Analyses on deformation and fracture evolution of zonal disintegration during axial overloading in 3D geomechanical model tests

2.1. Preparation of rectangular cemented sand models

In China, a distinct zonal disintegration phenomenon was first observed and monitored in a deep tunnel at –910 m level of Dingji coalmine, and the distribution of fracture zones and non-fracture zones are displayed in Fig. 1 [21].

Fig. 1A distinct zonal disintegration phenomenon in Dingji coalmine [21]

Hence, the deep tunnel at –910 m level is taken as the prototype for 3D geomechanical model tests, and capacity of deep rock breakage mechanics and supporting technique model test in State Key Laboratory of deep coal mining and environment protection, which is shown in Fig. 2, is employed to apply stress in three orthogonal directions independently.

Fig. 2Capacity of deep rock breakage mechanics and supporting technique model test in State Key Laboratory of deep coal mining and environment protection

Fig. 3Schematic diagram of rectangular cemented sand model and straight wall arch tunnel (Unit: mm)

In line with the capacity of deep rock breakage mechanics and supporting technique model test, the 3D geomechanical model is a rectangular solid with both length and width of 1000 mm and height of 400 mm. Two rectangular cemented sand models with the size of 1000×1000×200 mm are manufactured and bonded together after mounting strain gauges and fracture wires.

According to the monitoring results of Li et al. [21], the height and width of deep straight wall arch tunnel in Dingji coalmine is 3880 mm and 5000 mm. Then, the height and width of straight wall arch tunnel are set as 155 mm and 200 mm in 3D geomechanical model tests, which is illustrated in Fig. 3. Therefore, geometrical similarity coefficient is set as 25.

The selection of similar material should be ahead of 3D geomechanical model tests. Cemented sand similar material, a granular cementitious material, is adopted for 3D geomechanical model tests and it is a mixture of sand, Portland cement, gypsum, and water [22, 23]. As the density of cemented sand and deep rock is 1.81 g/cm³ and 2.62 g/cm³ respectively. Hence, density similarity coefficient is 1.44. According to Froude’s similarity law, stress similarity coefficient is 36.

In 3D geomechanical model tests, two mix proportions of cemented sand similar material are adopted to simulate two strengths of deep rock, 90 MPa and 120 MPa. The corresponding uniaxial compressive strength (UCS) of cemented sand similar material is 2.48 MPa and 3.31 MPa. In 3D geomechanical model tests, rectangular cemented sand model is marked as GMT-1 for UCS of 2.48 MPa and marked as GMT-2 for UCS of 3.31 MPa. After curing 14 d, cemented sand similar material reaches the design UCS, 2.48 MPa and 3.31 MPa, then rectangular cemented sand model is put into capacity of deep rock breakage mechanics and supporting technique model test to apply the initial geostress in three orthogonal dimensions independently.

2.2. Loading procedure of 3D geomechanical model tests

According to the researches of Zuo et al. [24], Yuan et al. [16] and Gu et al. [17], the maximum principal stress parallel to the axis of deep tunnel is vital for the formation of zonal disintegration. Based on the researches of Zhang et al. [18, 25], the zonal disintegration generates when the maximum principal stress exceeds 1.5 times UCS of surrounding rock mass. Considering the UCS of cemented sand similar material, the maximum applied stress along the axial direction of excavated tunnel is set as 4.96 MPa, which is 2.0 times UCS for GMT-1 and 1.5 times UCS for GMT-2. Two 3D geomechanical model tests are carried out in a loading procedure of first loading to initial geostress, then excavating the tunnel with blasting construction, and finally overloading the stress along the axis of excavated tunnel with the stresses on the other two directions constant. The loading procedure for GMT-1 and GMT-2 is the same and is shown in Fig. 4.

Fig. 4Procedure of 3D geomechanical model test

According to features of capacity of deep rock breakage mechanics and supporting technique model test, the rectangular cemented sand model is put in with the surface 1000×1000 mm side up. Seen from Fig. 3, stress ${\sigma }_z$ is applied along the axis of excavated tunnel, stress ${\sigma }_x$ and ${\sigma }_y$ are applied on the vault and sidewall of excavated tunnel respectively. The burial depth of simulated tunnel is 955 m, and unit weight of rock mass is 26.2 kN/m³. Therefore, initial geostress is achieved by applying $\gamma h$ for both ${\sigma }_x$ and ${\sigma }_y$ and 1.5$\gamma h$ for ${\sigma }_z$. Considering stress similarity coefficient, both ${\sigma }_x$ and ${\sigma }_y$ is equal to 0.69 MPa, and ${\sigma }_z$ is equal to 1.04 MPa. After stabilizing the applied initial geostress about 60 min, 400 mm model straight wall arch tunnel is excavated from top to bottom by blasting excavation. After blasting excavation about 60 min, axial overloading initiates from 1.04 MPa to 4.96 MPa. Axial stress overloads in a step of 0.41 MPa after first overloading to 1.66 MPa and keeps 10 min for each step.

2.3. Blasting excavation for model tunnel

Two blasting constructions are implemented for 400 mm model tunnel with an interval of 60 min. To control the range of blasting crater, 50 mm model tunnel is digging before blasting excavation [6]. After each blasting excavation, manual digging is adopted to shape the blasting tunnel. Layout of blastholes in blasting excavation for GMT-1 and GMT-2 are shown in Fig. 5.

Fig. 5Layout of blastholes in blasting excavation (Unit: mm)

a) GMT-1

b) GMT-2

Considering the different UCS of cemented sand similar material, three vertical blastholes are drilled for GMT-1, while six vertical blastholes with central blasthole uncharged are drilled for GMT-2. During blasting excavation, each blasthole with a diameter of 10 mm and a depth of 140 mm is charged with one electric detonator and the rest length of blasthole is filled with mud [20]. Electric detonators with diameter of 6 mm and length of 35 mm are charged with 0.3 g RDX. Protective measures, such as covering with the felt and thick rubber mat, are adopted to ensure safety during blasting excavation. After blasting excavation and manual digging, the final excavated model tunnel is shown in Fig. 6.

Fig. 6Final excavated model tunnel

2.4. Arrangement of measurement points around excavated tunnel

To monitor the deformation evolution around excavated tunnel during axial overloading, both strain gauges and fracture wires [26] are arranged at the vault, sidewall, and floor of excavated model tunnel on the top surface of bottom rectangular cemented sand model, as shown in Fig. 7.

Seen from Fig. 7, there are 8 strain measuring points at the vault and right sidewall and 9 strain measuring points at the floor. The space for first 6 strain measuring points is 20 mm, then it becomes 50 mm. Each strain measuring point is mounted two BX120-5AA foil resistance strain gauges, one in radial direction, the other in tangential direction. The strain gauges are connected with two XL2010G40 high-speed static strain indicators with a sampling interval of 5 s. The 4 strain measuring points with a space of 50 mm at the left sidewall are used for collecting blasting strain signals.

Due to the existing of the strain gauges, fracture wires are placed about 10 mm away from the center line. Fracture wire is a 190 mm long slim tin foil paper. There are 20 points on each fracture wire with a space of 10 mm. Adjacent two points are in a parallel connection with an LED light in a direct-current circuit. As the resistance of slim tin foil paper is very small, the LED light is off at the beginning. When the LED light is on, it indicates a crack between two points connecting with that LED light.

Fig. 7Arrangement of strain measuring points and fracture monitoring points around the excavated tunnel

a) Schematic diagram (Unit: mm)

b) Physical photo

2.5. Experimental procedure of 3D geomechanical model tests

3D geomechanical model tests are carried out in the following procedure.

Firstly, make two rectangular cemented sand models with size of 1000×1000×200 mm.

Secondly, mount strain gauges and fracture wires on the top surface of bottom rectangular cemented sand model.

Thirdly, bond two rectangular cemented sand models together and put them into the capacity of deep rock breakage mechanics and supporting technique model test.

Fourthly, connect strain gauges with XL2010G40 high-speed static strain indicators and fracture wires with LED circuit.

Fifthly, apply the initial geostress to 3D geomechanical model.

Sixthly, excavate model tunnel by blasting construction.

Seventhly, overload the stress along the tunnel axis.

Finally, take out the rectangular cemented sand model and cut it to observe the fracture distribution.

3.1. Strain evolution during axial overloading

Strain measuring results at point 4 of the vault for GMT-1 are taken to investigate the strain evolution during axial overloading process. Fig. 8 presents the time history of strain in both radial direction and tangential direction at point 4 of the vault. In Fig. 8, the start time is shift to zero at the beginning of axial overloading process.

Seen from Fig. 8, absolute value of strain in both radial direction and tangential direction increases with the growth of axial stress. Furthermore, strain in radial direction is a tensile strain, which indicates a tension. While strain in tangential direction is a compressive strain, which indicates a compression. Therefore, the surrounding rock around excavated tunnel is in a radial tension and circumferential compression stress state. According to the research of Wang et al. [27], when surrounding rock around the excavated tunnel in a stress state of radial tension and circumferential compression, parallel ring fracture is very easy to form and develop. Before axial stress overloading to 3.72 MPa, which is 1.5 times UCS for GMT-1, the strain variation for both radial direction and tangential direction is slow. While when axial stress exceeding 3.72 MPa, the strain variation becomes dramatic, which indicates a possible and potential ring fracture around excavated tunnel.

Fig. 8Strain evolution during axial overloading at point 4 of the vault for GMT-1

a) Radial direction

b) Tangential direction

3.2. Radial strain distribution around excavated tunnel

According to the research of Yuan et al. [16], the primary cause of zonal disintegration is the tension breakage in radial direction under high axial geostress. When axial stress overloading to 2.48 MPa, 3.72 MPa, and 4.96 MPa, radial strain distributions around excavated tunnel for GMT-1 are shown in Fig. 9. $L/B$ represents the ratio of distance $L$ between measuring point and tunnel boundary to width $B$ of excavated tunnel.

Fig. 9Radial tensile strain distribution around excavated tunnel for GMT-1

a) 2.48 MPa

b) 3.72 MPa

c) 4.96 MPa

Seen from Fig. 9, there are always a tension for radial strain around excavated tunnel due to the unloading effect. And radial tensile strain increase with continuous overloading of axial stress. With continuous overloading of axial stress, ring fractures appear due to large radial tensile strain. As a result of the unloading effect, the radial tensile strains near the excavated tunnel are much bigger than those away from it. For the existing free surface of excavated tunnel, the radial tensile strain at point near the boundary of excavated tunnel is small due to stress release after the appearance of first ring fracture. While in the inner part of surrounding rock, the free surface effect is weakened with the increasing distance from excavated tunnel boundary. At the end of axial overloading, an interval distribution of peaks and troughs are shown in the radial tensile strain distribution at the vault, sidewall, and floor of excavated tunnel with the increase of $L$/$B$. When ring fracture occurs, radial tensile strain at points in fracture zone goes through a rapid decrease due to stress release, while radial tensile strain at points near fracture zone undergoes an evident increase due to stress distribution. Therefore, peaks of radial tensile strain distribution correspond to fracture zones and troughs of radial tensile strain distribution correspond to non-fracture zones. For instance, the large radial tensile strains at point 3 and point 5 of the floor correspond to two ring fracture zones at the floor.

To investigate the influence of rock strength on zonal disintegration, radial tensile strain distribution around excavated tunnel for GMT-2 is shown in Fig. 10 when axial stress overloading to 4.96 MPa.

Fig. 10Radial tensile strain distribution around excavated tunnel for GMT-2

Comparing Fig. 9(c) with Fig. 10, radial tensile strain distribution for GMT-2 is a little difference with that for GMT-1. Due to the unloading effect, radial strains are all tensile strain, which indicates a radial deformation towards the excavated tunnel. Moreover, there is also an interval distribution of peaks and troughs around excavated tunnel for GMT-2. Similar with GMT-1, peaks of radial tensile strain distribution correspond to fracture zones, and troughs of radial tensile strain distribution correspond to non-fracture zones.