1. Introduction
Soil is commonly stabilized by mixing soil, water, and chemical stabilizers, such as cement, lime, or other chemical additives, to strengthen soft and collapsible ground [
1,
2]. These stabilizers have been proven to be effective but may have potential environmental drawbacks such as carbon emission issues and economic considerations due to their associated production costs [
3,
4,
5]. Specifically, cement manufacturing alone accounts for 5 to 8% of worldwide carbon emissions and 10% of global industrial energy consumption [
6,
7]. In recent years, there has been a growth in the popularity of alternative composite and eco-friendly binders, such as concrete applications [
8,
9,
10] or soil stabilizers [
11,
12,
13]. This can be attributed to their sustainability, cost-effectiveness, and distinctive chemical composition. Hence, the development of these construction materials represents a new frontier in the scientific quest to reduce the environmental impact of construction.
The biomass ash types previously explored as soil stabilizers include discharged waste from power plants [
14,
15,
16] and industrial by-products [
12,
17,
18,
19]. Studies suggest that the stabilizing effect of biomass ash depends on the parameters of the soil mixture (i.e., soil type, the unique chemical and mineral content of the binder, and the water-to-binder ratio), the specimen preparation, and the curing conditions [
13]. These biomass ashes contain some silica and alumina, exhibiting a soil-binding potential even when combined with moisture only [
16,
18,
20]. However, the contribution of biomass ash to the mechanical properties of stabilized soil varies [
21]. The combination of rice husk ash [
22] and sawdust ash with clay [
23] and wood ash-treated sand [
16] results in positive enhancements in the mechanical properties. In contrast, olive waste ash-stabilized marl soil [
24] and wood ash-stabilized clay [
18] were observed to have limited efficacy due to their short-lived strength and low cementation capacity. These findings suggest that the strength gain of biomass ash-stabilized soil depends on the host soil, its unique chemical composition, and its capacity to produce cementitious minerals [
13,
21].
Despite the differences in the stabilization efficiency of biomass ash, biomass fly ash (BFA) is a good candidate for alternative cementitious materials due to its distinct strength gain when used as an admixture and due to its positive environmental and economic advantages [
13,
25]. One type of BFA that is largely understudied is wood pellet fly ash (WA) [
26]. WA is a by-product of wood-pellet-based energy production. The use of wood pellets conforms with international trends regarding the use of biomass ash as a sustainable material for household and commercial heating fuel following ISO standards [
27] and as one of the responses to the global problem of climate change. In particular, the South Korean government has developed an energy policy [
28,
29,
30] to lessen the carbon emission issues associated with traditional fossil fuels [
31,
32]. As a result, the domestic demand for wood pellets is anticipated to increase [
33]. This transition to the use of wood pellets in domestic power plants may further lead to a surge in its by-product, WA. The generated ash by-products make up roughly 2% of the fuel’s weight, and since 2014, over 350,000 tons of wood pellet fly ash have been produced annually, leading to landfill saturation [
26]. Therefore, the excessive disposal of wood pellet fly ash in landfills has compelled researchers to develop recycling solutions.
However, despite the considerable presence of WA, there has been no significant advancement in the development of recycling technologies. Furthermore, in light of WA’s potential as a BFA, it is worth noting that there is currently a lack of research investigating WA’s effectiveness in soil stabilization. Additionally, the cementation mechanism of pure and blended BFA-stabilized albite-based weathered granite soil (WS) has rarely been analyzed. Thus, a unique challenge that this study aims to address is the excess production of WA by-products; the study deals with this issue by developing a wood pellet fly ash blend with low cement proportions while ensuring that the desired soil strength is achieved.
2. Previous Work
Kim et al. [
34,
35,
36] previously investigated the physical and chemical properties of raw and hardened wood pellet fly ash and the physical, chemical, and strength properties of WA ground granulated blast-furnace slag (GGBS)–cement blends for concrete and mortar applications. Their research findings show that raw wood pellet fly ash contains alkaline components, including K
2O, Na
2O, and MgO. In addition, a 27.8% CaO content was detected, suggesting its binding potential, especially for binders with 20% CaO (or greater) [
18,
37]. Interestingly, SO
3, which could promote a high reactivity with Ca-based compounds [
34,
35], was identified; K
2O, which is also beneficial to the strength gain [
38] and alkali aggregate reactions, was also identified [
34,
35,
36]. Moreover, raw WA contains low peaks of quartz and lime minerals; however, when WA is mixed with water, the hardened material shows a low peak of new hydrocalumite hydrate (based on XRD analysis). These findings imply that WA has unique hydraulic properties due to mineral changes during its production; the minimal strength contribution from WA’s hydration is due to its low hydrocalumite peak intensity.
In addition, a parallel study by Balagosa et al. [
39] was conducted to investigate the wood pellet fly ash binding effect (5% to 25% by dry mass of soil) on weathered granite soil (WS) through dry mixing. The findings of the unconfined compressive tests [
40] show that the measured strength gain of the WA-treated WS suggests premature and slow cementation for specimens with a low WA dosage (5% and 15% WA contents). Also, the slight 28-day strength decrease for 15% and 25% WA contents implies low WA hydration capacity when stabilizing WS [
34,
35,
36,
39]. Nevertheless, these findings motivated us to accelerate WA’s binding capacity; thus, we developed a sustainable construction material by creating a WA–GGBS–cement blend (predominantly made up of WA) and determining the optimal binder ratio through a series of experimental investigations. Firstly, cement activation tests were conducted on the WA–GGBS–cement blends (fresh and cured conditions) with a mixing ratio of WA (0% to 50%), cement (0% to 50%), GGBS (0% to 50%), and standard sand at a 1:3 ratio (binder-to-sand) and a water-to-binder (W/B) ratio of 50%; the blend was mixed to a slurry state in accordance with Korean Industrial Standards KS L ISO 679 [
41] for cement strength test methods through air curing conditions for 3, 7, and 28 days.
In the previous studies [
34,
35,
36], freshly blended binder specimens were evaluated through slump flow (KS F 2594 [
42]), setting time (KS F 2436 [
43]), and hydration temperature testing methods. On the other hand, a series of compression tests (KS L 5105 [
44]) and flexural strength tests (KS F 2408 [
45]) were conducted on the cured specimens. The experiment results of the freshly blended binders show that the hydration of WA–GGBS and WA–cement is lower than that of the specimens mixed with pure cement and GGBS–cement blends. These findings suggest that WA is more suitable as an alkali stimulant than a hydration contributor. Interestingly, the cured specimens with 50% WA, 30% GGBS, and 20% cement were determined as the optimal design mix in the WA–GGBS–cement blend due to high unconfined compressive strength (
qu) and flexural strength. Notably, the use of optimal
qu is typically adopted in characterizing soil improvements and as a design parameter basis for ground improvement projects [
46]. Thus, the present study adopted the 50% WA, 30% GGBS, and 20% cement ratio as a new sustainable soil stabilizer, called wood pellet fly ash blended binder (WABB), and extended its workability in natural soils in the field. It has been determined that the WA’s reactive properties are advantageous as an alkali stimulant, and the latent hydraulic characteristic of GGBS complements the strength gain in low-cement-based binders for concrete and mortar applications [
34,
35,
36]. However, comprehensive analyses of the WABB cementation mechanism on albite-based weathered granite soil and natural field soils are lacking.
4. Materials and Methods
The WS used in this study came from a quarry in Cheonan, South Korea, with a median particle size (D
50) of 1.21 mm [
47], designated as Material Type 2 by AASHTO T294-92 [
48] and classified as poorly graded sand (SP) by the Unified Soil Classification System (USCS). The WABB’s constituents are cement, GGBS, and wood pellet fly ash. The WA used in this study was discharged by Yeongdong Eco Power Plants, operated by Korea South-East Power Co., Ltd., (KOEN), wherein a hybrid dust electric precipitator at the top of the boiler gathers the by-products created during the pyrolysis of wood pellets for energy consumption. WA often has a dark gray color, a dry density of 2.31 g/cm
3, a fineness (measured using the Blaine method) of 3350 cm
2/g, and a pH of 12 or higher. The physical attribute of WA is characterized by its unique fine particles with a median size (
d50) of 15.347 × 10
−3 mm. BFA, with a particle size ranging from 0.004 to 0.1 mm [
16,
49,
50,
51], is a promising candidate as a soil stabilizer for the following reasons: BFA has a small particle size capable of effective strength enhancement when used in mortar applications [
52,
53] and as a soil stabilizer [
21,
54]. In relation to this, there may be lesser production costs when using BFA as a soil stabilizer since the pre-processing of coarse raw materials is required to increase binding efficiency [
53,
55]. Thus, this evidence suggests that WA is a promising sustainable construction material.
GGBS, on the other hand, is the by-product of processed iron from a blast furnace that can be used in the production of concrete in combination with commercial type 1 Portland cement.
Figure 1 and
Figure 2 show the particle size distribution (PSD) and SEM morphology of the materials used in this study.
Table 1 presents the index properties of the host soil (e.g., minimum void ratio, e
min; maximum void ratio, e
max; specific gravity, G
s, etc.) and binder constituents. The chemical components of the WABB constituents based on X-ray fluorescence (XRF) tests are shown in
Table 2. All the components of WABB contain high calcium oxide (CaO) and silicon dioxide (SiO
2), while GGBS has the highest aluminum oxide (Al
2O
3) content, all of which point to the possibility that it could react with active materials and form cementitious products. Notably, WA contains plenty of K, Na, and Ca, which are beneficial in supplying alkali ions to stimulate the latent hydration of GGBS and cement hydration. The WABB design mix was determined based on the best-performing cement mortar in terms of strength, as given in a previous study by Kim et al. [
26,
34,
35,
36].
The phase diffractions of the individual constituents of WS and WABB were analyzed using XRD and are presented in
Figure 3. The XRD reveals the WS silica (SiO
2) and alumina (Al
2O
3) components through quartz and albite, respectively. The mineralogical constituents of WA are quartz (SiO
2), lime (CaO), tripotassium sodium sulfate (K
3Na(SO
4)
2), and sodium divanadate (Na
4V
2O
7). The presence of lime in WA suggests its binding potential [
15,
23,
36]. GGBS contains calcite (CaCO
3), while cement is composed of tri-calcium silicate (C
3S), di-calcium silicate (C
2S), tri-calcium aluminate (C
3A), tetra-calcium aluminoferrite (C
4AF), and gypsum (CaSO
4). The presence of gypsum helps to increase the soil strength [
56]. The quartz minerals in WS, WA, and GGBS can enhance the Si-O-Si bonds, leading to a strength increase in stabilized soil [
57].
Initially, modified proctor tests [
58] were performed on weathered granite soil combined with wood pellet fly ash at 5%, 15%, and 25% by adding water to the dry host soil and mixing for two minutes to meet the optimum moisture content (OMC) and maximum dry density (MDD) in a separate project. As a result, the measured MDD and OMC for 5% WA were 1.932 g/cm
3 and 12.0%; those for 15% WA were 1.917 g/cm
3 and 10.5%; and those for 25% WA were 1.901 g/cm
3 and 9.6.% [
39]. These results indicate a minimal shift in OMC when combining WA and WS. Following that, the present study adapted the OMC and MDD for the WABB-stabilized weathered granite soil (WABB-WS) mix and employed dry mixing methods for the stabilized soil [
59,
60] to investigate the behavior change in compacted, weathered granite soil [
61].
Figure 4 illustrates the present study’s mixing condition and testing flow, including mechanical property and microstructure tests on 5%, 15%, and 25% of the WABB–WS specimens. In particular, the host soil was first sieved to eliminate coarse particles and then dried in an oven to make it ready for the experiments. Following the previously determined OMC, the dry soil was mixed with water and then with dry binders using a laboratory mixer for ten minutes until it became uniform. An 8 mm rod was used to tamp five layers of the soil–binder mixture into a cylindrical split mold, eventually aiming for a constant molding dry density of 95% relative compaction of 50 mm diameter and 100 mm high specimens. The specimens were finally prepared at a target dry density (
ρd) of 1.827 to 1.877 g/cm
3, 1.812 to 1.858 g/cm
3, and 1.796 to 1.834 g/cm
3 for the 5%, 15%, and 25% WABB mixtures, respectively. To prevent the specimens from drying out, the specimen molds were sealed with plastic and cured at room temperature (25 ± 1 °C) in a closed tank partially filled with water (without direct contact with the water). The curing chamber humidity was 85 ± 2% and was monitored using a humidity meter. After 3, 7, 14, and 28 days, the specimens were extracted from the molds and examined. Dry curing was used to simulate stabilized WS projects installed near the ground surface and above the water table with no access to more water.
Table 3 summarizes the testing type, mix proportions, and curing days for the WABB–WS specimens.
Unconfined compressive tests (UCT) [
40] were performed to evaluate the unconfined compressive strength (
qu) and the secant modulus (
E50). The specimens were vertically and uniaxially loaded at 1%/min. After UCT tests, the pH of the mixed soil (approximately 9 mm and below) with the WABB specimens was monitored at 3, 7, 14, and 28 days (the specimens are marked with UCS and pH in
Table 3) using a pH meter [
62]. Subsequently, microstructural analyses were conducted through a series of SEM-EDS and XRD tests. In particular, another set of fractures (after UCT tests at 28 days) was reduced to small pieces to fit the mold size for the SEM-EDS tests (Model MIRA LMH) operated at 5 to 20 kV and a resolution of 1 nm. Also, XRD qualitative analysis (Model MiniFlex600-Rigaku) was conducted by investigating the reflected peaks and their varying intensities, which were operated at 40 kV and 30 mA on a Cu target (clear peaks at the 2θ region of 0 to 5 were not observed, thus the 2θ range of 5 to 70 is presented). A semi-quantitative analysis was conducted by measuring the intensity counts on the detected minerals [
63] to evaluate the hydration of WABB at increasing dosage rates. These tests were conducted to validate the WABB’s contribution to the morphology of WS through the presence of cementitious materials relative to the mineral and chemical change over 28 curing days (see specimens marked under the MA column in
Table 3). Moreover, twelve additional WABB–WS specimens were prepared and cured for 7 and 14 days and used for two testing series of pH and total suction tests (see specimens marked under the ST and pH columns in
Table 3). These specimens were immediately extracted from the molds after the prescribed curing days to reduce air exposure. Then, the central cores of the specimens were divided into three parts consisting of the top, middle, and bottom portions. For Series 1, the core samples were carefully sliced into smaller intact specimens (approximately 9 mm and below). The specimens were immediately collected to maintain the initial conditions of the material, and the aluminum caps were filled to evaluate the total suction using a potentiometer (WP4C, METER Group) and the water content [
64,
65,
66,
67]. The use of WP4C for rapid total suction monitoring was effectively carried out to evaluate the hydraulic behavior of the unsaturated soils [
68,
69] and stabilized soils [
65,
70]. Furthermore, Series 2 testing was conducted by collecting the remaining intact samples and using them for pH tests.
6. Discussion: Cementation and Suction Development
To analyze the WABB–WS strength-gaining mechanism provided between the particle cementation and the suction (for unsaturated conditions), the rate of change in
Ψ,
qu, and
wc and its significance were investigated and defined as follows:
where Δ
Ψ = the rate of change in
Ψ;
Ψ(i) = the measured suction at a specified
ith curing day;
Ψ5%WABB(7) = the measured 7-day suction representing the early minimal suction contribution of 5% WABB–WS;
Δqu = the rate of change in
qu;
qu(i) = the measured strength at a specified
ith curing day;
qu(WS) = the measured 0-day strength of compacted weathered granite soil, representing the early strength of the pure soils under the minimal suction;
Δwc = the rate of change in
wc;
wc(i) = the measured water content at specified curing days; and
wc(m) = molding water content.
Figure 11 shows that as the WABB content and curing days increase,
Δqu and
ΔΨ increase while
Δwc decreases. In addition, the observed rapid desaturation rate of all the WABB–WS specimens at 7 days agrees with and supports Abdul-Hussain and Fall [
79] and Jing et al.’s [
103] claim regarding the effect of cement (or cementitious material) hydration to suction on the early
qu gains. Then, at 28 days, the measured relatively slower 28-day
Δwc suggests the contribution of the WABB’s cementation capacity since more water is inscribed between the materialized cementitious minerals and the WS grains.
As the number of curing days is extended, the WABB material gradually develops a dense skeleton structure; thus, the majority of the strength improvement can be associated with the enhancement of the WS internal structure. However, due to the concurrent densification of the WABB–WS matrix and pore refinement, the formation of suction is still discernible [
104]. Thus, the contribution of suction to the unsaturated WABB–WS strength gain mechanism slightly reduces. Despite the possibility of a suction-controlled strength gain, the results of this study agree with those of other cementitious soil studies, wherein the effect of suction on the strength gain of cementitious soils gradually decreases as the number of curing days increases [
59,
103,
105,
106]. The reduced contribution of suction is corroborated by the positive strength gain of the WABB–WS specimens due to the alteration in soil micropores through the newly developed cementitious minerals (i.e., C-S-H, C-A-S-H, Ettringite, and anorthite), leading to enhanced WS stiffness. Based on the aforementioned analysis, the Pearson correlation [
107] was employed to assess the relative contribution of the 7-to-28-day unconfined compressive strength of WABB–WS (
qu(7 to 28)WABB-WS) with WABB cementation over the curing days. Then, a regression model was developed to predict the
qu(7 to 28)WABB-WS using 70% of the total data as training data and 30% as cross-validation data. Lastly, the predicted and measured values were evaluated based on three factors: coefficient of determination (
R2), correlation coefficient (
R-value), and the average absolute percentage error (AAPE) [
108,
109], as follows:
where BDR = the ratio of dry wood pellet fly ash blended binder mass to dry soil mass in %; CD = curing days in the range of 7 to 28 days. The
qu(7 to 28)WABB-WS had
R-values of 0.18 and 0.96 for
BDR% and
CD, respectively.
Figure 12 compares the values predicted by Equation (4) to the measured values, and this comparison yields an
R2 of 98.68, an
R-value of 0.98, and an AAPE of 12.25%. These findings suggest that the predicted
qu(7 to 28)WABB-WS using the Equation (4) model is in good agreement with the experimental results as displayed in the cross-plots. In addition, the
E50 and
qu of WABB-WS were normalized with the average
qu and
E50 of the compacted WS and correlated to derive a linear relationship between
E50 and
qu (
m = 2.072;
n = 27;
R2 = 95.27), as shown in Equation (5):
where
E50(7to28)WABB-WS = the measured secant modulus of the WABB-stabilized WS from 7 to 28 days;
E50(WS-AVE) = the average secant modulus of the compacted WS without binders;
qu(7 to 28)WS-WAB = the measured unconfined compressive strength of the WABB-stabilized WS at 7 to 28 days; and
qu(WS-AVE) = the average unconfined compressive strength of the compacted WS without binders.
Figure 13 presents the simulation of model Equation (5) with the experimental data. The obtained
R-value of 0.97 and the AAPE of 15.71% indicate a good agreement between the predicted and measured values. Therefore, the
qu and
E50 of the compacted WS and
qu(7 to 28)WS-WABB can be used to estimate the 7-to-28-day secant moduli of the WABB–WS mixtures. However, the prediction model Equations (4) and (5) are only applicable with the compacted WABB–WS prepared at a target dry density range of 1.796 g/cm
3 to 1.877 g/cm
3, a minimal range of
wc(m) at 9.6% to 12.0%, and the testing conditions applied in the present study. Notably, the slower and premature strength gain of 5% WABB–WS and its microstructural developments further suggest that the optimal dosage rate for WABB applications in stabilizing WS should be 15 to 25% to achieve proper cementation. Nevertheless, the findings of the present study suggest the potential of the WABB in enhancing the stiffness of WS through its positive strength gain and the WABB’s cementation mechanism and in generating cementitious minerals with WS. These findings make the WABB utilization a cost-effective solution for developing BFA blended binders with a large volume of cement replacement [
54,
97,
110] that can stabilize albite-based weathered granite soil.