Research on Key Technologies of Construction Of Tunnel in Aeolian Stratum: A Case Study of Shenmu No. 1 Tunnel

: The naturally formed aeolian sand dunes in northern Shaanxi have unique 12 engineering characteristics. Several difficulties restrict the construction of road tunnels 13 under this stratum, such as the poor self-stabilization ability of the surrounding rock, 14 difficultly in injecting grout, and insufficient construction experience. Therefore, in this 15 study, a case study of the Shenmu No. 1 tunnel was conducted to investigate the 16 engineering characteristics of aeolian sand tunnels, compare the grouting effects of 17 commonly used grouting materials, and discuss the reinforcement effects of different 18 construction schemes in aeolian sand tunnels. Based on a field grouting test, it was 19 found that it is difficult to inject ordinary cement grout into an aeolian sand layer; 20 superfine cement grout and modified sodium silicate grout can be injected, but the 21 former has a poor reinforcement effect. Through numerical analysis, it is found that an 22 approach based on a concept of “horizontal jet grouting pile + benching partial 23 excavation method with a temporary invert” is suitable for the construction of tunnels 24 in aeolian sand in China. 25


Introduction 27
Deserts and sandy land cover an area of total 1.54 million square kilometers in China, 28 accounting for approximately 16% of the total territorial area. They are distributed in 29 Xinjiang, Inner Mongolia, Qinghai, Gansu, and other northwestern regions, and 30 accounting for 95.37% of these areas. An aeolian sand series results from the flow and 31 movement of sand, and is characterized by its strong fluidity and ease of slip (Zheng, has completed aeolian sand tunnel projects such as the Xingshumao Tunnel, Jingpeng 36 Tunnel, and Qiansongba Tunnel. However, because aeolian sand has strong fluidity and 37 significantly different engineering properties from those of general strata, the factors to 38 be considered during the design phase, construction phase, and accident management 39 are relatively complicated. Accordingly, progress on projects in aeolian sand strata has 40 been slow. summarized the mechanical characteristics and parameter change laws of aeolian sand. 53 Zhang (2020) conducted experimental and numerical analyses to study the influence of 54 sand deposition on a track structure. Li (2020) studied the utilization of aeolian sand 55 for concrete production, and analyzed its workability and mechanical properties. 56 Factors such as sand leakage and sliding in the aeolian sand stratum can easily cause 57 safety accidents such as large deformations of supports, collapses, and roof falls. As 58 such, the construction of aeolian sand tunnels is considered very difficult and risky, and 59 essentially can act as a bottleneck in engineering construction. In addition, there is no 60 special and complete construction plan for aeolian sand. The responses to 61 corresponding emergencies are insufficient. More generally, there is no well-formed, 62 systematic, and available construction mode. At present, the number of aeolian sand 63 tunnels being built (or that have been built) is relatively small; in most cases, the 64 excavation can be barely be completed, and usually requires some means of strong 65 support. However, with the increase in the number of tunnels being constructed in 66 aeolian sandy strata in China, the shortcomings (such as insufficient construction 67 experience and the lack of a well-formed systematic construction mode) have become 68 increasingly prominent. Therefore, relevant research is urgently needed to guide the 69 design and construction of tunnels in aeolian sand strata. sand tunnels based on a fuzzy evaluation method, expert investigation method, and 81 analytic hierarchy process, and determined risk indicators. Us Desert. The aeolian sand in that location is generally brownish-yellow, loose, and 106 slightly wet. It is primarily composed of feldspar (73%) and fine quartz sand (23%), 107 followed by silt sand and silty soil, which are concentrated at the tunnel exit and in the 108 surface above the tunnel, with a thickness of the covering layer between 15 and 35 m.

109
The main physical and mechanical indexes as obtained through field experiments are 110 listed in Table 1, and the particle size distributions are listed in From the physical and mechanical indexes and gradation of the aeolian sand, it can be 124 seen that its particles have the characteristics of a small cohesive force, poor gradation, 125 low compressibility, inclination to disturbances owing to excavation, strong water 126 permeability, and a relatively low shear strength. 127

Mechanical properties 128
(1) Compressibility 129 Aeolian sand has a single-grain structure, and its compression primarily depends on the 130 rearrangement and fragmentation of the particles. Under the action of low pressure, the 131 particles slip and roll, making the soil denser and more stable. The amount of 132 compression is determined by the frictional resistance between the particles against 133 displacement. The better the gradation and the higher the density, the greater the 134 resistance and the smaller the compression deformation. The compression process of 135 aeolian sand is almost instantaneous sinking, followed by long-term deformation with 136 a deceleration rate; this represents the process of gradually adjusting the position of the 137 particles to overcome resistance. The compressibility coefficient of aeolian sand is 138 generally small (less than 0.1 MPa -1 ), showing low compressibility. 139 (2) Strength characteristics 140 The strength characteristics, especially the density values under various water contents, 141 play an important role in grout injection. Under the same water content, when the dry 142 density of aeolian sand is higher, the corresponding internal friction angle and cohesion 143 are higher. Under the condition of the same dry density, when the water content is 144 smaller, the internal friction angle and cohesion are smaller. Furthermore, the higher 145 the dry density, the higher the shear strength. Under the same dry density and vertical 146 pressure, the water content has little effect on the shear strength; the difference in the 147 strength value is approximately 10 kPa. Under the influence of capillary force, given 148 that the vertical pressure is small, when the water content is less than a certain value 149 (14%), the shear strength increases with an increase in water content, and when the 150 water content is greater than a certain value (14%), the shear strength decreases with an 151 increase in water content. tunnel was constructed using a shallow tunneling method, with a steel arch, steel mesh, 158 and shotcrete as the initial support, and molded concrete as the secondary lining. The 159 horizontal profile is shown in Fig. 1. 160 According to the results from a ground survey, drilling, and geophysical prospecting, 161 there is Quaternary Holocene aeolian sand (Q4 eol ) overlying the tunnel site area, and 162 Triassic fine sandstone underlying the tunnel. The soil layers from top to bottom are as 163 follows: aeolian sand layer, and then fully weathered-strongly weathered fine sandstone. An advanced small pipe grouting support method was adopted in the early construction stage of the Shenmu No. 1 tunnel, in combination with a four-step excavation method. 169 During the construction, severe sand leakage and sand sliding occurred in the tunnel 170 face and side walls; the primary support sank overall and invaded the tunnel clearance, 171 and a depression cone (up to 10 m in diameter) and cracks (up to 3 cm wide) appeared 172 on the surface ( Fig. 2 and Fig. 2). There were two potential causes considered for the 173 sand leakage and development of cracks: (1) the sliding surface of the sand body 174 extended to the ground, indicating that the sliding surface of the sand body may have 175 exceeded the scope of the advance support and pre-reinforcement; and (2) the grouting 176 of the advance support failed to achieve the designed reinforcement effect, and the soil 177 above the vault slid into the tunnel along the gap between the grouting pipes. In summary, the key problem in the construction of aeolian sand tunnels is sand leakage.

181
This section discusses field grouting tests conducted using different grouting methods 182 and materials. The geological characteristics, grouting mechanism(s) of aeolian sand 183 strata, and existing problems in the grouting process are summarized and studied, 184 hoping to promote the further development of grouting technology, expand its 185 application range, and solve technical problems in similar projects. 186

Field grouting test 187
With respect to the geological conditions of the Shenmu No. 1 tunnel, the effect of 188 grouting is a key factor in controlling the surface settlement and ensuring construction 189 safety. Simply using an engineering analogy method and/or semi-empirical engineering 190 method to determine the grouting parameters will bring great risks. Therefore, to better 191 understand the grouting characteristics of the formation, obtain the necessary technical 192 and economic data, and demonstrate the rationality of the grouting scheme, a 193 representative section should be selected for field grouting tests before grouting 194 construction (Fig. 4). were adopted to carry out in-situ tests in the three pilot areas. All grouting construction 211 was completed on the first day, and the excavation was conducted the next day to 212 examine the grouting effect. 213 During the test, the value of the grouting pressure gauge increased rapidly and reached 214 the final pressure value within a short time (10 s-20 s). Once the grouting volume no 215 longer changed, it was found that the final volume was much smaller than the design 216 grouting amount. The grout failed to spread, and did not meet the design reinforcement 217 requirements. After excavation, it was revealed that there was no penetration of grout 218 into the aeolian sand layer. However, grout clusters with diameters of 2-10 cm appeared 219 around the holes of the small tremies. These grout clusters failed to form a whole 220 Although the grouting amounts of the three grouts were different, the reinforcement 222 effects were not significantly different. Ordinary Portland cement grout has difficulty 223 permeating in to aeolian sand formations; this is because ordinary Portland cement has 224 a large particle size, and is therefore difficult to inject into an aeolian sand layer with 225 pores of 10 μm. Furthermore, the cement particles are suspended in the slurry; thus, 226 even if the particles are smaller than the pores of the sand layer, owing to the filtering 227 effect of the sand layer, the penetration range is extremely small, and sometimes the 228 slurry consolidates around the grouting holes. Therefore, based on the principle of 229 particle size matching, this grout was considered as only suitable for broken rock layers 230 or coarse gravel sand layers, and was not suitable for aeolian sand layers. (2) Superfine cement grout 234 At K91+245 of the left tunnel, the ordinary cement was changed to superfine cement 235 for testing. Three grouts with various water-cement ratios, that is, : wc mm = 1:1, 1.5:1, 236 and 2:1, were adopted to conducted in-situ tests in the three pilot areas (Fig. 6). 237 238 Fig. 6 In-situ test of superfine cement grout 239 During the test, the value of the grouting pressure gauge increased rapidly and reached 240 the final pressure value in a relatively short time (20 s-30 s). Once the grouting volume 241 no longer changed, it was found that the final volume was smaller than the design 242 grouting amount. The grout failed to spread well, and it could not meet the design 243 reinforcement requirements. After excavation, it was revealed that a heterogeneous 244 penetration of grout existed in the aeolian sand layer, and that the radius of penetration 245 was small. Similar to the results from the test of the ordinary Portland cement grout, 246 grout clusters with diameters of 2-10 appeared cm around the holes of the small pipes. 247 Although the grouting amounts of the three grout were different, the reinforcement 248 effects were not significantly different. The superfine cement grout had a small particle 249 size, with an average particle size of 4 μm. Theoretically, it could be injected into the 250 aeolian sand layer with a pore size of 10 μm; however, the difference in the 251 reinforcement effect compared with that of ordinary cement grout was not large, 252 primarily because the superfine cement grout was prone to sediment, and the self-253 stability was poor. Superfine cement grout can penetrate aeolian sand formations, and 254 could potentially be adopted as a high-quality grouting material; however, its stability 255 and grouting technology need to be further improved to make good use of its advantages. 256 (3) Modified sodium silicate grout 257 After the superfine cement grouting test was completed, to avoid affecting the normal 258 construction in the tunnel, a modified sodium silicate grouting test was selected for the 259 roadbed outside the exit of the right tunnel. In particular, φ50 small pipes were placed 260 horizontally along the side slope of the roadbed. The pipe length was 4.5 m and the 261 spacing was 30 cm, for a total of 15 pipes. The test was divided into three groups, with 262 five small pipes in each group (Fig. 7). 263 During the test, the injectability of the grout was significantly improved, and the 264 permeability was better than the above two grouting tests in the aeolian sand formations; 265 however, the gelation time of the modified sodium silicate grout was difficult to control, 266 and the grouting process was complicated, causing the grout volume to be less than the 267 design grouting amount. 268 269 Fig. 7 In-situ test of modified sodium silicate grout 270 After excavation, it was revealed that there was uniform grout penetration in the aeolian 271 sand formation. Grout clusters with diameters of 8-10 cm around the holes of the small 272 pipes appeared; the strength of these was so low that they could be broken if subjected 273 to pressure by hand. Furthermore, the grout condensed into white flocs after being 274 exposed to air for a period of time. silicate grout are listed in Table 3. 281 The sand specimen was cured in air for two days. Its uniaxial compressive strength was 284 0.2 MPa, and its permeability coefficient was 6.83×10 -6 ; thus, it was essentially an 285 impermeable body. In addition, the durability values of the sand specimen were 286 different after curing in water, air, and buried sand. The sand specimen was buried in 287 the water and sand layers, and its strength did not decrease after three months. After 288 curing in air for one day, white crystals appeared on the surface of the specimen; 289 moreover, the surface was loose, and peeled off after one week. Increasing the fluidity of a grout is an effective measure for improving its permeability.

294
One commonly used method for cement grout is to increase the water-cement ratio.

295
This is generally based on using large water-cement ratio cement suspension, especially 296 for grouting in smaller cracks, so as to improve the fluidity and dispersion of the grout. 297 As shown in Table 4, when the water-cement ratio of the superfine cement grout was 298 1:1, the grouting volume was 16 L, and when the water-cement ratio was 1.5:1, the 299 grouting volume was 28.5 L. For ordinary Portland cement grout, the small increase 300 was ascribed to the poor injectability of the grout, i.e., the ordinary cement had large 301 particles, and could not penetrate into the gaps of the aeolian sand formations. 302 303 The deposition for the cement grout with the high water-cement ratio occurred 306 throughout the movement in the fissures. As the deposition thickness increased, the 307 pressure transmission and flow velocity changed until a certain section was blocked 308 and closed. Both the grouting pressure and consistency affected the compactness of the 309 filling body. When the water-cement ratio reached a certain level, the fluidity of the 310 cement grout was no longer significantly improved, as the deposition process had 311 become a controlling factor affecting its movement in the fissures. Increasing the water-312 cement ratio to a very high level did not make much sense for improving the fluidity of 313 the grout. As shown in Table 4, when the water-cement ratio of the superfine cement 314 grout increased from 1.5:1 to 2:1, the grouting volume only increased by 4.5%. 315 Table 5 shows the grouting volumes of the three different grouting materials for a single 317

Analysis of the injectability of different materials in aeolian sand layer 316
pipe. The water-cement ratios of the ordinary cement grout and superfine cement grout 318 were 1:1 and 1.5:1, respectively. With regard to injectability in aeolian sand formations, 319 the chemical grout was stronger than the suspension-type grout, and the grouting 320 volume was three to four times that of ordinary cement grout. 321 It can be seen from Table 5 that the permeability of the modified sodium silicate was 322 the best at up to 10 cm, followed by superfine cement at approximately 4 cm, and 323 ordinary cement grout had the smallest penetration radius of approximately 2 cm, i.e., 324 it was almost non-permeable. This was related to the type and particles of the grout. 325 In general, the permeability of a suspended grout composed of solid particulate 328 materials, for example, cement, clay, and fly ash, primarily depends on the particle size 329 and fluidity of the grout. When the sizes of the cracks or pores are smaller than the 330 diameter of the grout particles, effective grouting cannot be implemented. It is generally 331 believed that only when the cracks or pores are more than three times larger than the 332 coarsest particles in terms of size can it be used for grouting. If they are smaller than 333 this limit, the coarsest particles may be blocked in the cracks; thus, they will rapidly 334 form a filter layer, so that other smaller particles cannot penetrate. At present, the 335 ordinary cement produced in China has a particle diameter of approximately 50 μm as 336 the main component, and the thickest particles reach 80 μm. The particle diameter of 337 the aeolian sand of the Shenmu No. 1 Tunnel is primarily distributed between 75 and 338 500 μm, and the distance between particles is 10-20 μm. As such, particles of ordinary 339 cement grout cannot be injected; the grout consolidates around the steel pipe, and 340 cannot penetrate and diffuse. 341 The average particle size of the superfine cement grout was 4 μm. Therefore, according 342 to theoretical calculations, it could be injected. However, owing to the small particles 343 of the grout, segregation and sedimentation during the penetration process tended to 344 occur, and grout accumulated at the entrance of the fissure to form a grout layer. 345 Therefore, the penetration depth was limited, and the reinforcing effect was not 346 satisfactory. 347 Compared with suspended grout, the modified sodium silicate grout had low viscosity 348 and the best pourability, but the gel time had a great influence on the permeability. The 349 control of the gel time should be strengthened to obtain an ideal penetration range. 350

Construction technology of aeolian sand tunnel 351
An aeolian sand stratum has low cohesion and poor self-stability, collapses easily 352 during excavation, and construction therein is difficult. Until now, there have been 353 relatively few tunnel projects in the aeolian sand areas of China. In addition, many 354 issues need to be addressed urgently, such as a lack of construction experience, 355 imperfect construction technologies and methods, and a lack of well-formed, systematic, 356 and applicable construction modes. To discuss the construction technologies for aeolian 357 sand tunnels, we first investigated the construction approaches to related aeolian sand 358 tunnels. The projects that have been completed recently are listed in Table 6. 359 construction technology. In this study, FLAC 3D was adopted for the analysis. In the 365 establishment and analysis of the numerical model, we made the following 366 assumptions. First, the initial stress field of the formation did not consider tectonic 367 stress; only its geostatic stress was considered. The effects of the advanced support 368 and bolt reinforcement were achieved by increasing the surrounding rock parameters. 369 To fully reflect and compare the effect of the advanced support, the initial support 370 was activated after the excavation calculation was completed. 371 To ensure sufficient solution accuracy, the general model boundary was three to five times the tunnel diameter, and the semi-infinite boundary was simplified to a finite 373 boundary, so as to eliminate the influence of boundary effects. Simultaneously, the 374 normal displacement of each boundary surface was constrained around the model, the 375 bottom surface was completely constrained, and the top surface of the model was a free 376 surface. Based on this, the width direction (  Table 7. 386 horizontal jet grouting pile, and anchor (including the pipe diameter, pipe length, and 392 separation distance) and grouting diffusion radius. By combining the physical and 393 mechanical properties of aeolian sand formation and the purpose of numerical 394 simulation, the soil and the secondary lining adopted the Mohr-Coulomb model and the 395 elastic model, respectively, and solid elements were used for simulation. The initial 396 support adopted the shell element of the structural elements for the simulation. 397 The numerical model is shown in Figs. 9-14. Among them, the difference in the 398 advanced support model is only shown in the size of the reinforcement ring; therefore, 399 only one of them is listed as an example.    the compressive stress of the second lining increases, and the tensile stress decreases. 423 In short, it can be seen that the horizontal jet grouting pile method has the best control 424 effect on surrounding rock deformation and settlement, followed by the large pipe shed 425 grouting method; the small pipe grouting method has the worst effect. 426 In actual projects, the horizontal jet grouting pile has a complicated construction 427 technology and a long construction period, but its sand-fixing effect, deformation, and 428 settlement control effects are better than those of advanced small and large pipe sheds, 429 and can effectively solve the technical problems in the construction of aeolian sand 430 tunnels, while ensuring the stability of the surrounding rock and construction safety.

431
Compared with horizontal jet grouting piles, the construction technology of advanced 432 large pipe sheds is relatively simple and the construction period is relatively short, 433 meeting the practical requirements for many projects. The key to its successful 434 application in aeolian sand tunnels is the grouting effect. The annular distance and 435 length of the small pipe should be determined reasonably, depending on the diffusion 436 range and sliding surface of the sand body. 437

Comparison of construction methods 438
(1) Double-sided pit method 439 It can be seen from the numerical simulation results of the construction process of the 440 double-sided pit method that there are many divisions during the construction process 441 that cause large disturbances. The full-section closure time of the primary support is 442 long, but each section is closed immediately after excavation; therefore, the 443 deformation caused by construction is small, the ability to control the surrounding rock 444 deformation is strong, and the construction safety is high. Notably, the structure is 445 subject to complex forces, and part of the structure bears a relatively large force. The 446 following points should be noted during the construction. 447 The vault settlement caused by the excavation of the upper part of the middle head 448 pit accounts for approximately 65% of the final settlement. Attention should be paid 449 during construction, and temporary support should be added when necessary to control 450 the settlement of the vault in a timely manner. 451 During construction, the primary support bears large forces locally; thus, the 452 deformation monitoring and observation of the primary support should be strengthened. 453 Meanwhile, reinforcement must be performed when necessary to prevent local damage. 454 (2) Cross diaphragm (CRD) method 455 From analyzing the numerical results from the cross diaphragm (CRD) method, it can 456 be seen that the CRD method adopts partial construction, and adds temporary support. 457 Its ability to control the surrounding rock deformation is relatively strong, the 458 deformation is small, and the construction safety is high, but the structural force is also 459 relatively high and complex. Compared with the double-sided pit method, the CRD 460 method has a larger construction area, and the deformation caused by the construction 461 is also relatively large. In addition, the vertical temporary support is subject to greater 462 stress during construction; consequently, monitoring and protection should be 463 strengthened. 464 (3) Bench-cut method 465 By analyzing the numerical results from the bench cut method, it can be seen that the 466 bench cut method has poor control ability regarding the surrounding rock deformation, 467 leading to wide distribution area of the plastic zone. The following points should be 468 noted during such construction. 469 The trend of the vault and surrounding rock moving to the tunnel clearance is evident, 476 and the deformation is large; therefore, monitoring and measurement should be 477 strengthened to prevent invasion. 478 (4) Three-bench method with a temporary invert 479 By analyzing the numerical results of the three-bench method with a temporary invert, 480 it can be seen that this method increases the temporary invert, and its surrounding rock 481 deformation control ability is improved relative to that the bench cut method. However, 482 the deformation is still large, and the plastic zone is wider. The main reason is that the 483 settlement of the vault caused by construction primarily occurs during the construction 484 of the upper step, but the temporary invert is placed at the bottom of the middle step, 485 making it difficult to effectively play a role. The recommended construction precautions 486 are the same as those for the bench cut method. 487 (5) Benching partial excavation method with a temporary invert 488 From the analysis of the numerical simulation results of the construction process of the 489 benching partial excavation method with a temporary invert, it can be seen that the 490 surrounding rock deformation control ability is greatly improved relative to that of the 491 bench cut method and three-bench method with a temporary invert. The main reason is 492 that it places the temporary invert at the bottom of the upper step; thus, the deformation 493 of the surrounding rock and the settlement of the vault can be controlled in time. 494 Simultaneously, the core soil of the lower step provides effective support to the closed 495 support system of the upper step. 496 The settlement of the vault caused by the upper step construction accounts for 497 approximately 89% of the final settlement. Therefore, settlement control of the upper 498 step should be considered during such construction. In addition, the soil in the middle 499 of the lower step has a good supporting effect on the supporting system, and protection 500 should be strengthened during construction; shotcrete can be used for sealing if 501 necessary. 502

(6) Comparison of construction methods 503
From the perspective of the surrounding rock deformation, the double-sided pit method 504 has more section divisions and the best control effect on the surrounding rock 505 deformation. The CRD method has relatively large section divisions, and the settlement 506 control effect is the second-best. The bench cut method has the worst ability to control 507 deformation. The three-bench method with a temporary invert improves the control 508 effect compared with the bench cut method owing to the addition of the temporary 509 invert, but the disturbance of upper step excavation is an important part of its 510 deformation causes, and accounts for the largest proportion of the final deformation; 511 moreover, placing the temporary invert at the bottom of the middle step makes it miss 512 the best opportunity for deformation control. The benching partial excavation method 513 with a temporary invert shows improvement over the three-bench method with a 514 temporary invert, that is, the temporary invert is placed at the bottom of the upper step 515 to control the deformation of the surrounding rock in time; consequently, its 516 deformation control ability is greatly enhanced. (Table 12) 517 invert, and e denotes the benching partial excavation method with a temporary invert. 522 In terms of the initial support force, the double-sided pit method and CRD method 523 effectively control the surrounding rock deformation by reducing the excavation area 524 and increasing the temporary support. Simultaneously, the internal force of the initial 525 support increases, and the structural force becomes complicated. In the bench cut 526 method, the internal force of the structure is relatively small after the surrounding rock 527 pressure is released to a certain extent. 528 In summary, the double-sided pit method and CRD method have the best settlement 529 control effects, the highest construction safety, complex structural forces, and large 530 internal forces; the settlement control effect of the benching partial excavation method 531 with a temporary invert is the second-best, and the construction safety is general; the 532 three-bench method with a temporary invert has a poor settlement control effect and 533 poor construction safety; and the bench cut method has the worst settlement control 534 effect and worst construction safety. 535

Different construction schemes 537
(1) Advanced small pipe grouting (circular spacing 20 cm) + four-step method 538 At the beginning of construction, a comprehensive analysis was conducted according 539 to the tunnel geological conditions, construction period, project investment, and so on. 540 It was believed that the small pipe grouting combined with the four-step core soil 541 method has a low construction cost and fast excavation speed. Therefore, this scheme 542 was preferred for construction purposes. However, through practical application, it was 543 found that this scheme is not suitable for aeolian sand formations. During construction, 544 the sand leakage and sand sliding at the tunnel face and side walls were serious, 545 resulting in a funnel (approximately 2 m in diameter and 1.5 m in depth). There were 546 several transverse cracks on the surface in front of the face. The crack spacing was 547 approximately 0.5-1.0 m and the widths were 2-4 cm. After excavation of the tunnel, 548 the initial support settlement was excessively large, and the limit was severely invaded. 549 The maximum intrusion limit was 90 cm, and the steel arch changed. The construction 550 progress was very slow, and only 20 m was constructed for more than three months. 551 (2) Advanced large pipe shed grouting + three-bench method with a temporary invert 552 In response to the problems in the construction, the advanced small pipe grouting was 553 changed to advanced large pipe shed grouting. In terms of construction methods, i.e., 554 the comprehensive comparison of the double-sided pit method, CRD method, and 555 bench method with a temporary invert, it was believed that although the double-sided 556 pit method and CRD method could effectively control deformation, the construction 557 cost is high, and the process is relatively complicated and slow. Therefore, based on the 558 original bench cut method, a temporary invert was added, and the four steps were 559 changed to a three-step excavation, that is, the three-bench method with a temporary 560 invert. This method not only had the characteristics of a simple procedure and fast 561 progress of the bench method, but also added a temporary invert and enhanced the 562 ability to control deformation. 563 (3) Horizontal jet grouting piles + benching partial excavation method with a temporary 564 invert 565 Through timely adjustment, the large pipe shed was changed to horizontal jet grouting 566 piles. Through practical application, the sand-fixing effects of the horizontal jet 567 grouting piles were evident, and the phenomenon of sand leakage was effectively 568 contained. At the same time, the three steps were adjusted to two steps: improving the 569 initial support force of the upper step, reducing the initial support joints, increasing the 570 vertical braces, and controlling the settlement and deformation of the vault in time. The 571 actual application proved the expected results. 572

Comparison of controlling deformation at different construction schemes 573
Three representative tunnel sections were selected for analysis of the deformation 574 control ability (Table 13). The measured data regarding the peripheral displacement 575 convergence and vault settlement are shown in Fig. 15 and Fig. 16. 576 Through the numerical comparison of the convergence and the vault settlement, it can 583 be found that the vault settlements caused by the "small pipe grouting + four-step 584 method" and the "large pipe shed grouting + three-bench method with a temporary 585 invert" are 3.9 times and 4.2 times that caused by the "horizontal jet grouting pile + 586 benching partial excavation method with a temporary invert", respectively, and the data 587 for convergence are 3 times and 3.6 times, respectively. 588 Therefore, it can be considered that in the construction of aeolian sand tunnels, the 589 "horizontal jet grouting pile + benching partial excavation method with a temporary 590 invert" has the best effect in regards the control of settlement and deformation, and has 591 evident advantages relative to the other two schemes. 592

Suggestions for tunnel construction in aeolian sand strata 593
Through the above research, the following suggestions are provided. 594 (1) Advanced support 595 Horizontal jet grouting piles should be adopted for advance support; when conditions 596 are restricted, such as by capital or the construction period, under the premise of 597 ensuring the grouting effect, advanced large pipe shed grouting can be selected. The 598 advanced small pipe grouting method should be used as a supplement to solve the 599 problems of sand leakage caused by the increase in the spaces between pipe sheds with 600 the progress of the pipe shed system; when advanced small pipe grouting is used as the 601 primary advanced support method, it should be arranged densely, or in the double-layer 602 form. 603 (2) Construction method 604 In the case of strict control requirements for surface subsidence, the double-sided pit 605 method or CRD method should be adopted, and when there is no strict control 606 requirement for the surface subsidence, it is recommended to adopt the benching partial 607 excavation method with a temporary invert. These feature few excavation steps and 608 flexible construction systems that can control the settlement and deformation of the 609 vault in time; the three-bench temporary invert method and the bench method without 610 a temporary invert should not be used. 611

Conclusions 612
(1) Ordinary cement grout has a large particle size, and is difficult to inject into aeolian 613 sand layers with a pore size of 10 μm; moreover, the cement particles are suspended in 614 the slurry. Even if the particle sizes are smaller than the pores of the sand layer, the 615 filtration of the sand layer causes an extremely small permeability range, and the grout 616 may consolidate on the surroundings of the grouting holes. 617 (2) The superfine cement grout has a small particle size, with an average particle size 618 of 4 μm. Theoretically, it can be injected into an aeolian sand layer with a pore size of 619 10 μm, but actual field tests show that the grouting effect is not significantly different 620 from that of ordinary cement; this is because sedimentation is prone to occur, resulting 621 in poor stability. 622 (3) The modified sodium silicate grout has good permeability, but the gel time is short, 623 blockage in the pipe often occurs in the test, and the proportioning process is more 624 complicated. After grouting, the strength of the reinforced body is low (0.2 MPa), and 625 it can easily weather and break after exposure. 626 (4) Horizontal jet grouting piles should be adopted for advance support; to ensure the 627 grouting effect, advanced large pipe shed grouting should be selected, and the advanced 628 small pipe grouting method should be used as a supplement. 629 (5) In the case of strict control requirements for surface subsidence, the double-sided 630 pit method or CRD method should be adopted, and when there is no strict control 631 requirement for surface subsidence, it is recommended to adopt the benching partial 632 excavation method with a temporary invert. There are few excavation steps and flexible 633 construction systems that can control the settlement and deformation of the vault in 634 time. The three-bench temporary invert method and bench method without a temporary 635 invert should not be used. 636