Failure Mechanism and Optimization of Arch-Bolt Composite Support for Underground Mining Tunnel

Numerical simulation tests were performed on the arch-bolt combined supported mining tunnel through an improved numerical simulation approach..e typical soft rock roadway was took as the background, and the influencing factors such as ground stress level, lateral pressure coefficient, and support type and parameters were considered..e failuremechanismof a semicircular roadwaywith two straight walls was analyzed; results showed that the arch legs’ inward bending deformation and the arch-rock separation are the breakthrough of the global failure of the supporting system, and rock bolts breakage promoted the failure process..e effects of different controlling measures were analyzed including enlarging the bolt diameter, replacing the conventional bolts with energy-absorbing bolts, and setting arch locking bolts on the arch legs..e field test of the concrete-filled steel tube (CFST) arch-bolt composite support scheme was carried out in a high-stress soft rock roadway, and the results indicate the reliability of the main conclusions.


Introduction
In recent years, the mining depth is getting deeper and deeper, no matter in metal mines or in coal mines. ere are about 60 1 km-deep coal mines in China now [1,2]. With the adjustment of China's coal mining strategy, more and more coal mines are constructed in extremely soft rock strata (e.g., in Neimeng and Ningxia in west China), swelling soft rock strata (e.g., in the Longkou mining area in east China), water-rich strata, and other complex conditions. e roadway support is generally difficult in the above-mentioned situation, the commonly used rock bolt-shotcrete support is too weak to maintain the roadway stability. erefore, arch-bolt-shotcrete combined support became a common used reinforced support style for the above conditions. e supporting arches mainly include the traditional U-shaped steel arch [3][4][5] and the new developed concrete-filled steel tubular arch [6][7][8][9]. But sometimes this kind of high-strength support still cannot prevent the roadway deformation and failure.
In a roadway supported with the arch-bolt composite support system, the arches with higher stiffness can hardly couple with the other components, which are relatively flexible, such as rock bolts, anchor cables, and the shotcrete layer. In practice, the supporting members are usually broken down one by one, and the whole supporting system fails step by step. erefore, the key problem in an arch-bolt composite-supported roadway is how to achieve coordination among the various components in the arch-bolt composite support system and achieve coupling between the mechanical behaviors of surrounding rocks and supports. In fact, a lot of research has revealed the interaction mechanism between bolts and rocks in a roadway [10][11][12][13]. Some work has also discussed the failure mechanism of roadways with supporting arches such as steel set with I-or U-shaped cross sections [5,14,15]. Little effort has been spent on the combined support action supplied by bolts and arches on the surrounding rocks or interactions among the above three objects. e basic model of rock-support interaction can be established theoretically, and the fundamental interaction mechanism can be revealed, but in the theoretical method, simplifications cannot be avoided, such as simplifying the roadway to a circular shape and equalizing the supporting action of the arch, shotcrete, and bolt with linear support forces with certain stiffness [3,16]. erefore, the limitations of using the theoretical method to solve this complex problem are obvious. Model testing is an effective research tool for the above problem, but the cost of testing is too high to be used widely. us, the numerical simulation method, which has been greatly improved, is widely used for researching the problem. In the reported numerical simulation research, emphasis was placed on the description of the characteristics of the surrounding rocks [17][18][19][20][21], but considering the mechanical behaviors of supporting members is rare.
To solve the above defects, Li et al. [22] proposed an improved numerical simulation approach through additional development of FLAC 3D . e approach consists of four parts: a yieldable supporting arch module, a separable arch-rock interaction module, a breakable anchor bolt module, and a practicable surrounding rock module. e yield criterion of the supporting arch was proposed and embedded in the modified beam element via the FISH language programming. A separation criterion is proposed for the arch-rock interaction link, and the separable archrock interaction module is realized through modifying the normal-yield attachments of the built-in arch-rock links. A tensile breakage failure criterion is proposed for the anchor bolt using the ultimate elongation ratio of the whole anchorfree part as the criterion. Analysis showed that the improved numerical simulation approach is much more reliable for large deformation roadway behavior with arch-bolt composite support, especially concerning the bearing and failure behaviors of the supporting arch and anchor bolt. In this article, systematic simulation tests will be performed using this simulation approach, and the failure mechanism of the arch-bolt composite support in the mining roadways will be studied, and the optimized control measures will be proposed.

Engineering Background: An Extremely Soft Rock
Roadway. Liangjia Coal Mine is located in the eastern coast of China, with an output of 3000000 t per year, and some mining areas are under the sea. e gateway tunnels of the 4606 working face is the object of this article. e profile of the roadway consists of two straight walls and one semicircular arch. e height of the wall is 1.6 m, and the radius of the arch is 2.2 m. e roadway is located in the 4th coal seam, the top layers are coal and mudstone in turn, and the bottom layer is oil shale (Figure 1(a)). Because the rock mass is extremely soft and fragmented, the stability of roadway surrounding rock is very poor. e vertical in situ stress is 10 MPa, the horizontal major principal stress is 14.5 MPa, and the direction is basically parallel to the roadway axis; the horizontal minor principal stress is 10 MPa. e roadway is supported by U-steel arch + bolt + shotcrete, with a row of 16 bolts every 0.8 m, 2.25 m in length, 18 mm in diameter, 0.8 m in anchorage length, and 20 kN in preforce. e grade of shotcrete is C20, and the thickness is 0.12 m, the spacing of U36-steel (36 kg/m) arch is 0.8 m. Displacement of the roadway floor reached 900 mm, and the value on the side wall reached almost 500 mm at 120 d after excavation, and the rheological characteristics are obvious. Failures occurred on the legs of the supporting arches, as shown in Figure 1(b), and then the arch lost stability as a whole, as shown in Figure 1(c). Subsequently, the roadway surrounding rock deformation continued, and some of the rock bolts broke.

Designation of the Comparing Schemes.
Numerical simulation whose parameters are basically consistent with the above engineering case has been carried out, and the reliability of the numerical simulation method and parameters used were verified [22]. e test schemes shown in Table 1 are designed based on the above premise. e vertical ground stress values change in Schemes 1-5, and the lateral ground stress coefficient λ is set as 1.0; however, in Schemes 6-10, the lateral ground stress coefficient values change, and the vertical ground stress is set as 10 MPa. Here, the lateral ground stress indicates the horizontal ground stress that is vertical to the axial direction of the roadway model. While the horizontal ground stress that is parallel to the axial direction of the roadway model is set at the same value as the vertical ground stress, all of the schemes have the same supporting parameters as shown in Table 1. Because the U-steel arch is too easy to fail, the CFST support with higher strength is selected as the supporting arch type in the schemes, for more convenient comparative analysis. A supporting arch is generally composed of four concretefilled steel tubular components and three casing tubes as shown in Table 1. e supporting ability is much higher than the conventional U-steel-supporting arch and developed rapidly in recent years [9,23].

Modeling of Ground and Roadway.
e total size of the model is width × height × thickness � 40m × 40 m × 0.8 m, and the shape and sizes of the roadway are the same to the real one, as shown in Figure 2 ere is no displacement constrain set to the top surface, but the equivalent ground stress corresponding to the up strata is applied on it. e displacement in the normal direction of the other 4 surfaces are constrained. e mechanical model and parameters of the surrounding rock elements are set according to Table 2, and the ground stress is initialed according to Table 1.

Modeling of the Support Members
(1) Shotcrete layer: the shotcrete is modeled with solid elements with a thickness of 0.12 m, as shown in Figure 2 Figure 2(c). e anchorage parameters of the outer anchorage segment are bulk modulus 10000 MPa, cohesion 10000 MPa, and inner friction angle 25°; the anchorage parameters of the inner anchorage segment are bulk modulus 20 MPa, cohesion 20 MPa, and inner friction angle 45°; the anchorage parameters of the anchor-free segment are bulk modulus 0 MPa, cohesion 0 MPa, inner friction angle 0°, and the links between the cable element of anchor-free segment and rock zones are all deleted, as shown in Figure 2(c). According to the practice situation of anchor bolts in roadways [24], the breakage elongation ratio of the anchor-free segment is set as 10% [22].    Advances in Civil Engineering 3 length 600 mm placed on the crown and the two shoulders (45°below the arch crown point) of the arch. e geometric and the cross-sectional parameters are shown in Table 3, which is determined based on the field practice. e supporting arch was composed of 52 beam elements along the roadway profile.
e compression-bending yield functions used in the supporting arch model are also shown in Table 3, where m equals to bending moment M divided by M u , and n equals to axial force N divided by N u , and M u is the bearing capacity of the pure bending CFST beam, and N u is the bearing capacity of the compression CFST column; M and N are the bending moment and axial force of one section, respectively. k N ′ and k M ′ are valued as k N /1000 and k M /1000, respectively, where k M and k M ′ are the flexural rigidities before and after the bending yielding action, and k N and k N ′ are the axial stiffness before and after the compressive yielding action. e parameters of the separable links are F t � 0.2 kN, F c � − 1 × 10 17 kN, gap � on, and k � 5 × 10 6 , where F c is the compressive yield strength (force units), and F t is the tensile yield strength (force units), k is the stiffness per unit area, gap � on means a gap may form in both the positive and negative directions when yielding occurs in these directions. For more detailed information on modeling parameters, refer to Li et al. [22].       of the side wall, roof, and floor are similar; and the rock bolts are working with full loading. On the 10 th day, the plastic zone boundary exceeded the rock bolts by about 0.7 m (Figure 3(g)), and the deformations of the roadway became larger. In the above process, the axial forces and bending moments of the arch show linear growing pattern (Figures 3(b) and 3(c)). e axial force diagrams in Figures 3(f )-3(i) show that the axial force is highest on the arch crown, and the axial force decreases when the cross section drops, whereas the ultimate bending moment appears on the arch leg, and the value is apparently larger than that on the other sections. e deformation of the arch developed when the inner forces changed; the arch legs bent inward the excavation space because of the surrounding rock pressure. e bending deformation was easy to see at the 10 th day, and the arch leg started to be separated from the rocks (Figure 3(g)).
Because of the above arch-rock separation, the supporting forces of the arch legs reduced to 0 after fluctuation as shown in Figure 3(d).
e arch legs totally lost their support function on the rocks on the 10 th day, which accelerated the deformation and failure process of the side wall rocks and stretched the bolts on the side more significantly than the bolts in the other positions. e bolt located on the half height of the wall broke first on the 11 th day (Figures 3(e) and 3(h)). Also, on the 10 th day, the beam element on the half height of the leg (CID � 357, the CID refers to the serial number of the beam element) failed because the compression-bending yield criterion was satisfied. en, the beam CIDs 401, 358, 402, 356, and 400 (all located on the half height range of the arch legs) failed one by one. e arch legs' bending failure started the global failure of the supporting arch because the bending deformation of the legs developed faster, and the supporting arch began to fail globally. e failure of the legs leaded the axial forces and supporting forces of the arch decreased rapidly after the 10 th day. e values of the axial forces and supporting forces on the monitoring beam elements decreased to 0 after the 33th day ( Figure 3(d)), which means the arch lost its stability and supporting function totally. After that, although there was a rate reduction, the deforming rate of the surrounding rocks was still high. e bolts were almost all broken on the 37 th day, the arch-bolt composite support system failed globally, and the roadway was still deforming on the 50 th day. e above failure process can be summarized as follows: (a) First, the shape of the roadway (semicircular with two straight walls) dictated that the side walls significantly deformed inwardly at the beginning.
is action made the bending moments in the arch legs increase significantly, and the inner forces reached the compression-bending criterion quickly. is led to fast deformation of the arch legs and the global instability of the arch. (b) Second, the shape and the mechanical environment of the supporting arch meant that the deformation rate of the arch legs was greater than the rate of the side wall rocks, which caused the arch-rock separation. is means that the rocks on the side wall of the roadway lost support at an early stage, which caused further serious failure. (c) ird, the rock bolts breakage during the above evolution aggravated the failure of the roadway support. ere are two crucial points in the failure process: the compression-bending failure of the arch legs and the separation between the arch legs and the rocks. e above analysis shows the following support failure characteristics of a semicircular soft rock roadway with two straight walls, supported with an arch-bolt composite support system. (a) e rheological features of the surrounding rock deformation are obvious, and the duration time is long. (b) e serious deformations of the arch legs and the side walls are the critical points that led to the final deformation and failure of the whole roadway. (c) From the point of view of mechanics, the semicircular shape with two straight legs is adverse for stable control of the surrounding rock. e arch leg parts are prone to bending failure, and the macrocharacteristics of the global arch structure means that once the compression-bending failure happens on the arch legs, the support on the side wall rocks will be completely lost. (d) e bolts may be broken easily in this kind of roadway. . Some results of Scheme 1-5 were listed in Table 4, including surrounding rock displacements, plastic volumes, support structure forces, and working status. Figure 4 shows the developing curves of each monitored index of Scheme 2 (ground stress 5 MPa) and Scheme 3 (ground stress 10 MPa). Figure 5 shows the plastic zones of the surrounding rocks and support structure inner force diagrams of Schemes 1-4 with different ground stresses. e deformation and failure process are analyzed as follows.

Influences of Ground Stress (Comparing Analysis on
(1) In Scheme 2 (ground stress 5 MPa), the deformations are smaller, the roadway is basically stable at the 50 th day (Figure 2(a)), and the range of the plastic zone does not exceed the rock bolt inner end ( Figure 5(b)). e 6 bolts on the side walls reached the ultimate axial force but did not break ( Figure 5(b)). e arch inner forces were much lower than the extreme value, and the arch supporting forces are stable and function normally, and there is still a margin remaining. e roadway deformations of Scheme 1 (ground stress 2.5 MPa) were much smaller than those in Scheme 2, the plastic zone boundary did not exceed the rock bolt inner end ( Figure 5(a)), and the other laws shown in the results are similar to Scheme 2.
(2) In Scheme 3 (ground stress 10 MPa), the deformations and failure are much more serious, compared with Schemes 1 and 2. e roadway roof and side walls were still deforming at the 50 th day ( Figure 4F), the plastic zone boundary exceed the bolt inner end by about 1.5 m ( Figure 5(c)), and all rock bolts were fully loaded. e arch legs were bent inward, which gradually reduced the arch supporting force and even reduced the force to 0 on the arch legs, resulting in a resurgence in the rise of the roadway deformation curves, which were almost stable previously. When comparing the roadway deformation curves shown in Figures 3(a), 4(a), and 4F, we see that the greater the ground stress, the worse the control effect of the supporting system and the longer the duration of the deformation. In fact, the rheological rock deformations were basically out of control when the ground stress reached 15 MPa. Comparison of the bolts' axial force curves show that the forces were close to the ultimate capacity when the ground stress was low, the bolts were all fully loaded when the ground stress reached 10 MPa, and the bolts gradually broke when ground stress reached 15 MPa. e supporting arch did not fail when the ground stress was low, but the arch failure happened when the ground stress reached 10 MPa, and it happened as soon as the 10 th day in the 15 MPa scheme. e higher the ground stress, the earlier the failure of the arch, the faster the failure process, and the greater the reduction of supporting forces during the process. Figure 6 shows the variation curves of surrounding rock deformations and plastic volumes of Schemes 1-4. e deformation values and plastic zone volumes all increase with increases in ground stress, and the curves show exponential increases. When the ground stress is higher than 5 MPa, the side wall displacements become larger than the displacements of the roof and floor, and the differences are greater when the ground stress rises. . Some results of Scheme 6-10 were listed in Table 5. Figure 7 shows the evolution curves of each monitored index in Scheme 6 (lateral coefficient λ � 0.5) and Scheme 10 (λ �1.5). Figure 8 shows the plastic zone of the surrounding rocks and support structure inner force diagrams of Schemes 6, 7, 9, and 10. e deformation and failing processes are analyzed as follows.

Influences of Lateral Ground Stress Coefficients (Comparing Analysis on
In Scheme 6 (λ � 0.5), the bolts are fully loaded except the two on the roof, and the arch legs were bent slightly inward but had not failed, which is similar to that in Scheme 7 (λ � 0.75). In Scheme 8 (λ �1.0, the same scheme as Scheme 3), the arch legs yielded on the 15 th day and bent inward, and all bolts were fully loaded but not broken. In Scheme 9 (λ �1.25), all bolts were fully loaded and the arch failed seriously. In Scheme 10 (λ �1.5), there were two bolts broken on the side walls ( Figures 7J and 8(d)) and the arch failed seriously.
When λ is less than 1, the width of the plastic zone is larger than the height, whereas the width is smaller than the height when λ is greater than 1. When λ reaches 1.5, the height of the plastic zone is about 2 times the width. erefore, with an increase in λ, the shape of the plastic zone of the surrounding rock becomes taller and thinner.
When λ is less than 1, there is bending deformation on the arch legs, the inner force curves show that the arch legs did not yield, and the supporting force curves on the crown and shoulders did not show a reducing stage (Figures 7B-D). On the contrary, when λ is larger than 1, the arch yielded and failed and the inner force and supporting force curves show an obvious reducing stage (Figures 7G-I). erefore, the ground stress lateral coefficient λ obviously affects the inner forces and arch deformations, especially the inward bending deformation and yielding failure behaviors. e higher the value of λ, the higher the risk of losing the arch supporting forces.
As shown in Figure 9, the deformation values all increase with increases in λ, and the curves show an exponential increase. e curve slopes are not as significant as in Figure 6. Figure 9 also shows that the floor deformation is most obviously affected by λ, and the side wall is least affected. e volume of the plastic zone of the surrounding rock is larger when λ is 0.5, it is minimal when λ is 0.75, and then, it increases approximately linearly with increases in λ.
e comparison between Figures 9 and 6 shows that the influence of ground stress value on the roadway deformation is more significant than that of lateral ground stress coefficient. e response of the roadway side wall displacement is the strongest. When the ground stress increased from 2.5 MPa to 15 MPa, the roadway side wall displacement increased by 76.8 times. When the lateral ground stress coefficient increased within the common value range, the response of the floor displacement is the strongest, and the floor heave increased by 2.5 times.

Testing Schemes for Different Control Measures.
To explore control measures for roadways supported with an arch- Advances in Civil Engineering 7 bolt composite system that face problems similar to the above schemes, the test schemes listed in Table 6 were designed. Except the parameters listed in Table 6, the other parameters in the schemes are all the same as those in the schemes in Section 2.2. e breakage load of the Φ22 mm bolts is 260 kN, and the broken elongation ratio is also 10%. In Scheme 12, the energy-absorbing bolts (the broken elongation value is greater than the conventional one by 100 mm) are used in the side walls, replacing the conventional rock bolts. e energy absorption principle is mainly absorbing energy through large extension of the anchor-free section of the anchor rod. In this scheme, the energy-absorbing bolt numerical model is actually realized by   Table 7.

Comparing Analyses on Schemes 10-13 (Different
Supports with the Same Ground Stress: 10 MPa, λ � 1.5). Figure 10 shows the plastic zones and support structure inner force diagrams of Schemes 10-13, and Figure 11 shows        (1) e only difference between Scheme 11 and Scheme 10 is that the 18-mm-diameter rock bolts were replaced with 22-mm-diameter bolts, which obviously enhanced the surrounding rock controlling effect. No bolts were broken in Scheme 11, and the side wall inward displacement was reduced by 20%, but the arch still failed. e above results show that the highstrength bolts are necessary for soft rock roadways and explain why high-strength bolts are used widely today.
(2) e difference between Scheme 12 and Scheme 10 is that the conventional bolts on the side walls are replaced by energy-absorbing bolts. e comparison results show the obvious bolt protection effect of the energy-absorbing function, and the broken rock bolt problem was solved in this way. But the global control effect on the surrounding rocks was still weak. (3) Schemes 11-12 show that the broken bolt problem can be effectively solved by enlarging the bolt diameter or by replacing the conventional bolt with an energyabsorbing bolt. But the above measures can do nothing to prevent the failure of the support arch, which is the key point of roadway support failure as analyzed in Section 2.3.2. us, these measures cannot significantly enhance the controlling effect of the surrounding rocks; they can just improve it slightly. (4) Compared with all the other schemes, as shown in Figure 11, the controlling effect of the surrounding rock in Scheme 13 is the best; roof settlement and inward deformation of the side wall are the most convincing parameters, and the values are lower than those of Scheme 10 by 33.9% and 38%, respectively. In addition, the broken bolt and supporting arch failure problems are both solved. Because there are only two more bolts in Scheme 13 than in Scheme 11, the cost of the increased support is low.

Stabilizing Mechanism of the Scheme Using Arch
Locking Bolts (Scheme 13). Figure 12 shows the evolution curves of each index of Scheme 13. e following can be noted when comparing the analysis with Scheme 10 as shown in Figure 7(b).
(1) e axial forces on the arch in Scheme 10 show a risedown pattern, and the reduced axial forces happened first on the arch legs ( Figures 7G, 13(d)) and then on the arch shoulders and crown. Due to the loss of axial forces, the supporting forces supplied by the arch also show a similar phenomenon. As shown in Figure 7I, the supporting forces on the arch legs decreased to 0 before the 10 th day, and the supporting forces on the shoulders and crown of the arch also decreased. e reason can be determined by combining with Figure 13(a) as follows: e arch legs were bent toward the excavation space slightly at the 10 th day, which led to separation of the surrounding rocks and arch legs. Although the bending deformations were not obvious, they led to the loss of supporting forces on the roadway walls. en, the additional bending moments were generated due to the bending deformation of the arch legs (the inner forces on the half height of the legs approaches the compression-bending criterion very quickly as shown in Figure 7H), which caused the failure of the arch legs. After that, the bending deformation of the arch occurred rapidly, the arch lost its global structural mechanical property, and the arch failed overall. (2) In Scheme 13, global failure of the arch did not occur (Figures 13(a)∼13(d)), and the supporting forces remained at a high level until the end of the calculation. e deformation of the rock surrounding the roadway was well controlled because the deformation rate reduced to nearly 0 at the 20 th day. e controlling effects are more obvious on the side wall and roof of the roadway, compared with the results of Scheme 10. e axial force development curves of the arch in Scheme 13 (using locking bolts) are obviously different from that in Scheme 10. e curves show rise-stable pattern with no decreasing stage (Figure 12(a)). Similar pattern are shown by the supporting force curves as shown in Figure 12(c), except the one on the leg half height, which shows a short-term decline between the 3 rd and 18 th days. e working mechanism of arch locking bolts can be obtained through comparing Scheme 13 and Scheme 10. e locking bolts offered great restriction on the inner bending deformation of the arch legs, which limited the increase of the additional bending moment and prevented the strength failure of the section (the bending moments of the arch in Scheme 13 shown in Figure 12 those in Scheme 10). e arch maintained good arch shape, which is a great benefit to fully exploit the supporting ability of the arch. Roof settling was also controlled due to the well-maintained integrity of the arch structure.  zone of the surrounding rocks and support structure inner force diagrams of Schemes 14 at the 50 th day. Figure 14(d) shows the surrounding rock controlling effect columns compared with Scheme 4 (conventional supports with the same condition).

Analysis of the Scheme Using Arch Locking Bolts in a
(1) Because the locking bolts were used, the arch leg bending deformations, as shown in Figure 14(c), were much smaller than that in Scheme 4. e broken bolt problem was relieved but not solved (Figure 14(c)). e surrounding rock deformation and the plastic zone were significantly smaller than those in Scheme 4, especially the side wall displacement.
(2) As shown in Figure 14(c), the locking bolt shows an obvious control effect on the bending deformation of the arch leg. Because the surrounding rock pressure was too high, the arch crown became a new breakthrough which surrounding rock pressure was released, and the arch crown deformed seriously. Finally, the arch failed globally, the axial forces and supporting forces of the arch decreased to a very low level (Figures 14(a) and 14(b)). is is a very interesting phenomenon that the failure position of supporting arch changed as the ground stress rose. It tell us that some of the surrounding rock energy has to be released, especially in the deep ground; it is not a good way to stop the deformation of the surrounding rock blindly.
(3) A comprehensive comparison of  shows that for the soft rock roadway in a lower ground stress condition, the supporting structure and parameter optimization can be adopted as a functional tool for controlling the stability of the surrounding rock. For a higher ground stress condition, a semicircular roadway with two straight walls has obvious limitations compared with the curved wall shape, horseshoe shape, three-centered shape, and circular shape. In some conditions, it is very difficult to control the stability of the roadway only through optimization of the support structures and parameters. Cross-sectional shape optimization should be taken as an important option for roadway support design in high-ground stress conditions. (1) Although locking bolts were used, the U36 steel arch still experienced bending failure (on the lower position in the right leg of the arch). e results indicate that for the supporting structure design in roadway support, even if the potential supporting failure is concerned, the strength of the supporting component still needs serious consideration. (2) Figure 15(c) shows that there was less roadway deformation of the CFST arch scheme than there was with the U-steel arch scheme, which means that the support strength of the CFST arch was significantly improved compared with the U36 steel arch, which has similar steel volume as the CFST arch (150 mm × 8 mm). e CFST arch has much higher applicability in a roadway that has higher surrounding rock pressure or is buried in soft rock ground layers.  371  372  373  374 375 376 377 378 379380  381  382  383  384  385  386  387  388  389  390  391  392  393  394  395

Situations of the Roadway Using the Original Support.
Zhaolou coal mine is located in the middle of Juye Coal area in east China, and the designed production capacity is 3.0 Mt/a. e second downhill roadway of the Zhaolou coal mine is buried more than 1000 m deep, and its service period is 60 years. ere are remarkable geological structures in this area, and the practice area passes through the fault DF7, where the development of vertical tension fractures was significant. e lithology is changeable and mainly consists of mudstone, shale, and coal. e rock quality is poor, and the roadway stability is difficult to control. e initial support scheme of the roadway was rock bolt + steel mesh + shotcrete. But obvious deformation appeared in the roadway, part of shotcrete layer cracked and peeled off, and the initial support scheme failed to maintain roadway stability. In view of this problem, the U36 steel arches were used as the strengthened support (Figure 16(a)). e row spacing of U36 arches was 1.0 m. However, the deformation of the roadway still cannot be well controlled by the strengthened U-steel arch support. e monitoring results show that the U36 arches were distorted (Figure 16(b)), the roof settlement reached 218 mm, the deformation of the arch shoulder reached 322 mm, and the roadway was still deforming at the 150 th day (Figure 16(c)). e deformations of the roadway seriously affected the use of the roadway.

Effects of the New Support Scheme.
Based on the conclusions in Sections 2 and 3 of the article, the U-steel arches were replaced by the CFST arch in the test scheme as shown in Figure 17(a). e parameters of the rock bolts, steel mesh, and shotcrete layer are the same as in the original scheme.   Anchor cables (Φ22 mm × L6200 mm) were added on the roadway floor; the spacing and row spacing are 1500 mm and 1600 mm, respectively. e cross section of the CFST arch is square, the width of the steel tube is 150 mm, the thickness is 8 mm, and the filled concrete grade is C40. e length of the casing joint is 900 mm, and the width and thickness of the casing tube are 180 mm and 12 mm, respectively. e row spacing of the arch is 1 m, the width of the arch is 5 m, the height is 4.3 m, and the height of the arch legs is 1.8 m. Because the arch legs may bend and fail, the locking bolts are installed near the starting point of the arch as shown in Figure 16(c). e diameter of the locking bolt is 22 mm and the length is 3 m (Figures 17(b) and 17(c)). To equalize the pressure on the arch, the space between the arch and the rocks was filled with concrete grade C20. e monitoring devices and sensors were installed during the construction process (Figure 17(b)). Figure 18(a) shows the monitoring results of the new support scheme and the original U36 arch support scheme. e average value of radial pressure of CFST arch at 157 d was 12.57 kN, and the value of U36 arch was 10.14 kN. e average deformation of CFSTarch support roadway was 15.7 mm, only 28% of that of U36 arch support roadway. In addition, the roadway under the control of U36 arch support system was not stable, and the deformation was still continuing and yield failure appeared on some U36 arch. However, the yield failure problem did not appear on the CFST arches because their support strength is higher than the U36 arch. e stress monitoring of the back-filling concrete shows that the stress of the back-filling concrete was far lower than the compressive strength of C20 concrete, which effectively guarantees the safety of the support system. Figure 17(b) shows the roadway 1.5 years after excavation and support. Field test results show that the CFST arch and rock bolt composite support scheme with arch locking bolts provide much higher controlling capacity than the original scheme.

Conclusions
(1) e failure mechanism of the arch-bolt composite supported roadway in a semicircular roadway with two straight walls is discussed. First, side wall rocks deforming inward in the roadway space in the early stage leads to the compression-bending failure of the arch legs and finally induced the global instability of the supporting arch. Second, the arch legs separated from the side wall in the earlier stage caused the loss of the controlling effect on the side walls rocks and the global deformation and failure of the roadway. ird, the rock bolts breakage promoted the deformation and failure process. e side walls of the roadway and the legs of the supporting arch are the key points of failure in a semicircular roadway with two straight walls.
(2) e controlling measures research shows the following: (a) the bolt broken problem can be solved by enlarging the bolt diameter or replacing the bolts with energy-absorbing bolts, but control of the surrounding rock can hardly be enhanced. (b) e locking bolts can significantly restrict the bending deformation of the arch legs, which limited the increase of the additional bending moment, prevented strength failure, and preserved good arch shape so that the arch structure could be fully exploited. (c) In deep or soft rock roadways, the semicircular roadway shape with two straight walls has obvious limitation. It is difficult to control the stability of the roadway solely through optimization of the support structure and parameters. (d) e strength parameters of the supporting structure (especially the arch) are still the most important part of the design, and the CFSTarch is more suitable for a roadway with higher ground stress compared with the U-steel arch. (3) e field test of the CFST arch-bolt composite support scheme with arch locking bolts was carried out in a 1000-m-deep roadway in Zhaolou coal mine, and the results indicate that the conclusions drawn in this article are reliable.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.