1. Introduction
Fault ruptures are caused by sudden movement on a fault, when stress on the edge overcomes the friction, and energy in waves are released and travel through the soil. In recent years, there has been an increasing amount of research on this topic due to several major incidents, such as fault ruptures in Taiwan in 1999, Turkey in 1999, the USA in 2002, Japan in 2004, China in 2008, and in Iran in 1990 [
1,
2,
3,
4,
5,
6]. Furthermore, in 2015, an earthquake with a magnitude of 6 occurred in Sabah, Malaysia, which was recorded as the second largest earthquake in the country after the 1991 earthquake in Ranau, which was also in Sabah [
7]. A tectonic map of fault activity revealed that Sabah’s fault was part of a 200 km-long system of normal faults that crosses the eastern side of the Crocker Range, parallel to Sabah’s northwest coastline [
8].
Two types of physical models, namely, centrifuge and 1g models, have been used in previous studies on fault ruptures [
9,
10]. The centrifuge model has the ability to change the centrifuge’s acceleration,
Ng, where
N stands for the scaling factor and g stands for the acceleration of gravity. It means that a unit weight of soil in the centrifuge is
N times the unit weight at 1g [
11,
12,
13,
14,
15,
16]. It involves changing gravity to simulate a real situation. However, fault simulations with centrifuge tests were mostly used to investigate free fields (fields without any structures) and cohesionless soil [
13,
15,
17]. Several studies have examined the relationship between shear zone and damages, and the findings show that soil experiences the highest stress and strain in the shear zone, which causes damages to soil. Lee et al. used the centrifuge model for an experimental investigation to explore the shear zone when faults happen in a free field, and he showed that soil density can have a major influence on ground surface displacements [
18]. The 1g model is more preferred to be used for simulating the effect of soil properties on tunnels induced by faults. Firstly, unlike the centrifuge model, which is very sensitive to achieve certain gravity, natural gravity is used in the 1g model. Secondly, the process of increasing the vertical component in the 1g model can be manipulated, and also, more gauges can be installed in the 1g model due to its larger size. Thirdly, the effects of various soil properties can be investigated more accurately as it involves greater mass of soil.
Investigation on the effects of soil properties alone shows that deformations of wet and dry soil when a fault happens are different. It was found that wet soil shows a major increase in piston pressure compared to dry soil, illustrating that wet soil increases the soil strength [
19]. Soil density is another important aspect of soil failure. The shear zone in low-density soil is typically more complex than in high-density soil, and as the relative density of the ground model becomes more significant, the number of rupture planes decrease because soil strength does not allow more fractures in the soil [
9,
11]. Furthermore, soil properties such as cohesion, friction angle, particle size, density, and water contents are very important in any simulation and analysis involving soil [
20,
21]. Results from previous studies have shown that changing of soil friction angles and cohesion could affect the bearing capacity of soil [
22,
23]. The results from Khezri et al. showed that cohesion is very important to be measured in underground structures because cohesion increases with depth and each layer of soil has different cohesion [
24]. Furthermore, soil particle size and water content have been claimed to have critical impacts in designing foundations, roads, and other infrastructure [
25,
26]. Moreover, normal and reverse faults would also have different impacts on soil deformations. When a normal fault happens, at least one rupture propagation reaches the ground surface and damages building foundations, and there could be more impacts on the soil [
10,
27].
Tunnels and subways will be the first structures to experience damages when any movements happen, before the forces caused by faults reach the ground surface. Hence, the stability of tunnel structures should be given more consideration by engineers. Millions of people use underground public transportation every day, and hence, fault ruptures could pose a major hazard to their lives. Restoration of structures is another challenge because it can be very costly and is sometimes impossible. Thus, more research in this area is needed to reduce the damages to tunnels and buildings. Besides fault characteristics, soil properties also impose major influences and need particular attention. Several studies have previously been conducted in fault ruptures, in which none have considered soil properties as one of the major aspects of tunnel displacement. The simulation of normal and reverse faults can be shown by lifting the bottom of the physical model upward or downward, for simulating reverse and normal faults, respectively. Researchers have evaluated the limitations of displacements, and different data have been found, showing that
% (h stands for movements of the hanging wall and H stands for the depth of the sand box) can produce the range of 4% to 10%. The displacement will cause some changes to ground surfaces, which are different based on soil properties [
27,
28,
29]. In this study, the ranges were chosen between 2.7% to 8%. In previous studies, the effect of displacements was only measured on the ground surfaces; while in this study, the displacements also have been measured on an embedded tunnel.
In this paper, an experimental investigation was conducted to explore the effects of various soil properties, such as cohesion, water content, particle size, and friction angle, and various fault properties, such as fault types, angles, and tunnel depth, by using a 1g model. Furthermore, differences and similarities of normal and reverse faults, the effects of various fault angles, and tunnel depths on tunnels have also been investigated. Plane strain and axial symmetry (PLAXIS) software is a useful tool for analyzing two- and three-dimensional models, and has been used for different circumstances, such as modeling rock mass parameters, tunnels, foundations, and many more. In this paper, PLAXIS software was used for simulating the effect of faults on soft soil and rock, and also the influence of foundation on tunnel displacements. Furthermore, unlike other studies that only investigated free fields, this research focuses on a field with a tunnel. The outcome of this study is believed to help engineers in understanding the level of damage that might be imposed on tunnels at the area of a fault, with various properties of soil and faults.
2. Methodology
A physical model of 3000 mm length × 1000 mm width × 1000 mm height was fabricated in the Geomechanics laboratory, UPM, to simulate both types of normal and reverse faults, following the established boundary conditions of B = 4H, where H stands for thickness and B is the length of the physical model. The physical model consists of stable and movable parts, which can be moved upward and downward using a hydraulic jack to simulate reverse and normal faults respectively, see
Figure 1.
Three different water contents, particle sizes, cohesions, and friction angles, were assessed to evaluate the effects of various soil properties on the tunnel when a fault happens (
Table 1,
Table 2,
Table 3 and
Table 4). The fault angle of 60° remained constant in all tests at various soil properties and the soil type based on unified soil classification system (USCS) is poorly graded sand (SP). The natural water content was determined to be 5% and specimens with 2% water content were prepared by letting them air-dry in the laboratory for two weeks. Specimens with 9% and 12% water content were prepared with different amounts of water, as shown in
Table 5. To achieve a certain percentage of water content, after conducting several tests, the results showed that 95 kg of water could increase the soil water content of 3.7 tons of soil by approximately 3%. A compaction test was conducted to find the optimum water content (OMC), and the result obtained was an OMC of 13.9, as illustrated in
Figure 2. The reason for choosing the maximum water content of 12% was to achieve a soil density of 80% and soil friction angle of 39°.
The effects of fault angles was studied at 30°, 60°, and 90°, see
Table 6, and
Figure 3 illustrates how they were employed in this study. Furthermore, the tunnel depths were chosen to be located at 250 and 500 mm from the surface, see
Table 7. The values of soil properties for the soft soil and rock (in the United State Army Corps of Engineers, categorized as loose and dense sand) evaluated in PLAXIS have been collected from case studies, see
Table 8 [
23,
30]. The tunnel displacements were recorded by three strain gauges, and five linear variable differential transformer (LVDT) were used (
Figure 4) on top of the physical model. The five LVDTs were installed to show the displacements of the tunnel, and the three strain gauges were installed to show the ground surface displacements. Most of the gauges were installed in the middle of the physical model because forces are located where the fault is located, and hence the displacements could be recorded at its maximum.