Landslide Triggered by Orthogonal Tunnel Excavation and Prevention Measures in Jimei Village, Sichuan Province, China

: The axis of highway tunnels constructed in mountains under complex geological 15 conditions is usually orthogonal to the section of potential landslide. The tunnel construction may 16 lead to landslide, which then may result in the deformation and/or cracking of tunnels. Therefore, 17 it is very important and practical for tunnel projects to study the complex interaction mechanism 18 between orthogonal tunnel and landslide and provide appropriate prevention measures for tunnel. 19 This paper, on the base of geological survey, on - site monitoring and numerical simulation, 20 analyzed the deformation and reason of an ancient landslide revived by tunnel construction and 21 studied the prevention measures for tunnel. The results show that the reason for the revival of the 22 ancient landslide resulted mainly from the tunnel construction through sliding surface, and the 23 ancient landslide is generally stable because most landslide deformation occurred beyond the 24 tunnel and in the upper part of landslide. The numerical simulation was used to optimize the 25 tunnel prevention scheme by the analysis to the stability, stresses and deformation of landslide 26 based on stress - strain control theory. The original anti - slide pile design was cancelled and finally 27 the tunnel is reinforced by upper soil removal and moving upper soil into toe. This tunnel has 28 successfully completed and are under good operation. The used prevention measures were proven 29 to be effective according to the monitoring data about displacements and stress of landslide and 30 tunnel during operation period, and saved about seven million US$. The research results in this 31 paper may offer a beneficial reference to projects with similar geological conditions. 32

ancient landslide resulted mainly from the tunnel construction through sliding surface, and the 23 ancient landslide is generally stable because most landslide deformation occurred beyond the 24 tunnel and in the upper part of landslide. The numerical simulation was used to optimize the 25 tunnel prevention scheme by the analysis to the stability, stresses and deformation of landslide 26 based on stress-strain control theory. The original anti-slide pile design was cancelled and finally 27 the tunnel is reinforced by upper soil removal and moving upper soil into toe. This tunnel has 28 successfully completed and are under good operation. The used prevention measures were proven 29 to be effective according to the monitoring data about displacements and stress of landslide and 30 tunnel during operation period, and saved about seven million US$. The research results in this 31 paper may offer a beneficial reference to projects with similar geological conditions. 32

35
China suffers from abundant severe geological hazards each year, and more than 65% of 36 these hazards are statistically landslides (Technical Guidance Centre for Geological Disasters, 2019). 37 When various tunnels are constructed in the mountains in southwestern China, it is very difficult 38 to avoid potential large landslides due to the insufficient instigation on complex geological 39 conditions, although the distance between tunnels and potential landslides are set to be as far as 40 possible in the initial design stage. Tunnels may interact strongly with landslides during tunnel 41 construction and operation. Slopes may creep or even slide due to tunnel construction, and then 42 tunnels suffer from deformation and cracking due to slope failure, which could pose a great threat 43 to the construction safety and operation of tunnel (Karakus and Fowell, 2005;Tang et al., 2014;44 At present, some scholars have paid more attention to the interaction between tunnels and 46 landslides and the damage to engineering structures, and have obtained some achievements. 47 of about 21°. A nameless stream, which is the branch of Gulin River, flows along the slope toe. 123 The landslide is generally about 700m long and 400-900m wide, steep and wide in the upper part 124 and gentle and narrow in the lower part, with developed gentle platform in the middle and lower 125 part. The slope surface is mostly occupied by farmland, with a preliminary school and many 126 resident houses. The landslide deformation had a direct effect on the tunnel structure and buildings 127 on the slope surface. The ancient landslide was basically stable before tunnel excavation. 128 According to the geological survey and borehole data, the ancient landslide has different 129 characteristics in different zones. The landslide can be divided into two zones (zone I and zone 130 Ⅱ). Zone I, about 980m long and 330m wide, is located in ZK 12+506 -ZK 12+840 section, with 131 main sliding direction of about 354° and average slope of 18.5°, accounting for 53% of the whole 132 landslide area (Fig. 2). Zone Ⅱ, about 550m long and averagely 405m wide, is located in 133 ZK12+840 -ZK13+160 section, with main sliding direction of about 358° and average slope of 134 17.8°, accounting for 47% of the whole landslide area. The borehole data show that the ancient 135 landslide is up to 72.6m thick, averagely more than 50m thick, and the whole landslide is over 11 136 million m 3 in volume, being the giant traction-type deep-seated rocky bedding landslide. This 137 ancient landslide originally slid along weak structural plane, and subsequently suffered from 138 multi-zone multi-period and multi-layer slides due to long-term exterior forces. 139 140 Fig.2 The geological structure of the study area 141

Composition of landslide 142
The sliding body in zone I consists mainly of block stone and gravel, with a little silty clay, 143 and has uneven texture. There are a large number of undisintegrated rocks in zone I due to the 144 incomplete disintegration after slope failure ( Fig. 3 and Fig. 4). However, zone II is greatly 145 different from zone I in material components, which consists basically of gravel with breccia clay, 146 plastic to soft plastic, with multi-layer smooth surfaces, partially with 16.5-50.0m long bock stone. 147 The potential failure surface is basically identical to the original failure surface, and some 148 potential sliding surface is directly on the underlying bedrock. However, there are many differences between the two zones: (1) in longitudinal direction, the 163 sliding surface of zone I is gentle and smooth, with gentle undulation, and the middle and back 164 sections are steep; while that of zone II is greatly undulating, and the middle and back sections are 165 gentle. In transverse direction, zone I varies sharply in the sliding surface boundary, with bedrock 166 scarp found at the eastern boundary; while zone II changes greatly in thickness, largely undulating. 167 (2) All the sliding surfaces were found in the regions with adverse engineering geology, but for 168 zone II, the sliding surface is mainly along the weak oil shale which is the interface between J2 s 2 169 and J 2 s 1 ; for zone I, the sliding surface is generally along the weak structural plane in J2 s 2 . (3) The 170 bottom of sliding body in zone II is 5-25m lower than that in zone I, with more energy release, and 171 is above the adjacent riverbed. However, sliding surface of in zone I is below the riverbed, and 172 thus riverbed provides some resistance to this sliding body. Thus, the bottom of sliding body in 173 zone I has better stability. 174

Formation mechanism of ancient landslide 175
The ancient landslide was originally developed in a bedding bedrock form, which was 176 affected by many factors such as strata attitude, structure, groundwater, river corrosion, and slope 177

angle. 178
The landslide is located at a bedding slope in the southwest of Baiyangping syncline. The 179 slope has an attitude of 32°∠15° -26° with steep upper part and gentle lower part (Fig. 2), and the 180 slope consists of sandstone-mudstone interlayer of Jurassic Shaximiao Formation. There is 181 abundant fractures in the sandstone, which could help groundwater flow rapidly. In addition, there 182 are two sets of joints in the slope, approximately normal to each other, which tends to be subject to 183 tension and lateral shearing. The infiltration of surface water through fractures can lead to large 184 pore water pressure, retains on the impervious mudstone surface, and thus poses a great threat to 185 slope stability by softening the mudstone and reducing its mechanical parameters. Furthermore, 186 the long-term river corrosion against slope toe resulted in the high and steep scar, and finally 187 large-scale bedding landslide happened due to rainfall, forming the initial ancient landslide. 188 Subsequently, many years of weathering, rainfall, corrosion and human activities shaped the 189 current landslide. No deformation happened in the ancient slope since the last 100 years, and the 190 slope was generally stable before the tunnel excavation. 191

Deformation characteristics of landslide and tunnel 192
The slid bedrock has not yet fully disintegrated, and thus it is difficult to distinct the true 193 bedrock from false bedrock. Therefore, the tunnel did not completely avoid the ancient landslide,

Landslide deformation and emergency monitoring 204
Some arc-shaped cracks with 10-40cm width and 5-20cm depth occurred in the region about 205 270m right from tunnel, over 500m long intermittently, and generally transfixed at the landslide 206 tail. Much evidence indicated that partial ancient landslide on the right side of the tunnel in zone I 207 was reactivated by the tunnel excavation, resulting in deformable body with obvious outline of 208 basically transfixed cracks (Fig. 5). 209 To ensure the tunnel safety, a displacement monitoring system was installed within the 210 ancient landslide of zone I in January 2015 before tunnel excavation on the base of the 211 borehole-revealed sliding body shape and measured ground surface deformation as shown in Fig.  212 2. This system includes five ground monitoring sites, namely, JC01, JC02, JC05, JC06 and JC08, 213 and three inclinometer sites, namely, JC02, JC06 and JC08. The inclinometers, with 60m depth in 214 JC02, 75m depth in JC06, and 75m depth in JC08, were applied to observe the displacement 215 variation with depth, with one monitoring point every 1m depth. The measurement of vertical and 216 horizontal displacements was conducted every 24h interval with GPS.

Tunnel deformation and monitoring 237
The landslide can cause the tunnel to be subject to two kinds of deformation (Poisel et  On the same day, oblique cracks were found within the tunnel bottom and secondary lining in the 244 ZK12+829 to +817 section, and remarkable deformation was observed in the primary support 245 structure in ZK12+799 to +766 section, which resulted in twist break and deformation in the 246 tunnel top. The right wall deformed greatly beyond the limit by up to 65-69cm, and some concrete 247 dropped from primary support on the tunnel top. Some longitudinal and connected cracks, up to 248 3mm wide, were observed on the top right and bottom left of the right tunnel in K12+540 to +719 249 section. By the middle of December 2015, great deformation and many cracks were observed in 250 the right tunnel of K12+505 to +719 section, with some minor cracks in K12+719 to +799 section. 251 At the same time, cracks occurred in the left tunnel of ZK12+505 to +849 section, especially 252 abundant in ZK12+800 to +767 section. 253 Some monitoring gauges were installed immediately after the severe tunnel deformation, as 254 shown in Fig. 7, with displacement gauges at K12+560, K12+600 and K12+722, strain gauges for 255 temporary oblique support at K12+560, K12+580, K12+600 and K12+620, steel bar stress gauges 256 at K12+538, and concrete stress gauges for secondary lining at K12+554. Type and resolution of 257 the monitoring gauges were shown in Table 1. The tunnel displacement curves are shown in Fig. 8,  258 stress curves of temporary oblique support in Fig. 9, stress curves of lining steel bars in Fig. 10,  259 and stress curves of secondary lining concrete at K12+554 in Fig. 11. All the gauges began

Model and parameters 286
Engineering geological analysis, geological survey, field monitoring and numerical 287 simulation are often used to study the slope failure mechanism and predict the landslide movement. lining, and bedrock, as shown in Fig. 12. The FLAC 3D program, which can well simulate the 302 interaction between rock/soil and structures, was used to analyze the stress and deformation of 303 landslide. 304

Fig.12 3D numerical model of landslide and tunnel 306
For the initial stress condition, the tectonic stresses were ignored, and only gravity-induced 307 stresses were considered. For the boundary conditions, the model bottom was fixed in the vertical 308 direction, and the sides are fixed horizontally and vertically. The elastic-plastic constitutive 309 relation and Mohr-Coulomb failure criterion were applied for rock/soil, and the entity elements 310 were used for the primary and the secondary linings with linearly elastic model. 311 The model parameters should be determined in an appropriate way. The Poisson's ratio, unit 312 weight and elastic modulus were obtained by laboratory tests using samples from each borehole 313 (Fig.1). The soil samples were adopted every one meter in depth from each borehole. The strength 314 parameters were determined by back analysis utilizing uniform design, RBF neural network model 315 , and measured displacement data about landslide surface. Table 2 lists the 316 comparison between monitoring data and back analysis based results, indicating that all the errors 317 of monitoring points are less than 5% except that the errors of JC01 X displacement, JC06 Z 318 displacement and JC08 Z displacement are greater than 5%, which means that the parameters from 319 back analysis are appropriate. Table 3 lists the geotechnical parameters for numerical calculation. 320

Simulation results and analysis 324
Based on the 3D numerical calculation, the stress and deformation of the tunnel and sliding 325 body after tunnel reinforcement and soil removal was achieved. The Y-direction displacement 326 contour after reinforcement of tunnel is shown in Fig. 13, indicating that the sliding body 327 deformation basically occurs in the region over the tunnel, and changes gradually from landslide 328 tail to front in a decreasing way. The displacement contour (Fig. 14) and shear strain increment 329 contour (Fig. 15) are plotted along some typical sections (Fig. 13) to clearly observe the stress & 330 deformation within the slope. Fig. 14 suggests that the ancient landslide will fail again due to 331 tunnel excavation, and the deformation feature by numerical simulation is similar to on-site 332 investigation results of gradual deformation decreasing from tail to front. Additionally, great displacement occurs on the top and right side wall of let tunnel, with terrible inward invasion of 334 right wall. Fig. 15 indicates the distribution of maximum shear strain within the slope, with 335 maximum value of about 0.62, and the sliding surface was not thoroughly developed. According to on-site survey and numerical simulation results, the ancient landslide in zone II 362 are generally stable, without tunnel deformation and ground cracking, and thus no treatment is 363 needed for landslide in this zone, only with monitoring hereafter. The ancient landslide in zone I is 364 also generally stable, but the left tunnel is completely in the sliding body and the right tunnel is 365 just across the sliding surface, which resulted in the damage to right tunnel and upper slope. The 366 depth of the tunnel is greater than 60m, the adjacent strata affected is averagely 35-40m deep, and 367 thus it is not feasible to use the anti-slide piles as tunnel reinforcement. In addition, the anti-slide 368 piles would need very high cost and larger length of up to 80m. However, the lower landslide is 369 thick and upper landslide is thin, and hence it is suitable to stabilize the landslide by removing soil 370 from upper part into slope toe and conducting some reinforcement at lower part. (1) The stress and strain of surrounding rock and tunnel structure will be measured during 378 tunnel operation period. 379 (2) According to monitoring data, the stability and safety of tunnel and landslide will be 380 analyzed, and the range, speed and tendency of deformation for landslide and tunnel will be also 381 evaluated to safeguard the tunnel operation. 382 (3) The tunnel deformation limit will be judged as early as possible to provide early warning 383 to administration for determining the subsequent safety measures. 384 The monitoring data measured from October 2016 to March 2020 show that the displacement 385 at various depths are generally stable, and the change in stresses of steel bar, concrete and support 386 is little, indicating the treatment applied is successful. Thus, the anti-slide piles were not installed, 387 saving about seven million US$. There are few successful treatment cases of such tunnel-landslide 388 systems under complex conditions, so the analysis and design in this paper can provide some 389 beneficial reference to similar projects. 390

391
An ancient landslide was reactivated in October 2015 after the excavation of Jimei tunnel in 392 Sichuan Province, China, which triggered many severe geological problems such as tunnel 393 deformation, ground surface cracks and supporting structure damage. Such slope failure may have 394 great threat to residents and facilities, and sharply increase the construction cost. Several methods 395 were used to analyze the interaction between Jimei tunnel and the ancient landslide, and the 396 proposed reinforcement was evaluated in this paper. 397 At first, the composition and distribution of the ancient landslide was analyzed based on the 398 UAV aerial photography, geological prospecting and field geological survey. Secondly, the 399 parameters of the sliding zone and sliding body were determined with back analysis according to 400 the monitoring data about the surface deformation. Finally, the landslide failure mechanism due to 401 tunnel excavation was discussed by 3D numerical simulation, and the evaluation on the effect of 402 proposed reinforcement was carried out. 403 The borehole data show that the Jimei landslide, as an ancient landslide, is more than 50m 404 thick on average, with over 11 million m 3 in volume, being the giant deep-seated landslide. The 405 sliding body consists basically of block stone and gravel, and the slope failure generally resulted 406 from the tunnel excavation through the sliding belt. Most displacement of the sliding body mainly 407 occurred above the tunnel because the potential sliding surface under the tunnel is gentle and the 408 deep-seated sliding body prevented further sliding. In addition, the range and value of overall 409 displacement of the sliding body by back analysis match well the field survey and measured data. 410 In view of tunnel construction period, sliding body thickness and deformation characteristics, 411 some soil above the tunnel in zone I was suggested to be removed. Numerical simulation results 412 show that the tunnel and sliding body will be subject to small displacements and keep stable after 413 soil is removed. The monitoring data measured during tunnel operation period testified the success 414 of the treatment, and this paper may present a beneficial reference to similar projects. 415 Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. The geological structure of the study area Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.    territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.    Stress curves for temporary oblique support    Maximum shear strain increment contour after soil removal along section I-I Figure 18 Bird view of construction site after soil removal Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.