the particles of such rocks do not allow the introduction of a gamma emitter into the ground. While it is still possible to measure backscattering, the soil volume tested for influence is inevitably smaller. Therefore, the measurements performed belong to the most superficial area, which is where the impact of the compaction energy is higher.
Wan-Huan et al. [
12] estimated the soil–water characteristic curve (SWCC) of soils with different initial dry densities.
Based on several experiments using highway test sections, Fernández et al. [
5] concluded that the results obtained from plate bearing tests show scatter.
For the wheel-tracking test, seat measurements are performed before and after carriage passes at ten points that are 1m apart from each other.
Sun et al. [
13] carried out certain experiments on 75 × 75 × 87 cm crushed rock samples subjected to vertical cyclic loading. Three coarsely crushed rock samples with initial grain sizes of 16–40, 25–50 and 50–80 mm were used to measure the corresponding parameters.
Garcia et al. [
14] analyzed the granular sub-base of the railway. Thus, Ev
2 is not associated with compaction, using Ev
1 as a reference. Moreover, cycling vibration loading can cause particle breakage and abrasion. The second/first modulus ratio at load bearing test (k) below 2.2. is established for the calibration of fine soils, which are very different from random fill, where the use of other parameters, such as wheel-tracking and plate bearing tests, prove more useful. The second modulus provides no information on the degree of compaction, so that other criteria based on the first modulus are considered more appropriate. This underdevelopment suggests the need for a new compaction control procedure, which entails the need for different functional parameters, such as automatic online complete process monitoring or specific loading plate diameters. The strongly anisotropic properties of slate make it suitable for its use in random fill or sub-base layers for any traffic volume.
2. Materials and Methods
The random fill material and field tests were carried out on the Spanish motorway A-66 "Ruta de la Plata”.
This research focuses on slate rocks. The reason for choosing the slate family is that these rocks are usually obtained from highway diggings, similar to those obtained from the demolition of buildings or pavements. Thus, the scientific knowledge was focused on slates because is a rock that is widely abundant and, therefore, the results on this material have wide applicability.
Table 2 provides a summary, including examples of the tests that were conducted on the slate alluvial material during excavation, with the last row showing average values.
The technical standard requirements for the slate alluvial materials used in the laboratory experiments were the particle size by screening, UNE 103101. [
15]; determination of the liquid limit and plastic limit of a soil, UNE 103103 [
16] and UNE 103104 [
17]; modified Proctor compaction test UNE 103501 [
18] and California Bearing Ratio (CBR), UNE 103502 [
19].
These soils come from the alteration of slates and are associated with low to medium plasticity. According to the Unified Soil Classification System (USCS), most of them belong to the GC group of the coarse-grained soils wrapped in a clay matrix. Large sizes of the parent rock remain, the percentage after sifting through a 20 mm sieve being 68.8%, and, at the same time, there is an important percentage of fine fractions, with an average of 29.3% after using a fine sieve (0.075 mm). Bedrock weathering variations resulted in the classification of a significant number of samples within the group of high plasticity silts (MH). The existence of coarse sizes implies that CBR testing yielded high values, with an average of 20.9.
In this research, the modified compaction control tests according to Teijón-López-Zuazo et al. [
20] have been used, which modify the test procedures in the wheel-tracking test and in the topographic settlements. The study is particularized to the random filling of the crown with slate rocks. To facilitate interpretation, the crown includes the two upper layers of the filling, with thicknesses in the penultimate layer between 60 and 40 cm in the last. All the tests that were used in the experiment are shown in
Table 3.
X-ray powder diffraction (XRD) is an analytical technique that we are not choosing to perform because it does not have a strong relation with the compaction quality control, which is strongly associated to different characteristics like mechanical properties.
The wheel-tracking test is carried out with a metal structure on which it is measured. These are welded profiles known as "H" shapes. The wheel-tracking test provides the measurement points in compaction batches, which are between 100 and 200 m. The truck should be conducted through topographic leveling pegs. The test result is the average value between different of measurements before and after the passage of the truck. The pegs reduce the possibility of extreme erroneous observations and the chance of any potential errors.
The other trial reviewed is the topographical settlement. It measures the seats after roller passes. This control method and its limitations were thoroughly revised in the research [
19]. The criteria suggested for quality control in the crown is grouped in
Table 4.
The degree of compaction proposed is associated with a modified Proctor compaction energy level. All the tests were performed under the same moisture conditions to prevent soil stiffness increases and noticeable dry density decreases in the PLT as a result of decreases in water content to below optimum levels. The ANOVA statistical analysis and Levene’s F test have been done. As a large sample size was obtained, the Kolmogorov–Sminornov test was used to check for normal distribution. Alternatively, the Shapiro–Wilk test was used. When processing road geotechnical tests, a strong association between variables is considered when the value of parameter R2 is higher than 0.70. To summarize, the multivariate analysis ANOVA offered a generalized, single, linear model for the adjustment. There is no difference between dependent and independent indicators with the highest goodness-of-fit.
4. Discussion
The topographic settlement test usually measures the first and last pass of the compaction roller. In the revised procedure, measurements are also taken on the penultimate and last pass of the compactor. There is a strong correlation of the topographic settlement improved with the PLT (ɸ 600mm), so one of them can easily be deduced from the other.
The wheel impression test lacks precision. The test distance is only 10 meters and the measurements are made on the ground. The test has been revised to improve on these deficiencies by using metal picks, doubling the number of measurement points and a test distance of 50 meters. As a dependency relationship was found between the revised test and the PLT (ɸ 600mm), this allows the wheel impression test to be replaced by the PLT (ɸ 600mm).
As the maximum dry density is obtained by laboratory compaction using a modified Proctor test, the degree of compaction is obtained from the field of dry density. However, average density control using nuclear methods is characterized by its high heterogeneity, low performance and testing of only low degrees of thickness, making plate bearing tests necessary to assess stiffness. To evaluate the quality of compacted soil only from the results of plate bearing tests, these were performed using compacted soils with a moisture content within a specific interval (−2, +1%) above modified Proctor optimum water content (wopt). A decrease in the water content from wopt according to modified Proctor means an increase in stiffness according to PBT, whereas dry density decreases. The PBT is a test where the highest pressure of the load is on the surface, providing surface measurements and strongly associated with surface moisture. Therefore, surface moisture is the main parameter in the result of the test. Due to this, all the PBT were carried out immediately after nuclear tests. In other words, density and PBT were defined at the same moisture content. Hence, the results from the in situ density test and the PLT (ɸ 600mm) provide an evaluation of the quality of compacted soil in terms of the degree of compaction requirements. Additionally, analyzed tests have yielded excellent results, supporting the possibility of using sizes larger than fine grain soils in random fill at the crown level.