Title Page
ABSTRACT
국문 초록
PREFACE
Contents
CHAPTER 1. INTRODUCTION 27
1.1. Background 27
1.2. Thesis organization 31
CHAPTER 2. REVIEW OF THE PREDICTION METHOD WITH GEOPHYSICAL EXPLORATION 35
2.1. Geophysical Exploration methods 35
2.1.1. Applied seismology 36
2.1.2. Electrical methods 40
2.1.3. Electromagnetic methods 42
2.1.4. Other methods 45
2.1.5. Selection of optimal methods for predicting geological risks 48
2.2. Theory of electrical methods 52
2.2.1. Electrical resistivity survey 52
2.2.2. Electrode array 56
2.2.3. Time-domain IP survey 60
2.2.4. State of the Art in Electrical methods applied to TBMs 67
CHAPTER 3. LABORATORY SIMULATIONS ON ELECTRICAL METHODS TO PREDICT GEOLOGICAL RISKS AHEAD OF A TBM TUNNEL 76
3.1. Chapter overview 76
3.2. Lab-scale experiments 80
3.2.1. Test apparatus setup 80
3.2.2. Soil and Rock specimens 88
3.2.3. Experimental cases: geological risks during TBM tunneling 92
3.3. Discussion of the experimental result 98
3.3.1. Effect of fault zone (Case 1) 98
3.3.2. Effect of water intrusion zone (Case 2) 101
3.3.3. Effect of geological transition (Case 3) 103
3.3.4. Effect of embedded core stones (Case 4) 107
3.3.5. Effect of cavity existence (Case 5) 109
3.3.6. Application of experimental results for real TBM tunneling 111
3.4. Chapter conclusion 116
CHAPTER 4. NUMERICAL SIMULATIONS ON ELECTRICAL METHODS TO PREDICT MIXED GROUND AHEAD OF A TBM TUNNEL 118
4.1. Chapter overview 118
4.2. Analytical solution of electrical resistivity 121
4.3. FE numerical simulations 124
4.3.1. Verification of developed FE model 124
4.3.2. Effect of tunnel geometry 130
4.4. Parametric study 134
4.4.1. Effect of geological conditions 134
4.4.2. Effect of sensor geometric conditions 138
4.5. Chapter conclusion 145
CHAPTER 5. PREDICTING GEOLOGICAL RISKS AHEAD OF A TUNNEL FACE USING ELECTRICAL RESISTIVITY SURVEY: A NUMERICAL AND MACHINE LEARNING APPROACH 147
5.1. Chapter overview 147
5.2. Machine learning model 150
5.3. FE numerical simulation 154
5.3.1. Verification of FE model 154
5.3.2. Simulation of geological risks ahead of a tunnel face 160
5.4. Model implementation 167
5.4.1. Analysis of the input parameters 167
5.4.2. Implementation procedure 171
5.4.3. Results and Discussions 175
5.4.4. Application of prediction results in practice 183
5.5. Conclusion 185
CHAPTER 6. OPTIMIZATION OF ELECTRICAL METHODS UTILIZING MODIFIED HARMONY SEARCH ALGORITHM 187
6.1. Chapter overview 187
6.2. Research background 190
6.2.1. Analytical solution of resistivity 190
6.2.2. Modified harmony search algorithm 197
6.3. Optimization 205
6.3.1. Optimization system 205
6.3.2. Inverse analysis process using MHS algorithm 207
6.3.3. Optimum MHS algorithm parameters 211
6.4. Lab-scale experiment 214
6.4.1. Experimental setup 214
6.4.2. Experimental results and discussions 219
6.5. Field application 228
6.5.1. Field overview 228
6.5.2. Field application and results 230
6.6. Chapter conclusion 237
CHAPTER 7. HARMONY SEARCH-BASED OPTIMIZATION FOR ELECTRICAL EXPLORATION IN DETECTING GEOLOGICAL TRANSITIONS AT A TUNNELING SITE 239
7.1. Chapter overview 239
7.2. Research background 242
7.2.1. Analytical solution of resistivity 242
7.2.2. Harmony search algorithm 247
7.3. Optimization process 252
7.3.1. Inverse analysis process 252
7.3.2. Optimum HS algorithm parameters 256
7.4. Laboratory experiment 261
7.4.1. Test setup 261
7.4.2. Results and discussion 264
7.5. Application for field tunneling 272
7.6. Chapter conclusion 275
CHAPTER 8. SUMMARY AND CONCLUSIONS 277
REFERENCES 285
Table 2.1. Geological risks prediction methods applicable to tunneling sites 51
Table 2.2. Relationship between geological conditions and electrical resistivity 53
Table 2.3. Characteristic of each electrode array 57
Table 3.1. Rock and soil specimen properties 91
Table 3.2. Experimental cases for predicting geological risks 97
Table 3.3. Guideline on predicting hazardous ground conditions utilizing electrical method during TBM tunnel excavation 115
Table 4.1. Material properties adopted in FE numerical analysis and experiment 126
Table 5.1. Electrical and mechanical properties of the materials 158
Table 5.2. Numerical analysis cases for predicting geological risks 166
Table 5.3. Statistical descriptions of electrical resistivity in the numerical database 169
Table 5.4. Geological risks prediction performance and optimal hyperparameters of algorithms 176
Table 5.5. Application results of the developed model 180
Table 6.1. Experimental variables for validation 194
Table 6.2. Initial values of the experiments for sensitivity analysis 194
Table 6.3. MHS algorithm parameters for parametric studies 211
Table 6.4. Experimental cases 218
Table 6.5. Optimization results from the inverse analysis considering the distance (daz)[이미지참조] 221
Table 6.6. Optimization results from the inverse analysis considering the thickness (taz)[이미지참조] 224
Table 6.7. Optimization results from the inverse analysis considering the gouges (ρaz)[이미지참조] 227
Table 6.8. Applied known variables of the field tests 231
Table 6.9. Optimization results for real tunneling sites 233
Table 7.1. Initial values of the experiments for sensitivity analysis 244
Table 7.2. HS algorithm parameters for parametric studies 257
Table 7.3. Experimental cases regarding to two types of geological transition 263
Table 7.4. Optimization results from the inverse analysis considering the distance (dtrans)[이미지참조] 267
Table 7.5. Optimization results from the inverse analysis considering the soil substances 270
Table 7.6. Optimization results for verifying field applicability 273
Figure 1.1. Damage in tunneling sites caused by tunnel excavation: (a) Tunnel face collapse; (b) Water intrusion at the tunnel face;... 30
Figure 1.2. Organization of thesis 34
Figure 2.1. Schematic view of seismic surveys: (a) reflective methods; (b) refractive methods 38
Figure 2.2 Schematic view of seismic methods applicable to tunneling: (a) TSP (Tunnel Seismic Prediction) methods (Dickmann and Sander, 1996) (b) SSP-E (Sonic Soft ground... 39
Figure 2.3. Schematic view of electromagnetic methods applicable to tunneling: (a) Radar exploration methods (Ba-Ro-Tec Inc.); (b) Electromagnetic wave exploration methods... 44
Figure 2.4. Statistical methods based on the machine database 46
Figure 2.5. ML algorithms based on the machine and geological database 47
Figure 2.6. Results of expert surveys: (a) Geological risks requiring prediction during tunneling; (b) Required depth of investigation for predicting geological risks during tunneling 49
Figure 2.7. Typical range of electrical resistivity of saturated rocks and soils 54
Figure 2.8. Schematics of electrode arrays: (a) Dipole-Dipole array; (b) Pole-Pole array; (c) Schlumberger array; (d) Wenner array 58
Figure 2.9. Concept of electrical methods on a tunnel face utilizing the Wenner array 59
Figure 2.10. Schematics of principle of IP surveys: (a) Accumulation of cations inside the ground (overvoltage effect); (b) Voltage decay curve in time-domain IP method 61
Figure 2.11. Conceptual pore model: (a) 'silt' sequence representing interconnected pores : (b) Concentric circles 65
Figure 2.12. Schematics of the multi-layer ground formation with electrical flow lines : (a) Upper ground layer with higher chargeability; (b) Upper ground layer with... 66
Figure 2.13. Bore-tunneling Electrical Ahead Monitoring system (BEAM) : (a) Schematics of the apparatus; (b) Rock type and water inflow indication... 69
Figure 2.14. Tunnel Electrical Resistivity Prospecting System (TEPS) : (a) Schematics of the apparatus; (b): Example of test results 71
Figure 2.15. TBM Resistivity Prediction system (TRP): (a) Schematics of the apparatus; (b) Photograph of the system 73
Figure 2.16. Numerical FE model simulating the electrical methods: (a) TBM environment geometry and material definition; (b) Finite element domain... 75
Figure 3.1. Photograph and schematics of the test apparatus: (a) Photograph of test apparatus; (b) Schematics of test apparatus 81
Figure 3.2. Schematic diagram of the TBM model and electrode: (a) TBM model; (b) Electrode 82
Figure 3.3. Validation of the experimental setup by the analytical solution: (a) Front view of the Wenner array to the two layers earth model; (b) Schematics of the test for verifying the experimental setup; (c) Comparison results 85
Figure 3.4. Experimental setup for investigating boundary effect: (a) Plan view; (b) Front view 86
Figure 3.5. Examination of boundary effect in soil chamber test: (a) Movement of electrodes along the x axis; (b) Movement of electrodes along the y axis; (c) Movement of electrodes along the z axis (vertical direction) 87
Figure 3.6. Schematic diagram of rock mass and rock blocks in chamber test: (a) Rock mass; (b) Rock blocks 89
Figure 3.7. Effect of rock blocks in measuring electrical resistivity and chargeability: (a) Resistivity; (b) Chargeability 90
Figure 3.8. Geological risks ahead of a TBM tunnel 93
Figure 3.9. Illustration of laboratory test cases for predicting geological risks: (a) Fault; (b) Water intrusion; (c) Rock to soil transition; (d) Soil to rock transition; (e) Embedded core-... 96
Figure 3.10. Measured resistivity and chargeability data as the TBM model approaches the fault zone; (a) Electrical resistivity (granite); (b) Chargeability (granite); (c) Electrical... 100
Figure 3.11. Measured resistivity and chargeability as the TBM model approaches the water intrusion zone; (a) Electrical resistivity (tap water intrusion); (b) Chargeability (tap water... 102
Figure 3.12. Measured resistivity and chargeability as the TBM model approaches the rock-to-soil transition zone; (a) Electrical resistivity (granite); (b) Chargeability (granite); (c)... 105
Figure 3.13. Measured resistivity and chargeability as the TBM model approaches the soil-to-rock transition zone; (a) Electrical resistivity (granite); (b) Chargeability (granite); (c)... 106
Figure 3.14. Measured resistivity and chargeability as the TBM model approaches embedded core stones in soil formations; (a) Electrical resistivity; (b) Chargeability 108
Figure 3.15. Measured resistivity and chargeability as the TBM model approaches cavity in soil formations; (a) Electrical resistivity; (b) Chargeability 110
Figure 3.16. Example of installation of electrodes ahead of a tunnel face; (a) Electrode advancing apparatus in EPB-TBM; (b) Electrode advancing apparatus; (c)... 112
Figure 4.1. Front view of the Wenner array to a vertical fault 123
Figure 4.2. Modeling for simulating mixed ground condition ahead of a tunnel face; (a) Simplified model; (b) Tunnel excavation model 127
Figure 4.3. FE model for simulating laboratory experiment of the resistivity survey (Simplified model); (a) Perspective view; (b) Plan view 128
Figure 4.4. Comparison of numerical analysis results with experimental results and analytical solution (Simplified model) 129
Figure 4.5. Numerical modeling of simulating mixed ground condition ahead of a tunnel face (Tunnel excavation model); (a) Tunnel excavation model outline; (b) Mesh configurations of tunnel excavation model 132
Figure 4.6. Estimation on effect of real tunnel geometry; (a) Comparison of tunnel excavation model to simplified model; (b) Effect of tunnel depth (H/D) 133
Figure 4.7. Effect of interface slopes on electrical resistivity survey; (a) Geometry of numerical model; (b) Comparison of effect of interface slopes 136
Figure 4.8. Effect of difference in electrical resistivity between two ground formations; (a) Comparison of effect of R (R〉1); (b) Comparison of effect of R (R〈1) 137
Figure 4.9. Geometry of numerical model for parametric study (electrode spacing (a)) 140
Figure 4.10. Geometry of numerical model for parametric study (location of electrode array (x)); (a) Case 1; (b) Case 2 141
Figure 4.11. Effect of electrode spacing on electrical resistivity survey: (a) θ = 10° ; (b) θ = 30° ; (c) θ = 45° ; (d) θ = 75° 142
Figure 4.12. Effect of location of electrode array on electrical resistivity survey (Case 1): (a) θ = 10° ; (b) θ = 30° ; (c) θ = 45° ; (d) θ = 75° 143
Figure 4.13. Effect of location of electrode array on electrical resistivity survey (Case 2): (a) θ = 10° ; (b) θ = 30° ; (c) θ = 45° ; (d) θ = 75° 144
Figure 5.1. Graphical description of machine learning algorithms; (a) KNN algorithm; (b) SVM algorithm 151
Figure 5.2. Workflow of RF and XGB algorithms; (a) RF algorithm; (b) XGB algorithm 153
Figure 5.3. Schematic of experimental setup 156
Figure 5.4. FE model for simulating lab-scale experiment of the electrical resistivity survey; (a) Perspective view; (b) Plan view 157
Figure 5.5. Comparison of numerical results with experimental results 159
Figure 5.6. Hazardous ground conditions during tunnel construction 161
Figure 5.7. Numerical modeling of simulating geological risks in front of a tunnel face; (a) FE numerical model outline; (b) Plan view 162
Figure 5.8. Illustration of numerical analysis cases for predicting geological risks: (a) Fault, water intrusion; (b) Geological transition; (c) Mixed ground; (d) Cavity 165
Figure 5.9. Histogram of the database applied to ML algorithm 168
Figure 5.10. Five-fold cross-validation procedure 173
Figure 5.11. Flowchart of development procedure for optimal model 174
Figure 5.12. Confusion matrix of ML algorithms indicating true and predicted labels: (a) KNN; (b) SVM; (c) RF; (d) XGB 177
Figure 5.13. Evaluation of the accuracy as the prediction performance 178
Figure 5.14. Feature importance of RF algorithm 179
Figure 5.15. Application cases for classifying geological risks: (a) Cases 1, 2 (Mixed ground); (b) Cases 3, 4 (Fault) 181
Figure 5.16. Application results 182
Figure 5.17. Flowchart for application of geological risk prediction model in practice 184
Figure 6.1. Schematics representing the analytical solutions: (a) Multi-layer earth model; (b) An anomaly ahead of a tunnel 192
Figure 6.2. Validation of the analytical solution: (a) Schematics of experiment; (b) Experimental results 195
Figure 6.3. Sensitivity analysis for variables in the analytical solution 196
Figure 6.4. Flowchart of MHS algorithm 199
Figure 6.5. Mutation process for improvising new harmony 202
Figure 6.6. Schematics of optimization system 206
Figure 6.7. Objectives of the inverse analysis 209
Figure 6.8. Pseudo code of the MHS algorithm 210
Figure 6.9. Results of the parametric studies for selected parameters: (a) Prediction errors (HMCR); (b) Optimal convergence (HMCR); (c) Prediction errors (PAR); (d) Optimal... 213
Figure 6.10. Schematic of test setup 215
Figure 6.11. Schematic of the lab-scale experiment 217
Figure 6.12. Effect of the distance (daz) on the optimization: (a) Prediction error graph; (b) Convergence curve[이미지참조] 220
Figure 6.13. Effect of the thickness (taz) on the optimization: (a) Prediction error graph; (b) Convergence curve[이미지참조] 223
Figure 6.14. Effect of the fault gouges (ρaz) on the optimization: (a) Prediction error graph; (b) Convergence curve[이미지참조] 226
Figure 6.15. Geological profile along the tunnel alignment of the tunneling sites: (a) First site (NATM tunnel); (b) Second site (NATM tunnel); (c) Third site (TBM tunnel) 229
Figure 6.16. Convergence curves for optimization results 234
Figure 7.1. Schematics representing the analytical solutions: (a) Multi-layer earth model; (b) geological transitions ahead of a tunnel 243
Figure 7.2. Sensitivity analysis for variables in the analytical solution 245
Figure 7.3. Flowchart of HS algorithm 251
Figure 7.4. Objectives of the inverse analysis 254
Figure 7.5. Inverse analysis process of the developed system 255
Figure 7.6. Schematics of experiments for determining the optimal parameters 258
Figure 7.7. Results of the parametric studies for selected parameters: (a) Prediction errors (HMS); (b) Optimal convergence (HMS); (c) Prediction errors (HMCR); (d) Optimal... 260
Figure 7.8. Illustration of laboratory test cases for predicting geological transitions; (a) rock to soil transition; (b) soil to rock transition 262
Figure 7.9. Effect of the distance (dtrαns) on the optimization: (a) prediction error graph (rock to soil); (b) convergence curve (rock to soil); (c) prediction error graph (soil to rock);...[이미지참조] 266
Figure 7.10. Effect of the soil materials on the optimization: (a) prediction error graph (rock to soil); (b) convergence curve (rock to soil); (c) prediction error graph (soil to rock); (d)... 269
Figure 7.11. Convergence curve for verifying field applicability 274