Compaction Procedures and Associated Environmental Impacts Analysis for Application of Steel Slag in Road Base Layer

: In recent years, recycling steel slag is receiving growing interest in the road base layer construction ﬁeld due to its role in alleviating land occupation and resource shortages. However, the mixture compaction and its environmental impact on practical construction sites remain unclear, which may hinder the application of steel slags in road layers. This study investigates the pavement construction of the ‘Baotou-Maoming’ motorway, located in Inner Mongolia, China, analyzing the compaction procedures and assessing the environmental impacts caused by the road base layer containing steel slag. Firstly, mechanical properties and texture appearances of the steel slag aggregates are characterized. Afterwards, the comparative assessments for steel slag and andesite layers compaction are quantiﬁed from equivalent CO 2 emission and energy consumption aspects, respectively. The results show that the steel slag has a better surface texture than the natural aggregates; physical properties including compactness, ﬂatness and compressive strength comply with the requirements for applying steel slag to a hydraulically bound mixture. Compared to the base layer using andesite aggregates, the compaction vibration period of the course containing steel slags should be reduced to achieve a proper density due to the “hard-to-hard” effect that occurs between the adjacent steel slag particles. Consequently, the additional energy and the equivalent CO 2 are generated at 2.67 MJ/m 3 and 0.20 kg/m 3 , respectively.


Introduction
Road networks have been rapidly developing in recent decades, which is strictly linked to modernization and urbanization. The United States Department of Transportation announced that the national highway system was composed of over 164,000 miles of highway in 2011 [1], while the length increased to 223,257 miles in 2017 [2]. At the end of 2018, the total length of national highways in China reached 4.85 million kilometers [3], reaching 5.01 million kilometers in 2019 [4]. In many studies, the rapid expansion and improvement of new road constructions can be seen, especially in developing countries [5]. Nevertheless, the sustainability of road construction and development is increasingly questioned due to various environmental issues, such as the large consumptions of natural aggregate resources.
With the exploration of solid waste recycling, steel slag has become a potential material to be increasingly developed and used as a manufactured aggregate for civil engineering applications [6]. Generally, steel slag is a by-product of the steelmaking industry. In the steel slag and natural mineral aggregates in asphalt pavement construction, stating that approximately 39.77 TJ of energy and 2653 tons of CO 2 can be reduced during the material production process [23]. Václavík et al. replaced natural aggregates with steel slag in cement concrete and investigated associated environmental impacts using an LCA. The results showed that steel slag aggregate can diminish 7% of environmental burdens [24]. Anastasiou et al. considered the sustainability of utilizing industrial by-products (fly ash and steel slag) in cement mixtures. The results found that cement clinker content is critical for GHG emissions, and the transport distances of steel slag aggregate could strikingly affect the environmental burdens in the material production stage [25].
The majority of research explores the potentials of steel slag in laboratories, while the specific construction procedures related to practical pavement engineering are seldom researched and reported, which may hinder the application of steel slag in pavement construction. To address the research gap, this study has investigated a stretch of motorway located in Inner Mongolia, China. The surface's micro-texture and the composition of the manufactured aggregates were analyzed in the laboratory before application in the field. The effect on basic road performance was analyzed for the different compaction procedures of a hydraulically bound mixture containing steel slag aggregate, compared with a reference material andesite, while the base course compaction procedure was thoroughly investigated in the field by means of three types of compaction plants. Eventually, the associated equivalent CO 2 emissions and energy consumptions were analyzed to assess the overall environmental impacts when steel slag was applied into the road base layer.

Materials and Methods for the Case Study
This study has considered the construction of a hydraulically bound mixture for a road base layer along the "Baotou-Maoming" motorway located in Inner Mongolia, China. The base course was designed with a width of 7 m and a thickness of 20 cm.

Physical Properties of Steel Slag and Andesite
The steel slags, obtained from Lucheng and supplied by the Baotou City Steel Slag Road Performance and Engineering Application Research Center, underwent a three-stage process to make them into usable manufactured aggregates: initially, the tailings (blast furnace slags) generated by the steel industry were placed into a hot furnace and cooled down; afterwards, the steel tailings were aged by climate conditions, aiming to decrease the expansion and shrinkage effects; then, the stable slags were crushed, and sieved for appropriate sizes according to the actual requirements.
Lucheng steel slags and andesite were used in the hydraulically bound mixture as coarse and fine aggregates (the size designating the limit between coarse and fine dimensions was considered equal to 4.75 mm), respectively [26]. For comparison, the study also investigated a traditional base layer which only comprised of andesite aggregates for the hydraulically bound mixture. Portland cement with a compressive strength grade of 32.5 MPa was employed as binder (Table 1). The physical and mechanical properties of the steel slags were investigated according to the Chinese standard of Testing Methods of Aggregate for Highway Engineering (JTG E42-2005) [27] and displayed in Table 2. In addition, the content of free calcium oxide (f-CaO), usually regarded as a major issue due to the phenomenon of swelling in water [28,29], is listed in Table 2, which is below the suggested limit [30]. The physical properties of the aggregates are closely related to the load capacity of the hydraulically bound mixture, as well as to the adhesion with the cement binder [31]. The surfaces of both the Lucheng steel slag and andesite were investigated by means of an Aggregates Image Measurement System (AIMS) device ( Figure 1). The Texture Index (TI) is commonly used for quantifying the aggregate's surface texture. TI values, obtained as reported in Equation (1), range from 0 to 1000: a bigger TI indicates a rougher surface, leading to a higher shear resistance and load capacity of the mixture [32,33]. Twenty aggregates with particle sizes comprised between 10 mm and 16 mm were investigated and given a TI value.
where n is the decomposition level during measurement, N is the total number of coefficients in the detected image, the i value is 1, 2 or 3 based on the image analysis dimension (horizontal, vertical, and diagonal), the j value corresponds to the wavelet coefficient index, D is the compiled decomposition function, and x and y locate the coefficients in a transformed domain.  Furthermore, this study employed a Mean Free Path-3 Dimensions-Scannin Force Probe (MFP-3D-SA) to investigate the microscopic characteristics of the a interface. The fractal dimension was utilized to describe the irregularity of each the scanned surfaces, and the logarithmic curve of the incremental variable was to analyze the slope and the shape of the aggregate interface. The width and dep image was 20 μm and the height was 2.5 μm. Figure 3a,b show the 3D microscop of Lucheng steel slag and andesite, respectively. The obvious differences in struc microscopic characteristics illustrated that the lighter color which appeared in F accounts for a higher percentage, and its emulated structure also presents a grea larity than Figure 3b. Therefore, compared to manufactured aggregate, the surfac of natural aggregate is more regular, indicating that the selected steel slag aggre better surface characteristics to prepare a hydraulically bound mixture. Four different TI ranges were used to describe the surface roughness, namely 0-200 (low), 200-500 (moderate), 500-750 (high) and 750-1000 (extreme). Figure 2 shows the cumulative TI percentage of Lucheng steel slag and andesite: most of the natural aggregates belong to a moderate level, while the largest part of the manufactured steel slag aggregates lies within a high level. The average TI values for Lucheng steel slag and andesite were 566.8 and 437.5, respectively. A greater TI value of an aggregate can obtain a stronger load capacity of the hydraulically bound mixture. Hence, Lucheng steel slag offers a better surface texture for cement workability.   interface. The fractal dimension was utilized to describe the irregularity of each point on the scanned surfaces, and the logarithmic curve of the incremental variable was adopted to analyze the slope and the shape of the aggregate interface. The width and depth of the image was 20 µm and the height was 2.5 µm. Figure 3a,b show the 3D microscopic texture of Lucheng steel slag and andesite, respectively. The obvious differences in structure and microscopic characteristics illustrated that the lighter color which appeared in Figure 3a accounts for a higher percentage, and its emulated structure also presents a greater angularity than Figure 3b. Therefore, compared to manufactured aggregate, the surface texture of natural aggregate is more regular, indicating that the selected steel slag aggregate has better surface characteristics to prepare a hydraulically bound mixture.  Figure 4 illustrates the research diagram of this study. The surface textures of both of the base layers containing manufactured aggregates and natural aggregates were initially analyzed by means of an Aggregates Image Measurement System (AIMS) and a Mean Free Path-3 Dimensions-Scanning Atomic Force Probe (MFP-3D-SA). Successively, the following in-service properties were assessed: surface flatness, layer compaction and compressive strength of hydraulically bound mixture specimens. Eventually, the research quantified the equivalent CO2 emissions and the energy consumption related to the compaction operations.

Mechanical Properties of Base Layer
Steel slag was used for the coarse aggregate fractions (size bigger than 4.75 mm) and

Mechanical Properties of Base Layer
Steel slag was used for the coarse aggregate fractions (size bigger than 4.75 mm) and andesite for the fine aggregate fractions during the production of the hydraulically bound mixture. Table 3 presents the gradation design of the aggregates. The percentage of added cement was 5% by mass, and water content applied here was 4.6%. In this investigated case, three different types of compactors were used in the field for the base layer containing steel slags: Single Steel Wheel Roller (SSWR), Double Steel Wheel Roller (DSWR) and Pneumatic Tired Roller (PTR). Table 4 presents the specifications of the used compactors. The density achieved by each compaction procedure was evaluated using sand (φ 15 mm, loose density 1.38 g/cm 3 ) according to JTGE60-2008 [34]. The first investigated compaction procedure (C1) was conducted on andesite aggregates, composed of three vibration cycles (weak, strong, weak) for a total length of 60 m back and forth performed by a SSWR; successively, cycles two and one were accomplished by PTR and DSWR (static mode), respectively. After this compaction procedure, we measured the compactness (K) in two different locations with the maximum densities at 93.0% and 96.2%. These results did not fulfil the practical requirements (98.0%) based on the code of JTGE60-2008 [34].
Specifically, under the strong vibration, the surface of the steel slag mixture underwent deformation, meanwhile, under high frequency vibration, the "hard-to-hard" effect occurred between the adjacent steel slag particles, resulting in the loosening of the compacted mixture. The surface containing hard steel slag particles made the compaction process more sensitive than andesite. Hence, the compaction procedure (C1) could obtain a good compaction for andesite, but failed to achieve the required effects for steel slag. Figure 5 shows the "hard-to-hard" phenomenon for steel slag. effect generated by the weak and the strong vibration is clearly visible in terms of loose surface; on the other hand, a proper density was achieved due to reduced vibration frequency and an increased number of cycles with static pressure.

Surface Flatness
The flatness (σ) of the base layer was analyzed after compaction by means of a Continuous Flatness Meter (CFM). The device wheel was equipped to measure the surface flatness for all of the test sections [35]. The device automatically recorded the vertical displacement deviations every 10 cm (unit measurement interval), and the final flatness index was evaluated every 100 m (Equation (2)). The CFM was towed for the entire base layer length (500 m) at the speed (V) of 6.8 km/h. 200B Based on the results obtained for the first compaction procedure, an optimized compaction procedure (C2) was proposed to alleviate the "hard-to-hard" effect. The operations were performed according to the following sequence: one SSWR cycle applying static pressure forth and strong vibration back; then, three SSWR cycles (static pressure) and two PTR cycles; eventually, one more SSWR cycle (static pressure). Following this procedure, the average compactness achieved 95.8%. Although the code requirements were not met again, the higher values proved the beneficial effect generated by the weaker vibration.
Eventually, a third compaction procedure (C3) was investigated to increase the base layer density: one cycle with a SSWR (static pressure) plus three weak cycles with a SSWR (vibration); then, four cycles with a PTR; finally, one cycle completed with a DSWR (static pressure). The compactness values of C3 measured in three different locations (L1, L2, L3) were now equal to 97.8%, 98.8% and 99.1%, respectively ( Table 5). The average value was 98.6%, which lies within the code threshold. The surfaces of the base layer compacted according to the three procedures (C1, C2, C3) are displayed in Figure 6. The detrimental effect generated by the weak and the strong vibration is clearly visible in terms of loose surface; on the other hand, a proper density was achieved due to reduced vibration frequency and an increased number of cycles with static pressure.  6. Surface of base layer compacted according to the C1-C3.

Surface Flatness
The flatness (σ) of the base layer was analyzed after compaction by means of a Continuous Flatness Meter (CFM). The device wheel was equipped to measure the surface flatness for all of the test sections [35]. The device automatically recorded the vertical displacement deviations every 10 cm (unit measurement interval), and the final flatness index was evaluated every 100 m (Equation (2)). The CFM was towed for the entire base layer length (500 m) at the speed (V) of 6.8 km/h. 200B

Surface Flatness
The flatness (σ) of the base layer was analyzed after compaction by means of a Continuous Flatness Meter (CFM). The device wheel was equipped to measure the surface flatness for all of the test sections [35]. The device automatically recorded the vertical displacement deviations every 10 cm (unit measurement interval), and the final flatness index was evaluated every 100 m (Equation (2)). The CFM was towed for the entire base layer length (500 m) at the speed (V) of 6.8 km/h.
where σ i is the standard deviation of flatness for 100 m (mm), d i represents the displacement deviation for the unit measurement interval (mm), and N is the total number of recorded data. The flatness (σ) of all of the five test sections exceeded 1.2 mm, and the average value was 1.3 mm ( Figure 7); therefore, the standard requirements were fulfilled [36].

FOR PEER REVIEW 9 of 15
where σi is the standard deviation of flatness for 100 m (mm), di represents the displacement deviation for the unit measurement interval (mm), and N is the total number of recorded data. The flatness (σ) of all of the five test sections exceeded 1.2 mm, and the average value was 1.3 mm ( Figure 7); therefore, the standard requirements were fulfilled [36].

Hydraulically Bound Mixture Strength
The compressive strength of the hydraulically bound mixture containing steel slag was evaluated with 26 cylindrical samples, which were randomly divided into two groups (A and B). All of the specimens were molded with a diameter and height of 150 mm using the universal testing machine WE-1000. The samples were stored at room temperature for

Hydraulically Bound Mixture Strength
The compressive strength of the hydraulically bound mixture containing steel slag was evaluated with 26 cylindrical samples, which were randomly divided into two groups (A and B). All of the specimens were molded with a diameter and height of 150 mm using the universal testing machine WE-1000. The samples were stored at room temperature for six days and soaked for one day. Eventually, the unconfined compressive strength was assessed using the universal testing machine WE-300B according to the Chinese standard code JTG E51 [36,37]. The compressive strength varied between 8 MPa and 11 MPa as showed in Figure 8 and Table 6.
the universal testing machine WE-1000. The samples were stored at room temp six days and soaked for one day. Eventually, the unconfined compressive st assessed using the universal testing machine WE-300B according to the Chine code JTG E51 [36,37]. The compressive strength varied between 8 MPa and showed in Figure 8 and Table 6.
The limited variability between the two groups indicated that the c strength results were relatively stable. The compressive strength of the two gro ples was higher than 9 MPa within the 95% confidence interval. Compared to aggregate mixture (8.0 MPa within the 95% confidence interval), the steel slag m achieve a higher strength, indicating that steel slag mixture is able to properl compressive stresses generated by road traffic loads [38].   The limited variability between the two groups indicated that the compressive strength results were relatively stable. The compressive strength of the two groups of samples was higher than 9 MPa within the 95% confidence interval. Compared to the natural aggregate mixture (8.0 MPa within the 95% confidence interval), the steel slag mixture can achieve a higher strength, indicating that steel slag mixture is able to properly resist the compressive stresses generated by road traffic loads [38].

Energy Consumption
This study investigated the diesel consumption of the four different roller compactors, and the functional unit was assumed to be the road base layer with a length of 30 m. The assessment of energy consumption and equivalent CO 2 emissions referred to the applied compaction procedures (C1 for andesite mixture and C3 for steel slag mixture) previously reported in Section 3.1.1.
Equations (3) and (4) express the energy consumption associated to the vibration process and driving process, respectively. Equation (5) specifies the total energy consumption.
where E 1 is the energy consumption for the vibration process (kJ), F represents the vibration force (kN), A is the vibration amplitude (mm), f indicates the vibration frequency (Hz), L is the working length in one cycle (m), n represents the number of compactors, D is the working width of the base layer (m), v indicates the driving speed (m/s) and d is the compacting width (m). E 2 is the energy consumption for driving process (kJ) and P is the operation power (kW). E represents the total energy consumption during compaction (kJ) and α is the energy utilization rate of diesel compactors (%). Two types of SSWRs were adopted for compaction, namely KS225HD-2 (SSWR-1) and CLG 620H (SSWR-2). SSWRs consume a different amount of diesel when they operate according to strong (S) or weak (W) vibration mode (Table 7): the fuel consumption was calculated based on the number of SSWRs and the working time. The compacting width for DSWR and PTR was 2.1 m and 2.7 m, respectively. Based on past work [39,40], the average thermodynamic efficiency α for diesel consumption was assumed to be 40%.  Figure 9 reports that the total energy consumption for the compaction of the functional unit containing steel slag aggregate or andesite aggregate is 364.27 MJ or 251.99 MJ, respectively. Considering the base layer containing natural aggregate, the highest energy amount was consumed by SSWR-1 (107.78 MJ); regarding the hydraulically bound mixture containing Lucheng steel slag, the PTR entailed the biggest energy consumption (140 MJ) because the vibration period of the SSWR was shortened and the static pressure of the PTR was increased.
Sustainability 2021, 13, x FOR PEER REVIEW 11 of 1 MJ) because the vibration period of the SSWR was shortened and the static pressure of the PTR was increased.
On one hand, the use of steel slag as a manufactured aggregate in pavement con struction is beneficial given the increasing shortage of natural aggregates; on the othe hand, the compaction on site would require more energy than the amount demanded fo a traditional base layer to ensure appropriate density and surface flatness. Compared to the compaction of the andesite base course, the application of steel slag generated an ad ditional 2.67 MJ/m 3 energy consumption.

Equivalent CO2 Emission
This study referred to the China Statistical Yearbook [41] and the Chinese Life Cycle Database (CLCD) [42] to retrieve data related to the diesel supplied during compaction On one hand, the use of steel slag as a manufactured aggregate in pavement construction is beneficial given the increasing shortage of natural aggregates; on the other hand, the compaction on site would require more energy than the amount demanded for a traditional base layer to ensure appropriate density and surface flatness. Compared to the compaction of the andesite base course, the application of steel slag generated an additional 2.67 MJ/m 3 energy consumption.

Equivalent CO 2 Emission
This study referred to the China Statistical Yearbook [41] and the Chinese Life Cycle Database (CLCD) [42] to retrieve data related to the diesel supplied during compaction. The equivalent CO 2 emission was calculated with Equation (6).
where G indicates the equivalent CO 2 emission (kg), E represents the total energy consumption during compaction (kJ), G 0 is the standard diesel emission factor (kg/kg) and E 0 is the standard diesel calorific rate (MJ/kg). Figure 10 displays the equivalent CO 2 emissions related to the four types of compactors. Overall, the compaction of the functional unit containing manufactured aggregates or natural aggregates engendered the release of 27.08 kg and 18.73 kg equivalent CO 2 , respectively. Moreover, the PTR generated the largest quantity of equivalent CO 2 for steel slag mixture (10.41 kg), while the SSWR-1 was the major GHG contributor for andesite mixture (8.01 kg). In our investigated case, compacting steel slag mixture caused higher equivalent CO 2 emissions than andesite mixture by 0.20 kg/m 3 .   According to the elaborated results in Figures 9 and 10, steel slag and andesite mixtures presented the same ratio distribution of energy consumption and GHG emissions. To assess the influence of the four compactors, their occupied percentages of GHG emission are illustrated in Figures 11 and 12 for steel slag and andesite base layers, respectively. SSWRs played a significative role in the compaction of steel slag (55%) and andesite (62%) mixtures, followed by PTRs and DSWRs; the PTR had undertaken a higher proportion of work in steel slag mixture, by 10%, because its static pressure alleviated the "hard-to-hard" effect. Furthermore, compactor's driving operation accounts for most of the diesel consumption.   From a life cycle perspective, most studies have reported that using manufact aggregates such as steel slag in lieu of natural aggregates can positively contribute t reduction of environmental depletion [43,44]. However, the calculated results obta from this investigated case show that when only focusing on the environmental imp    From a life cycle perspective, most studies have reported that using manufact aggregates such as steel slag in lieu of natural aggregates can positively contribute t reduction of environmental depletion [43,44]. However, the calculated results obta from this investigated case show that when only focusing on the environmental im From a life cycle perspective, most studies have reported that using manufactured aggregates such as steel slag in lieu of natural aggregates can positively contribute to the reduction of environmental depletion [43,44]. However, the calculated results obtained from this investigated case show that when only focusing on the environmental impacts in compaction procedures, steel slag aggregates cause higher energy consumption and GHG emissions than andesite. Hence, reducing the environmental impacts of compaction can distinctly strengthen the road sustainability by employing manufactured aggregates instead of natural aggregates.

Conclusions
Steel slag is a solid waste which can be conveniently recycled in civil engineering applications. This study characterized the use of steel slag in a hydraulically bound mixture for road base layers. Furthermore, an additional traditional base layer containing andesite (a natural aggregate) was also investigated for comparison. The compaction procedures of the base layers have been thoroughly investigated in the field by means of four different compactors; furthermore, surface flatness, layer density and the compressive strength of the hydraulically bound mixture were also evaluated. Eventually, the environmental impacts of the compaction operations were assessed.
Based on the performed analyses, results, and discussion, the following conclusions can be drawn: (1) Steel slag aggregate offers better roughness and surface micro-morphology properties than andesite aggregate in this investigated case. (2) The base layer containing steel slag aggregate fulfils the road construction requirements specified by the standard code, namely compactness, surface flatness and compressive strength of the hydraulically bound mixture.
(3) Strong vibrations should be avoided or reduced during compaction to mitigate the adverse "hard-to-hard" effect between adjacent steel slag particles; the static pressure exerted by the rolling compactors should be increased.