Evaluation of Mechanical and Microstructural Properties of Waste Tire Improved Cemented Clay

ABSTRACT

Each year, an enormous number of tires approach the end of their useful lives, posing threats to human health and the environment. On the other hand, cement is frequently used to treat soils in geotechnical applications, while its production causes a significant environmental impact. Thus, this study provides a disposal alternative by investigating the influence of tire rubber fiber (TRF) as a partial replacement of cement in artificially cemented soils. Unconfined compressive strength (qu) and ultrasonic pulse velocity (UPV) were measured on the various mixtures to assess strength, stiffness, and ductility index. Statistical analysis and regression models were conducted, and a novel approach to estimate rubberized-cemented-clay was proposed to determine G_o, q_u, and E from a single nondestructive test. Moreover, SEM was performed to observe the interaction of the TRF and cement with the clay on a microscopic scale. The results showed that the 2.5% TRF content improves rubberized cemented clay’s strength, stiffness, and ductility index by around 12–15%. Furthermore, up to 10% of the TRF blends achieved the minimum requirements for rammed earth, base, and sub-base treated soils. However, 20% of TRF reduces G_o, and q_u, by around 20–30% while improving the ductility index by 35–40%.

摘要

每年，大量轮胎的使用寿命即将结束，对人类健康和环境构成威胁. 另一方面，水泥在岩土工程应用中经常用于处理土壤，而其生产会对环境产生重大影响. 因此，本研究通过研究轮胎橡胶纤维（TRF）作为人工水泥土中水泥的部分替代品的影响，提供了一种处理方案. 测量了各种混合物的无侧限抗压强度（qu）和超声脉冲速度（UPV），以评估强度、刚度和延性指数. 进行了统计分析和回归模型，并提出了一种新的方法来估计橡胶胶结粘土，以从单个无损检测中确定Go、qu和E. 此外，用扫描电镜在微观尺度上观察了TRF和水泥与粘土的相互作用. 结果表明，2.5%的TRF含量可使橡胶胶结粘土的强度、刚度和延性指数提高约12-15%. 此外，高达10%的TRF混合物达到了夯实土、基底和底基层处理土壤的最低要求. 然而，20%的TRF将Go和qu降低约20–30%，同时将延性指数提高35–40%.

KEYWORDS:

• Clays
• soil stabilization
• compressive strength
• stiffness
• waste

关键词:

• 粘土
• 土壤稳定
• 抗压强度
• 刚度
• 浪费

Introduction

During the transition from hunter-gatherer to settled life, people preferred fertile lands and alluvium clays rich with minerals, whilst homes were built on firmer grounds. However, increasing population lead to an increase in housing and infrastructure needs. With such demand, people were forced to build on problematic soils that covered large land areas worldwide. Such soils are not preferable for development due to the problems encountered in their use (Cooper and Rose 1999; Młynarek, Stefaniak, and Wierzbicki 2012). Low strength, excessive settlement, high compressibility, and their tendency to shrink-expand due to seasonal cycles are mainly the issues directly reflected in the strength and durability of infrastructures resting on them.

Li and Zornberg (2003), in one of the earliest studies on fiber-reinforced soils, showed that introducing fibers into cohesive and granular soils increases peak shear strength. Another study by Li and Zornberg (2005) and Freilich et al. (2010) demonstrated that fibers more evenly distribute pressures across the soil matrix. Authors (Ekinci and Ferreira 2012; Freilich, Li, and Zornberg 2010; Özkul and Baykal 2007; Tang et al. 2007) studied the deformation modes of cohesive soils and reported that clay alone samples developed distinct slip planes, while the addition of fibers results in a barreling type of failure mode. Mirzababaei et al. (2013) conducted a more exhaustive study on carpet waste-reinforced clay and reported that the addition of carpet fiber increased compressive strength, decreased post-peak strength loss, and altered the failure behavior from brittle to ductile and the failure pattern from localized to uniform barrel-shaped failure.

The demand for traditional soil stabilizers such as cement and lime has substantially increased due to their significant benefits for ground improvement. However, the production of such materials has significant environmental impacts. Many countries and associations have taken the initiative to reduce the effects of the construction industry on the environment. The EU, in 2020, approved a policy initiative, “Green Deal,” to reach climate neutrality by 2050, and one of its strategies is a sustainable built environment (European Commision 2020). The incorporation of industrial wastes in the ground has attracted the attention of researchers because it has two-way benefits: enhancing poor soils’ performance and using a waste disposal method. Previous studies such as (Afrakoti et al. 2020; Fakhrabadi et al. 2021; Janalizadeh Choobbasti et al. 2019; Kutanaei, Afrakoti, and Choobbasti 2021) investigated the mechanical characteristics of cemented soils when copper sludge or coal waste is added. Janalizadeh Choobbasti et al. (2019) used an ultrasonic pulse velocity test to determine the influence of copper sludge on the properties of cemented clay. They found that a 15% addition of copper sludge is considered an optimal amount.

Nevertheless, the use of waste rubber tires in manufacturing construction materials has been growing. The ground waste rubber tire was recently engineered to create an innovative cement-based sandwich composite material with promising energy dissipation properties and crack resistance against impact load (Valente et al. 2022). One of the earliest studies of tire clay composite was performed by Al-Tabbaa and Aravinthan (1998) on naturally over-consolidated fissured clays mixed with a shredded tire. The authors studied the composite’s mechanical behavior and reported a 40% reduction in compressive strength compared to clay alone. The authors further noted that the initial stiffness of the tire-stabilized clay was half that of the clay specimens.

Researchers later incorporated waste tire into cemented clays to achieve strength requirements, allowing the reinforced soil to be used as a base, sub-base of roads, tamped earth, and lightweight fill behind the retaining walls and embankments. For instance, the minimum unconfined compressive strength requirement, suggested by the FHWA (2014), for subgrade cemented soils of pavement roads is 1.4 MPa after 7 curing days. The PCA (1992) recommended a range of 2.07–5.52 MPa strength after 7 curing days for base and sub-base of medium to high volume roads. For rammed earth, the USA Housing and Construction standard (New Mexico 2009) specified 2.068 MPa (300 psi) as the minimum compressive strength in rammed earth soils (treated or untreated), and suggested a 1.38 MPa (200 psi) strength for treated soils after 7 curing days Chan (2012) used granulated rubber waste in cement-stabilized clayey sand with 0–4% cement and rubber shreds or chips. The authors investigated the strength and stiffness response of the mixes via unconfined compressive strength and bender element tests and reported that both specimen’s strength and stiffness were controlled by cement content and the increase in the rubber content resulted in an increase in ductility of the stabilized material. They also reported that a correlation between the strength (q_u) and stiffness (G₀ and E) parameters exists for the mixes Yadav and Tiwari (2017a) utilized crumb rubber in cement treated soft clay. Authors used 0.8–2 mm size 2.5%, 5%, 7.5%, and 10% rubber crumb and 3% and 6% cement content by the dry weight of the specimens. The authors demonstrate that the addition of rubber crumbs decreases the unconfined compressive and split tensile strength of specimens compared to clay-cement treated samples while increasing the ductility index. The authors demonstrate that adding rubber crumbs decreases the unconfined compressive and split tensile strength of specimens compared to clay-cement treated samples while increasing the ductility index. Bekhiti et al. (2019) used 0.5, 1, and 2% rubber fiber content with 5%, 7.5%, and 10% cement content to examine the unconfined compressive strength and ductility of cement-stabilized bentonite clay soil. Authors stated that maximum strength and ductility results were obtained at 2% rubber fiber content and further reported that ductility increases with the increase of fiber content. Lately, Akbarimehr et al. (2020), investigated the geotechnical behavior of three different forms of waste tire, including granular, fiber, and chips, by performing various tests, including compaction, unconfined compressive strength, direct shear and triaxial. It was reported by the authors that rubber fiber provided higher increase in strength among all other forms of rubber tire used. Besides a minor increase in the strength of rubberized cemented clay, a reduction with a low rate of up to 5% inclusion of tire rubber was reported by He et al. (2023).

This study investigates an alternative and useful utilization of TRF in cemented soils. This study focuses on the strength, stiffness, ductility, and microstructure of clay reinforced with tire rubber. It also explores the potential uses of rubberized clay in geotechnical applications, such as sub-base layers, earth dams, and shallow foundations. Correlation models are developed to assess strength and stiffness using the porosity/binder index. Additionally, it proposes correlations between geotechnical parameters to estimate fundamental properties from a single nondestructive test. Overall, the investigation found correlations backed by experiments to determine the best TRF dosage to replace cement in clayey soils without affecting their mechanical properties significantly.

Experimental program

To evaluate the mechanical and chemical behavior of the proposed mix, a total of 180 samples were prepared.

Materials

Soil

The soil was collected from an excavation at a depth of around 3 m; the pit is located in the Iskele district, northern part of Cyprus. Sufficient disturbed soil samples were collected and dried at 60 C° before use. Several laboratory experiments were conducted to assess the soil’s physical properties per ASTM standards; the results of these experiments are summarized in Table 1 below. These experiments included a test for Atterberg limits (ASTM D4318-17e1 2017), a sieve analysis (ASTM D6913/D6913M–17 2017), and a specific gravity (ASTM D854–14 2014). The clay treated is categorized as inorganic clay with low to medium plasticity (CL) following the Unified Soil Classification System (USCS) (ASTM D4318-17e1 2017). Moreover, the soil’s particle size distribution is presented in Figure 1.

Figure 1. The grain size distribution (%) of studied clay.

Table 1. Physical properties of the alluvial clay.

Properties	Value
Liquid limit (%)	46
Plastic limit (%)	20
Plasticity index (%)	26
Specific gravity	2.66
Fine gravel (4.75 mm < diameter <20 mm) (%)	–
Coarse sand (2.00 mm < diameter <4.75 mm) (%)	–
Medium sand (0.425 < diameter <2.00 mm) (%)	–
Fine sand (0.075 mm < diameter <0.425 mm) (%)	6
Silt (0.002 mm < diameter <0.075 mm) (%)	54
Clay (diameter <0.002 mm) (%)	40
D50 particle diameter (mm)	0.005
USCS class	CL

The X-ray fluorescence spectrometry results for the chemical composition of the clay are presented in Table 2. Clay is composed mainly of silica (36%), with smaller amounts of calcium (17.3%) and alumina (11.7%) also present.

Table 2. Chemical analysis of both clay and cement.

Oxides (%)	Cement (Type I)	Clay
CaO	63.6	17.3
SiO2	21.7	36
AL2O3	4.8	11.7
Fe2O3	3.9	6.67
SO3	1.4	1.13
MgO	0.3	6.38
K2O	0.3	1.81
LOI	2.1	16.4
Others*	1.9	2.61

End-of-life Tire Rubber Fibre (TRF)

The tire used in this research was shredded in a fiber-like form (TRF) with a length ranging from 5 mm to 20 mm and a diameter of 1 mm to 3 mm (Figure 2). A study was performed measuring lengths of 100 random TRF, and results show that there is an even distribution of all TRF sizes. The TRF was obtained from a recycling factory in the Nicosia district of the northern part of Cyprus. Before using it in the mixtures, it was washed to remove any dust and impurities covering the surface and air-dried to eliminate the moisture. The specific gravity of the tire shreds was determined as 1.13. Moreover, the chemical composition of the tire was determined through energy-dispersive X-ray spectroscopy. The primary chemical element forming the tire is carbon with 88.1%, oxygen with 8·6% and the rest is accumulated of zinc, sulfur, silicon, magnesium, and aluminum.

Figure 2. The utilized shredded waste rubber fiber (TRF).

Cement

Cement (Type I – Ordinary Portland) specified by (ASTM C150/C150M–20 2020) is used in this study. The specific gravity of the cement was found as 3.15. The cement has a Blaine fineness of 305 m²/kg and 2.1 of loss on ignition. The chemical composition compounds of the cement are summarized in Table 2. The minimum cement content was determined as 7% where below specimens were disaggregating when submerged in water. Additionally, USACE (1994) states that CL-type soils are estimated to require 9% (dry weight of soil) cement for stabilization.

Methods

Specimen preparation

Based on the findings of a standard compaction test (ASTM D698-12e2 2012), the dry density for the material was determined to be 1810 kg/m³, and the ideal moisture content was determined to be 17.35%. To achieve the desired compacted densities of 1800 kg/m³ and 1600 kg/m³, a 100 mm × 50 mm cylindrical split mold was designed and built. The impact of compaction on various binder types at 100% saturation was emphasized using two distinct dry densities. After determining the dry mass of clay soil (${\mathrm{M}}_{\mathrm{S}}$) based on the predefined dry density (${\mathit{\rho }}_{\mathrm{d}}$), the proportion of cement content (${\mathrm{M}}_{\mathrm{c}}$) was measured as a percentage of the dry mass of soil where the TRF content (${\mathrm{M}}_{\mathrm{TRF}}$) was the partial replacement of the mass of cement content. The blend’s dry components were stirred until homogeneous, and then the prescribed amount of water was gradually introduced while mixing until a consistent paste was obtained. The blend’s dry components were stirred until homogeneous, and then the prescribed amount of water was steadily introduced while mixing until a consistent paste was obtained. The mixture was statically compacted inside the mold in three layers, according to Ladd (1978). Before adding the new layer, its surface was scarred to ensure a perfect bond between the layers. After producing the sample, the mass and dimension measurements were recorded. All the samples were cured inside a humidity chamber with a relative humidity of about 95% at 24° ± 2°C (ASTM C511–19 2019). A three-decimal point balance was used to measure the weight of the samples with an accuracy of ± 1% and 0.5% of the water content. The samples’ height and diameter were measured by a digital caliper with two decimal places and an accuracy tolerance of ±0.5 mm for diameter and 1±mm for sample length.

The porosity was determined following Equation (1)(1) $\mathit{\eta }=100-100\left[\frac{{\mathit{\rho }}_d}{\mathit{total}\mathit{\hspace{0.17em}}\mathit{mass}\mathit{\hspace{0.17em}}\mathit{of}\mathit{\hspace{0.17em}}\mathit{solid}}\right]\left[\frac{M_S}{\mathit{G}{s}_S}+\frac{M_c}{\mathit{G}{s}_c}+\frac{{\mathit{M}}_{\mathit{TRF}}}{\mathit{G}{s}_{\mathit{TRF}}}\right]$(1) ,

(1) $\mathit{\eta }=100-100\left[\frac{{\mathit{\rho }}_d}{\mathit{total}\mathit{\hspace{0.17em}}\mathit{mass}\mathit{\hspace{0.17em}}\mathit{of}\mathit{\hspace{0.17em}}\mathit{solid}}\right]\left[\frac{M_S}{\mathit{G}{s}_S}+\frac{M_c}{\mathit{G}{s}_c}+\frac{{\mathit{M}}_{\mathit{TRF}}}{\mathit{G}{s}_{\mathit{TRF}}}\right]$(1)

where $\mathit{G}{s}_S$, $\mathit{G}{s}_c$, $\mathit{G}{s}_{\mathit{TRF}}$ are the subsequent specific gravities of soil, cement, and tyre rubber fiber, respectively.

The cement content highly controls the geotechnical parameters of the artificially cemented soils. The porosity/cement index (ŋ/C_iv), proposed by Consoli et al. (2016), was later modified by Ekinci et al. (2019) to include the effect of multiple binders (TRF in this case). X_iv is the porosity/binders index that considers the impact of cement and TRF binders on the treated soils’ strength and stiffness properties, and it is calculated using Equation (2)(2) $X_{\mathit{iv}}=\mathit{\hspace{0.17em}}\frac{V_C+V_{\mathit{TRF}}}{\mathit{V}}$(2) .

(2) $X_{\mathit{iv}}=\mathit{\hspace{0.17em}}\frac{V_C+V_{\mathit{TRF}}}{\mathit{V}}$(2)

Where $V_{\mathit{TRF}}\mathit{\hspace{0.17em}},V_C$, and $\mathit{V}$ are the tire rubber fiber, volumes of cement, and the total volume, respectively.

Table 3 shows the blends’ proportions, the curing periods, and the tests conducted for each mix’s specimens. Consoli et al. (2007), suggested modifying the porosity/cement index exponentially to give a better assessment of the strength of granular soils treated with cement. In this study, $X_{\mathit{iv}}$ is modified by an exponent of 0.32, in accordance with other studies (Baldovino et al. 2021; Consoli et al. 2007; Ekinci, Scheuermann Filho, and Consoli 2019). The range of η/X_iv^exp or η/C_iv^exp found to be between 20% and 40% (Ekinci, Scheuermann Filho, and Consoli 2019) and 20–45% for fiber-reinforced cemented soils (Mirzababaei et al. 2018).

Table 3. Details of molding, curing, and normalization parameters of all blends.

Soil Type	Cement content (%)	Tyre rubber Fiber (TRF) as replacement of Cement (%)	Moulding dry density (kg/m^3)	Curing Period (Days)	Test Type	Normalization Index (∇)	qu for normalization (kPa)	G0 for normalization (kPa)	E for normalization (MPa)
Clay	7, 10, and 13	0%	1600 and 1800	7	qu, G0, and SEM	η/(Xiv)^0.32 = 25	2380.33	4524.16	312.90
				28			3988.55	5551.43	510.01
				60			4542.32	6030.50	591.12
	7, 10, and 13	2.5%	1600 and 1800	7	qu, G0, and SEM	η/(Xiv)^0.32 = 25	2984.58	4716.09	302.67
				28			4702.49	5830.56	465.04
				60			5242.88	6509.94	529.92
	7, 10, and 13	5%	1600 and 1800	7	qu, G0,and SEM	η/(Xiv)^0.32 = 25	2726.23	4601.76	269.39
				28			4222.97	5839.10	433.33
				60			4821.76	6415.76	491.89
	7, 10, and 13	10%	1600 and 1800	7	qu, G0, and SEM	η/(Xiv)^0.32 = 25	2787.40	4473.83	281.22
				18			3711.04	5506.36	405.81
				60			4116.92	6096.10	466.70
	7, 10, and 13	20%	1600 and 1800	7	qu, G0, and SEM	η/(Xiv)^0.32 = 25	2328.02	4169.00	240.83
				28			3424.05	5075.68	330.93
				60			3761.03	5590.22	407.87

Moreover, the strength and stiffness outcomes have been normalized following a model developed by Consoli et al. (2017), inspired by Diambra et al. (2017). The researchers utilized a method of dividing the formed powered equations by a specific value of strength obtained at a particular value of η/X_iv^0.32. This approach has proven effective in evaluating the mechanical properties of cement-treated soils cured at various ages. Table 3 shows the determined values of q_u, G_0, and E at η/X_iv^0.32 = ∇ = 25 applied to normalize all the different blends.

Unconfined compression strength

The unconfined compressive strength test was done according to the standard (ASTM D1633–17 2017) to see how partial TRF cement replacement affected the compressive strength of cement-treated soil blends. All the samples were submerged in water for 24 h before testing to ensure saturation. A computer-controlled, 23 kN load frame was used to conduct the unconfined compressive strength test (Figure 3(a)). Load frame is equipped with a load cell with an accuracy of 5 N, and LVDTs were installed on both sides of the sample to measure the axial deformation up to 0.001 mm. The strain rate was set as 1% axial strain/minute per standard (ASTM D1633–17 2017). The mass and dimensions of the specimens were collected again before performing the strength test. The whole process was repeated for the other specimens. At the linear region of the vertical stress-strain diagram, Equation (3)(3) $\mathit{E}=\frac{\mathit{\Delta \sigma }}{\mathit{\Delta \epsilon }}$(3) was used to calculate the modulus of elasticity (E) and strength at failure of each sample (Figure 4).

(3) $\mathit{E}=\frac{\mathit{\Delta \sigma }}{\mathit{\Delta \epsilon }}$(3)

Figure 3. Tests’ setups; (a) load frame for unconfined compressive strength and (b) pundit for ultrasonic pulse velocity (UPV).

Figure 4. Typical stress strain diagrams showing the determination of elastic modulus.

To emphasize the effectiveness of the tire rubber fiber on the deformation response of cemented soils, the ductility index ($\mathit{D}$) was determined following the method proposed by Bekhiti et al. (2019), using Equation (4)(4) $\mathit{D}=\frac{{\mathit{\Delta }}_{\text{withTRF}}}{{\Delta }_{\text{withoutTRF}}}$(4) .

(4) $\mathit{D}=\frac{{\mathit{\Delta }}_{\text{withTRF}}}{{\Delta }_{\text{withoutTRF}}}$(4)

Where ${\mathrm{\Delta }}_{\mathit{withTRF}}$ and ${\mathrm{\Delta }}_{\mathit{withoutTRF}}$are the strains at peak vertical stress of TRF-cement-clay and cement-clay specimens, respectively.

Shear wave velocity

Due to its nondestructive nature, the Ultrasonic Pulse Velocity (UPV) test was performed on all samples before the unconfined compressive strength experiments were performed. The (ASTM C597–02 2002) standard was followed to carry out this test, using a Model C368 MATEST pundit device. Two transducers (a shear wave sender and a receiver) are connected to the device (Figure 3(b)). An ultrasonic wave is sent across the sample by transducers, which are mounted to the sample’s parallel sides using silicon grease. The duration of the wave’s journey from source to target was measured, and the shear wave velocity (Vs) was computed based on the wave travel distance (sample length). Using the following Equation (5)(5) $G_0=\mathit{\hspace{0.17em}}\mathit{\rho }\ast {V_s}^2$(5) .

(5) $G_0=\mathit{\hspace{0.17em}}\mathit{\rho }\ast {V_s}^2$(5)

Where $\mathit{\rho }$ and G_o are the sample’s density and the maximum shear velocity, respectively.

Microstructural analysis

The microstructure analysis was performed to investigate the pozzolanic reaction of the cement as well as the interaction of the TRF in the blends. A QUANTA 400F Field Emission with 1.2 mm resolution and supplemented with an Energy Dispersive System (EDS) was utilized to obtain images of approximately 10 mm pieces of the samples with a range of magnification between 1K to 10k. Scanning Electron Microscopy (SEM) was executed on specimens containing 2.5% TRF as replacement of 7, 10, and 13% of the cement content, compacted at both densities and cured for 60 days.

Results and discussions

Influence of controllable factors on the mechanical behaviour

The analysis of variance (ANOVA) was used to statistically evaluate the influence of different variables of the tests on the resulting mechanical properties of TRF cemented clay. StatGraphics 18 program was used for this purpose. Two and three factor analyses were conducted to examine the influence of the dry density (1600 kg/m³ and 1800 kg/m³), the curing period (7, 28, and 60 days), the cement content (7, 10, and 13%) and tire rubber fiber content (0, 2.5, 5, 10, and 20%) on the q_u, G₀ and E of the tire rubber fiber-reinforced cemented clay.

All of the main factors and the two-factor and three-factor interactions were shown to be statistically significant, with p-values lower than 5%, indicating a confidence level of 95%. Table 4 shows the relationship between controllable factors (dry density, curing period, cement content, and tire rubber fiber content) and q_u, G₀ and E. The higher the sum of the square value, the higher the deviation from the mean. Therefore, high divergency of controllable factors results in more effect on the q_u, G₀ and E. For q_u, the most important factors were the curing period, cement content, dry density, and TRF content. For the G₀ the affecting factors were in the order of dry density, curing period, cement content, and TRF content. For the E, the affecting factors were in the order of curing period, dry density, cement content, and TRF content. It can be summarized that since the TRF addition is a replacement for cement, it is least likely to affect the measured parameters, where density and curing periods are highly dominant in controlling measured parameters.

Table 4. ANOVA table of responses regarding unconfined compressive strength, initial shear modulus and elastic modulus.

	Source	Sum of Squares	Df	Mean Square	F-Ratio	P-Value
Unconfined Compressive Strength, qu	MAIN EFFECTS
	A: Dry Density (kg/m^3)	122586000	1	122586000	3077.35	0.0000
	B: Cement (%)	124272000	2	62135900	1559.83	0.0000
	C: Tyre Rubber Fiber (%)	60385900	4	15096500	378.97	0.0000
	D: Curing (Days)	163743000	2	81871700	2055.27	0.0000
	INTERACTIONS
	AB	3530980	2	1765490	44.32	0.0000
	AC	19591800	4	4897960	122.96	0.0000
	AD	15839900	2	7919950	198.82	0.0000
	BC	10849800	8	1356230	34.05	0.0000
	BD	6147820	4	1536960	38.58	0.0000
	CD	26487100	8	3310890	83.12	0.0000
	ABC	2376580	8	297072	7.46	0.0000
	ABD	1339010	4	334751	8.4	0.0000
	ACD	6858890	8	857361	21.52	0.0000
	BCD	4568140	16	285509	7.17	0.0000
	RESIDUAL	4222510	106	39835
	TOTAL (CORRECTED)	572800000	179
Initial Shear modulus, Go	MAIN EFFECTS
	A: Dry Density (kg/m^3)	101007000	1	101007000	3055.77	0.0000
	B: Cement (%)	47886100	2	23943100	724.35	0.0000
	C: Tyre Rubber Fiber (%)	20344300	4	5086080	153.87	0.0000
	D: Curing (Days)	88042400	2	44021200	1331.78	0.0000
	INTERACTIONS
	AB	1136910	2	568456	17.2	0.0000
	AC	2402250	4	600562	18.17	0.0000
	AD	1665500	2	832751	25.19	0.0000
	BC	2167240	8	270905	8.2	0.0000
	BD	1380420	4	345104	10.44	0.0000
	CD	1629340	8	203668	6.16	0.0000
	ABC	628476	8	78559.5	2.38	0.0214
	ABD	951092	4	237773	7.19	0.0000
	ACD	1885740	8	235718	7.13	0.0000
	BCD	1242790	16	77674.6	2.35	0.0051
	RESIDUAL	3503770	106	33054.4
	TOTAL (CORRECTED)	275873000	179
Elastic modulus, E	MAIN EFFECTS
	A: Dry Density (kg/m^3)	1145390	1	1145390	1789.57	0.0000
	B: Cement (%)	881730	2	440865	688.81	0.0000
	C: Tyre Rubber Fiber (%)	569087	4	142272	222.29	0.0000
	D: Curing (Days)	1399580	2	699791	1093.36	0.0000
	INTERACTIONS
	AB	38721	2	19360.40	30.25	0.0000
	AC	28857	4	7214.23	11.27	0.0000
	AD	56597	2	28298.30	44.21	0.0000
	BC	32319	8	4039.93	6.31	0.0000
	BD	24495	4	6123.63	9.57	0.0000
	CD	99485	8	12435.60	19.43	0.0000
	ABC	13213	8	1651.60	2.58	0.0125
	ABD	13718	4	3429.47	5.36	0.0005
	ACD	9739	8	1217.41	1.9	0.0661
	RESIDUAL	74244	116	640.04
	TOTAL (CORRECTED)	4489620	173

The main effects plot for all strength and stiffness tests performed here are shown in Figure 5 (i.e. curing periods, cement content, tire content, and combined analysis). All graphs in Figure 5 show that 1800 kg/m³ specimens have the highest performance in terms of cementation, mobilization of fibers and encouraging soil–fiber interactions, given the higher results in terms of q_u, G₀ and E values in comparison to the specimens molded at 1600 kg/m³ density. It is also clear that the longer curing time and higher cement content lead to an increase in q_u, G₀ and E, when one looks at the curing period (Figures 5(a), 5(d) and 5(g)), can realize that the rate of increase on all parameters is slowed down between 28 and 60 days. However, when considering the effect of cement content (Figures 5(b), 5(e) and 5(h)), there is an apparent linear increase in all three parameters for the lower-density samples but not for the high-density samples. It is clear on Figure 5(c) that, at higher density, the addition of TRF gradually reduces the compressive strength. The same can be seen for the lower density mixtures when the percentage is higher than 2.5%, as in this percentage, an increase in strength was seen with respect to the 0%. Similarly, in Figure 5(f), it is clear that the initial shear modulus is at peak value when mixing 2.5% of TRF for both densities of specimens and reduces afterward. It is also monitored in Figure 5(i) that adding fiber degrades the elastic modulus with the increase of TRF content at both densities. However, at 2.5% of TRF inclusion and 1600 kg/m³ density, there is an evidential contribution of TRF replacement on all parameters when compared to unreinforced specimens.

Figure 5. The interaction between controllable factors curing period, cement content, TRF content and density for q_u, G₀ and E.

ANOVA analysis approach was proved to provide insight into the degree of contribution of each test variable to the mechanical properties of treated soils. The same approach was utilized by (Bak et al. 2021) to analyze the effect of various parameters on the bio-cementation process of the soil–steel interface.

Mechanical behaviour of the blends with respect to the porosity/binder index

Figure 6 represents the unconfined compressive strength (q_u) of all the blends as a function of the modified porosity/binder index (ɳ/X_iv^0.32). The Figs. 6(a)–(e) shows the clay mixed with 7, 10, and 13% of cement, prepared at 1600, and 1800 kg/m³, cured for 7, 28, and 60 days, and a partial replacement of the cement by 0, 2.5, 5, 10, and 20% TRF (Figures 6(a)–(e), respectively). At each TRF ratio, powered curves were plotted to fit the strength and porosity/binder index’s relationship for different curing ages. A degree of significance higher than 90% is observed in all the relationships in Figure 6. Overall, the strength (q_u) increases exponentially with the decrease of the porosity/binder ratio (ɳ/X_iv^0.32). Previous investigations (Baldovino et al. 2021; Consoli et al. 2007; Ekinci, Scheuermann Filho, and Consoli 2019) have reported similar observations. Samples compacted at 1800 kg/m³ dry density have less porosity and thus exhibit higher compressive strength than 1600 kg/m³ because of the increase of particle contacts through compaction. As expected, curing the samples for longer periods gives more time for the chemical reactions in the pozzolanic material therefore higher strengths are observed. Figure 6(f) shows that at 7 days of curing, the strength was improved by replacing cement with 2.5 to 10% TRF, which can be referred to the contribution of the fiber overcoming the influence of the cement replaced at early ages. However, at 28, and 60 day curing periods, only 2.5% and 5% TRF, at low porosity (1600 kg/m³), showed an improvement in the strength compared to 0% TRF blends and further substitutions significantly decreased the strength of the cement-treated soils. As the substitution of cement by TRF increases, a reduction in the pozzolanic material on the blends leads to lower strengths. The reason for such a drop in the strength is regarded to the low mass density and the poor bond between cemented clay and the rubber fiber. Furthermore, the elastic behavior of TRF caused the rubber tire to return to its original shape after it was compacted. This led to the creation of microcracks, thus weakening the bond between TRF and the cement-clay mixture.

Figure 6. The unconfined compressive strength (q_U) and adjusted porosity/binder index (ηl<X_iv)⁰·³² correlations for all curing days and cement percentages in both dry density specimens with (a) 0% TRF, (b) 2.5% TRF, (c) 5% TRF, (d) 10% TRF, (e) 20% TRF cement replacements, and (f) all the correlations.

Figure 7 shows the initial shear modulus (G₀) of all the mixes as a function of the adjusted porosity/binder index (ɳ/X_iv^0.32). Figure 7(a–e) represent the blends containing 0, 2.5, 5, 10, 20% of TRF, respectively, as a partial substitution of the cement content. The Fig. also includes the results of the mixes with 7, 10, 13% of cement and the two different compacted densities. As in the previous Fig., a powered curve was fitted to describe the trend of the relationship of G₀ and (ɳ/X_iv^0.32) at each curing period (i.e., 7, 28, 60 days). The regression coefficients of all the relationships in Figure 7 are higher than 94%, with only one exception at 91% and two at 93%. In Figs. 7(a–f), it can be seen that less porous specimens yielded higher initial shear modulus, and the longer the curing period, the higher the initial stiffness. Figure 7(f) contains all the curves, comparing the effect of TRF on G₀ of all the blends. It can be observed that G_o followed more or less the same behavior as the q_u results when concerned with the impact of TRF contents. The replacement of 2.5% and 5% TRF showed improvement along all the blends’ porosities and at almost all the curing ages, compared to 0% TRF. However, more than 10% cement replacement of TRF considerably decreases the initial shear modulus (G₀).

Figure 7. The shear modulus (G) and adjusted porosity/binder index (η/(X_iv)⁰·³²) correlations for all curing days and cement percentages in both dry density specimens with (a) 0% TRF, (b) 2.5% TRF, (c) 5% TRF, (d) 10% TRF, (e) 20% TRF cement replacements, and (t) all the correlations.

Figure 8 follows the same procedure described for Figures 6 and 7, for the relationship between E versus the porosity/binder index (ɳ/X_iv^0.32). The power curves and the equations of each group have a degree of significance of more than 90%. Figure 8(f) shows all the curves for all the groups. Altogether, the modulus of elasticity (E) results are generally consistent with q_u and G_o. A marginal change in the trend might be caused by the method of obtaining E when compared to q_u and G_o.

Figure 8. The modulus of elasticity (E) and adjusted porosity/binder index (η/(X_v)⁰·³²) correlations for all curing days and cement percentages in both dry density specimens with (a) 0% TRF, (b) 2.5% TRF, (c) 5% TRF, (d) 10% TRF, (e) 20% TRF cement replacements, and (t) all the correlation.

Nevertheless, specimens with 2.5 to 10% TRF showed a moderate decrease in the modulus of elasticity compared to samples with cement only. Beyond 20% TRF replacement, E starts to drop considerably. This is simply due to increased fibers limiting the cementation process.

Considering the previous discussion, it can be settled that replacing cement with TRF up to 2.5% can effectively improve the cemented clay’s compressive strength (q_u), which also has been observed by Yadav and Tiwari (Yadav and Tiwari 2017a). Similarly, the inclusion of up to 5% rubber tire has a minor reduction on the strength of cemented soils, as reported by (He et al. 2023). The elastic and initial shear modulus have been marginally enhanced at 2.5% TRF. Further increase of TRF content reduces the strength and stiffness values due to the lack of bonding between the tire and artificially cemented clay (Kim and Kang 2011). Higher proportions of TRF transform the behavior of the combinations from brittle to ductile, which can be beneficial in places with energy absorption demands. The compressible performance of the TRF can provide an alternative layer between the base and the asphalt in the roads to give more shrinkage resistance and prevent pavement cracking (Cho, Lee, and Ryu 2006).

Strength and stiffness normalisation

The qu, G_o, and E results were normalized, as mentioned in the methods Section (“Specimen preparation”). The results of the normalization of all the blends against the porosity/binder index (ɳ/X_iv^0.32) are shown in Figure 9. The power curves representing the relationship trends show a high degree of significance (R² = 88 and 89%). The resulting correlations significantly contribute to estimating the geotechnical parameters (q_u, G_o, and E) of cement-treated soils with any replacement proportion of TRF, cured for a specific time, through conducting the tests for one blend. Inspired by the literature in similar research recently (Ekinci, Hanafi, and Aydin 2020; Ekinci, Scheuermann Filho, and Consoli 2019; Festugato et al. 2018); the mix is suggested to have a predetermined value of (ɳ/X_iv^0.32 = ∇) close to 25 at which the graphs have been normalized, and it divides the range of the porosity/binder index values. Equation (6)(6) $q_{\mathit{u}\mathit{\hspace{0.17em}}}={q_u}_{\left(\mathit{\eta }/X_{iv}^{0.32}=25\right)}\mathrm{.1405.9}{\left(\eta /X_{iv}^{0.32}\right)}^{-2.25}\mathit{\hspace{0.17em}}{R}^2=0.89$(6) , (7(7) $G_{\mathrm{o}}={G_0}_{\left(\mathit{\eta }/X_{iv}^{0.32}=25\right)}\mathrm{.55.48}{\left(\eta /X_{iv}^{0.32}\right)}^{-1.25}\mathit{\hspace{0.17em}}{R}^2=0.89$(7) ), and (8(8) $\mathrm{E}={\mathrm{E}}_{\left(\eta /X_{iv}^{0.32}=25\right)}\mathrm{.798.28}{\left(\eta /X_{iv}^{0.32}\right)}^{-2.08}\mathit{\hspace{0.17em}}{R}^2=0.89$(8) ) can be used to find q_u, G_o, and E of TRF-cement-clay mix having a predetermined porosity/binder index (∇) and knowing the parameters $(q_u{ }_{\left(\eta /X_{iv}^{0.32}=25\right)},G_0{ }_{\left(\eta /X_{iv}^{0.32}=25\right)},{\text{E}}_{\left(\eta /X_{iv}^{0.32}=25\right)}$ from a tested sample having ∇ = 25.

(6) $q_{\mathit{u}\mathit{\hspace{0.17em}}}={q_u}_{\left(\mathit{\eta }/X_{iv}^{0.32}=25\right)}\mathrm{.1405.9}{\left(\eta /X_{iv}^{0.32}\right)}^{-2.25}\mathit{\hspace{0.17em}}{R}^2=0.89$(6)

(7) $G_{\mathrm{o}}={G_0}_{\left(\mathit{\eta }/X_{iv}^{0.32}=25\right)}\mathrm{.55.48}{\left(\eta /X_{iv}^{0.32}\right)}^{-1.25}\mathit{\hspace{0.17em}}{R}^2=0.89$(7)

(8) $\mathrm{E}={\mathrm{E}}_{\left(\eta /X_{iv}^{0.32}=25\right)}\mathrm{.798.28}{\left(\eta /X_{iv}^{0.32}\right)}^{-2.08}\mathit{\hspace{0.17em}}{R}^2=0.89$(8)

Figure 9. The unconfined compressive strength (q_u), initial shear modulus (G),and elastic modulus (E) normalized with the modified porosity/binder index (η/(X_iv⁰·³²) for all the blends.

The normalization of strength and stiffness against different mix variables combines their effect in a single variable, which is the porosity/binder index (ɳ/X_iv^0.32). In this way, the equations above can be used to estimate the strength and stiffness of TRF cemented clay by performing only a single test for each property at the designated value of (ɳ/X_iv^0.32).

Obtaining geotechnical parameters via non-destructive tests

Nondestructive tests are widely used in geotechnical works; such techniques are convenient, economical, and efficient to be applied and, in this study, an ultrasonic pulse velocity (UPV) test has been utilized as a nondestructive test to determine the initial shear modulus of all the blends. The resulting q_u and E as a function of the G_o are presented in Figures 10 and 11, respectively. The blends shown in Figures 10 and 11 have been compacted to 1600 and 1800 kg/m3 dry densities after curing for 7, 28, and 60 days, respectively, and comprise 7, 10, and 13% cement with 0, 2.5, 5, 10, and 20% TRF. Both relationships are illustrated by powered curves with high significance rates of 87% and 91% (Figures 10 and 11, respectively). Equation (9)(9) $q_{\mathit{u}\mathit{\hspace{0.17em}}}=0.00124{G_0}^{1.73}\mathit{\hspace{0.17em}}{R}^2=0.87$(9) and (10(10) $\mathit{E}=0.00021{G_0}^{1.68}\mathit{\hspace{0.17em}}{R}^2=0.91$(10) ) can be used to obtain the strength ${\mathrm{q}}_{\mathrm{u}}$ and elastic modulus $\mathrm{E}$ of any TRF-cemented-clay blend cured for a specific time by performing only one non-destructive (UPV) test. G_o in both equations can be determined following Equation (7)(7) $G_{\mathrm{o}}={G_0}_{\left(\mathit{\eta }/X_{iv}^{0.32}=25\right)}\mathrm{.55.48}{\left(\eta /X_{iv}^{0.32}\right)}^{-1.25}\mathit{\hspace{0.17em}}{R}^2=0.89$(7) described in the previous Section (”‎Strength and stiffness normalisation”).

(9) $q_{\mathit{u}\mathit{\hspace{0.17em}}}=0.00124{G_0}^{1.73}\mathit{\hspace{0.17em}}{R}^2=0.87$(9)

(10) $\mathit{E}=0.00021{G_0}^{1.68}\mathit{\hspace{0.17em}}{R}^2=0.91$(10)

Figure 10. The unconfined compressive strength (q_u) as a function of the initial shear modulus (G₀) for all the blends considering the cement replacement percentages of (0, 2.5, 5, 10, and 20%) TRF.

Figure 11. The elastic modulus (E) as a function of the initial shear modulus (G₀) for all the blends considering the dry density of the specimens.

Eventually, the essential strength and stiffness properties of TRF-cemented soil can be estimated by performing a single undestructive test. Such method needs to be validated by doing similar studies with different soil types and conditions.

Deformation behaviour

The axial strain results at peak strength of the specimens treated with 7, 10, and 13% of cement and 0, 2.5, 5, 10 and 20% rubber tire fiber replacement are presented in Figure 12. As can be seen in Figure 12, the axial strain at peak increases as the percentage of rubber tire fibers increases. Since the rubber tire fiber was used as cement replacement, the axial strain at peak values was increased with increasing cement content. At 0%, the difference between the axial strains measured at the peak was relatively narrow, increasing with the increase in the percentage of TRF.

Figure 12. Effects of fiber ratio on axial strain at peak strength for different mixtures.

Figure 13 introduces the concept of ductility, further exploring the observed impact of rubber tire fibers. It was observed from the ANOVA analyses that the 2.5% of TRF inclusion prepared at the lowest density (1600 kg/m³) showed higher parameter values in comparison, resulting in a contribution in contrast to unreinforced specimens for all parameters. Therefore, ductility was evaluated on specimens prepared at 1600 kg/m³ density and cured for 28 days. As shown in Figure 13, the ductility of the specimens has improved from 1 to 1.5 as the percentage of rubber tire fibers has risen. The obtained ductility values of the various cement contents are relatively comparable, suggesting that the increase in cement content is not a controlling factor that impacts the ductility of the specimens. Therefore, increases in the TRF percentage increase the ductility of the blends; this was also encouraged by the fact that TRF is being used as a cement replacement. Materials with high ductility can be used as an option to determine crack reflection on highways. Such material can also be used between the base and the surface to provide stress relaxation instead of geotextile or chip seal (Adaska and Luhr 2004; Scullion 2002).

Figure 13. Influence of fiber and cement content on the ductility (D) of different mixtures.

The failure modes of the 2.5 and 20% TRF-reinforced soil, prepared at both densities and cured for 28 days are illustrated in Figure 14. Mid-age range was chosen to perform this study. Once the unconfined compressive tests were complete, the test specimens were inspected to provide visual evidence of the failure modes. At 2.5% TRF-reinforced specimens, a few small cracks become visible at the center of the reinforced samples in a vertical orientation. As observed in the ductility analysis, due to brittle behavior and little plastic material (TRF) contents, samples fail due to the formation of a main crack. However, the 20% TRF-reinforced samples show the formation of multiple shear planes. Agglomerates of clay-cement-rubber form within the specimen after a high percentage of rubber tire fiber reinforcement is added. This leads to the development of preferential failure planes along the contacts of these large lumps, which ultimately leads to a significantly weaker specimen with a bulging failure mode when the TRF content is 20% or higher. As a result, fiber reinforcement up to a particular percentage prevents the formation of a shear zone and modifies the samples’ failure mode. Recent studies also reported similar findings (Freilich, Li, and Zornberg 2010; Özkul and Baykal 2007; Zare et al. 2020).

Figure 14. Images of failure modes in unconfined compressive strength tests for specimens prepared with 2.5 and 20% TRF replacement, both densities and 28-days curing.

Microstructure

Figs. 15(a–f) shows the SEM images of the samples prepared at the densities of 1600 kg/m³ and 1800 kg/m³ and 60-day curing with 2.5% TRF content and 7, 10 and 13% cement content. It is clear from the SEM images that an increase in cement content results in a denser microstructure. Furthermore, as density increases, the pores become smaller, forming a denser structure. It is also clear that the rise of cement content from 7 to 13% results in a more pronounced Portlandite phase as small cylindrical white deposits on SEM images. As expected, samples prepared at lower density (1600 kg/m³) have fewer pores and more hydrated silicate and aluminum phases classified as CSH and CASH, as shown in Figs. 15(a)–(f).

Figure 15. SEM micrographs for blends a)1600kglm3, 7% cement and 2.5% TRF, b) 1600 kg/m³, 10% cement and 2.5% TRF, c) 1600kglm³, 13% cement and 2.5% TRF, d) 1800 kg/m³, 7% cement and 2.5% TRF, e) 1800 kg/m³, 10% cement and 2.5% TRF, t) 1800kglm3, 13% cement and 2.5% TRF at 60-days.

An all-around enhanced flocculated and solidified matrix can be seen in Figures 15(c) and 15(f) due to the interaction between the cement’s positive and the soil’s negative charges. The strength increase is observed due to the clustering of the clay particles and improving bonds amongst soil particles. Interestingly, the pozzolanic reactions begin when the pore fluid reaches a particular alkalinity level and produces a significant quantity of hydroxide particles. The secondary reactions might last for quite a while and are significantly slower than the hydration processes.

SEM images in Figure 16 present a sample prepared at 1600 kg/m³ density, cured for 60 days with 2.5% TRF content and 10% cement content. It can be seen in Figures 16(a,b) that a hard shell developed around the rubber particles through the cement hydration, increasing the stiffness compatibility between rubber and clay. However, as seen in Figures 16(c,d), microcracks and cavities on the surface of the cement might prevent the clay and rubber fibers from sticking together, which might result in the creation of gaps at the interface. With an increase in fiber content above 2.5%, the strength may have decreased due to the poor interfacial contact between the artificially cemented clay and tire.

Figure 16. SEM micrographs for a–b) hard shell development around the rubber, c) cavity and d) microcracks on the clay rubber interface.

Conclusions

This study investigates the effect of TRF as a cement replacement of alluvial clay treated with soil. The objective is to provide an eco-friendly method of disposing of TRF while benefiting from its properties in cemented soils. Various blends of different cement and TRF proportions, compacted at two dry densities and cured for three periods, were experimentally tested. The results have been evaluated, and the outcomes of the discussion can be concluded as follows:

• According to the statistical analysis findings, the most significant factors that determine the q_u, G₀ and E of a tire-fiber-reinforced clay are the dry density and the curing period, respectively.

• 2.5% TRF of partial cement replacement can be selected as the optimum dosage to improve the cemented soils’ mechanical properties. 5% to 10% showed a moderate drop in the blends’ strength and stiffness, whilst a significant decrease was observed for 20%.

• At 7 days curing period, the 2.5 to 10% TRF replacement content has contributed more than cement-only treated samples to the blends’ strength and stiffness. However, 28 and 60-day cured specimens showed a considerable drop in the q_u, G_o, and E, especially at more than 10% TRF content.

• Denser samples showed a higher increase in q_u, G_o, and E than less compacted ones considering the significant particle contacts of blends’ particles in denser samples.

• The porosity/binders index (ɳ/X_iv^0.32) is a decisive factor in determining the mechanical parameters of the artificially cemented tire-stabilized soil.

• The regression model makes a significant contribution since it permits the estimation of geotechnical characteristics of tire-reinforced clay with any cement content, TRF content, density, and curing duration using just one test result. This is a considerable advancement in the field.

• The tire rubber increases the blends’ ductility behavior with high energy absorption, an advantageous characteristic in various practices such as providing stress relation in pavements.

• The hard shell developed around the rubber particles increases the compatibility in stiffness between tire and clay. However, the loss in strength with a rise in TRF content beyond 2.5% may be caused by the cavity and micro cracks on the tire’s surface.

The mixes with different cement and TRF contents can comply with the minimum requirements for rammed earth, specified in various standards as expressed in the introduction section, by providing a lightweight construction material. Moreover, up to 10% TRF content blends satisfy the minimum requirement for the base and subbase cemented soils suggested by various standards. The proposed methods to estimate the mechanical properties of tire rubber cement soils should also be validated by conducting similar research on different soil types and conditions.

Highlights

• The most significant are the dry density and the curing period, respectively.

• 2.5% TRF of partial cement replacement is the optimum dosage.

• Denser samples showed higher increase in q_u, G_o, and E than less compacted ones.

• Porosity/binders index (ɳ/X_iv^0.32) is a strong factor to determine the mechanical parameters.

• The tire rubber increases the blends’ ductility behavior with high energy absorption.

• The hard shell developed around the rubber particles increases the compatibility in stiffness.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The corresponding author may provide some or all of the data utilized upon request.

Additional information

Funding

This project received fund from Middle East Technical University (FEN-20-YG-4).