IMPACTS OF DIFFERENT FACTORS ON SEEPAGE AND LAND UPLIFT DUE TO COMPRESSED-AIR INJECTION Abstract In this study, using an in-situ, air-flow test in Essen, the impacts of different factors on multiphase flow and land uplift during and after compressed-air injection were investigated using numerical simulations. A loosely coupled, two-phase flow and geo-mechanical modeling approach, linking two numerical codes (TOUGH2/EOS3 and FLAC3D) was employed to simulate the in-situ, air-flow test for comparing the simulated and measured results. As the compressed air is injected, it flows upwards and laterally, and the vertical effective stress near and above the injection zones decreases owing to the pore pressure increasing here, causing an expansion of the soil skeleton in the corresponding zones. The land uplift, induced mainly by support actions from lower deformed soils, is relevant to the distribution of the porosity increments in the soil interior, and it increases rapidly during air injection. After the compressed-air injection stops, the land uplift decreases gradually to zero due to the overpressure dissipation. With a combination of intensive rainfall, the land uplift is slightly greater near the borehole, but it is significantly greater at a distance from the borehole than the land uplift with no or low rainfall, but the air-injection rate remains almost unchanged due to the unchangeable pore pressure near the injection region. As the intrinsic permeability increases or the air entry pressure decreases in the injected strata, both the land uplift and the air-injection rate increase, but the time required for the land uplift to become zero is slightly advanced with either a small permeability or a high air entry pressure. Zang Yongge Tianjin University, State Key Laboratory of Hydraulic Engineering Simulation and Safety Tianjin 300072, China E-mail: zangyongge2011 @163.com Sun Dongmei (corresponding author) Tianjin University, State Key Laboratory of Hydraulic Engineering Simulation and Safety Tianjin 300072, China E-mail: sundongmei@tju.edu.cn Feng Ping Tianjin University, State Key Laboratory of Hydraulic Engineering Simulation and Safety Tianjin 300072, China E-mail: fengping@tju.edu.cn Stephan Semprich Graz University of Technology, Institute of Soil Mechanics and Foundation Engineering 8010 Graz, Austria E-mail: stephan.semprich@tugraz.at Keywords loosely coupled two-phase flow and geo-mechanical model; in-situ, air-flow test; compressed-air injection; multiphase flow; land uplift; air loss 1 INTRODUCTION During tunnel construction below the groundwater table, due to the relatively higher groundwater pressure and the greater hydraulic conductivity, the pore-water in soil voids flows into the work space through the excavated surface, which may hinder the progress of the construction or even cause the collapse of the tunnel. It would, therefore, be advantageous to apply the compressed-air technique during the tunnel's construction [1-2]. By introducing pressurized air into the tunnel space, the groundwater inflow through the excavated surface could be prevented and the surface settlement can also be reduced, which is very critical in an urban area where the damage of settlements on the existing buildings must be concerned. It is important to note that the applied compressed-air pressure in the tunnel space must be equal to or greater than the magnitude of the groundwater pressure at the tunnel invert, so that the compressed air can permeate into the surrounding soils due to the pressure gradient. In Acta Geotechnica Slovenica, 2017/2 19. Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection addition to this application, subsurface fluid injection has been applied extensively for energy development and environmental management, such as enhancing oil production, storing useful gas or oil in depleted gas/ oil fields, recharging an over-drafted aquifer system, arresting or mitigating land subsidence, and disposing of contaminants and hazardous wastes, and other applications [3-6]. Actually, these applications of pressurized fluid injection all involve an interaction between multiphase fluid flow and geo-mechanical processes, and they could affect the variation of the seepage and the stress state in the porous media [5, 7-9]. The simultaneous consideration of the gas phase, liquid phase, and solid phase underground could therefore produce a more realistic result. Numerical analyses, compared to analytical solutions, can dispose of the complicated initial and boundary conditions, the multi-layered soils and the complex geometry of many engineering problems, and is a better choice to analyze the coupling processes between multiphase fluid flows and soil deformation. Selvadurai and Kim developed a mathematical solution to study the caprock-storage formation interactions during the injection of fluids into a poroelastic storage formation and the ground subsidence caused by the uniform extraction of fluids from a disc-shaped region [10-11]. A loosely coupled methodology, linking two numerical codes (TOUGH2, used for solving multi-phase multi-component flow equations [12]; and FLAC3D, used for solving geo-mechanical, stress-strain equations [13]), was proposed by Rutqvist et al. [14] and Rutqvist and Tsang [15] to simulate the interactive processes between the geo-mechanical and fluid-flow processes. This coupled simulator has been widely applied in many geo-environmental situations, such as nuclear waste disposal, CO2 sequestration, geothermal energy extraction, naturally occurring CO2 upwelling with surface deformations, and gas production from hydrate-bearing sediments [16-20]. Therefore, in this study, on the platform of the in-situ, air-flow test in Essen conducted by Kramer and Semprich [1], the water-air, two-phase flow processes and soil deformation during and after compressed-air injection were investigated using the coupled TOUGH2-FLAC3D simulator. Notably, the in-situ, air-flow test was carried out to explore the behavior of the outcropping soils before the subway construction using the compressed-air technique in Essen, Germany. Then, the impacts of the different factors on the multiphase flow and land uplift during and after compressed-air injection were analyzed, including the occurrence of a rain event, and a sensibility analysis of the values of the permeability and air entry pressure of the injected strata. 2 COUPLING PROCEDURES TOUGH2/EOS3 is a module in TOUGH2 for non-isothermal, water-air, two-phase flow in three-dimensional, unsaturated-saturated porous and fractured media, in which the transformation and dissolution processes occurring between the liquid and gas phases are explained by mass-balance equations. These balance equations are discretized in space by the integral finite difference and in time by the first-order finite difference. FLAC3D is a three-dimensional and explicit finite-difference computer code for solving geo-mechanical stress-strain equations. In the coupled two-phase flow and geo-mechanical process, the seepage process affects the stress field by changing the pore pressure and the effective stress, whereas the stress field affects the seepage through changing the porosity, the capillary pressure and the intrinsic permeability. 2.1 Update of Geo-mechanical Variables In the loosely coupled procedure, the pore water pressure pi, the pore air pressure p„, and the liquid saturation 5/ provided by TOUGH2 (the pore pressure is referred to the local atmospheric pressure, and the same below) are sent to FLAC3D to calculate the average pore pressure p[ 14]: p = S,p, + (l-S,)pg (l) This expression for the average pore pressure is applied to porous sedimentary rock [14, 21], and the medium was assumed to be porous media in this study. Then, the average pore pressure is incorporated into the calculations of the soil skeleton stress, the strain, and the effective stress = d(Vp !Vt) = {VtdVp - VpdVt) /Vt2 = =dvp/v,-t0dvl/vt (2) where Vp is the pore volume, V, is the total volume, and 0 is the initial porosity at the initial stress. The total volume Vt is equal to the sum of the pore volume Vp and the solid grain volume Vs. Here, it is assumed that the deformation of the solid grains is much less than that of the soil skeleton, and can be negligible. Therefore, the change in the total volume V, is equal to the change in the pore volume Vp, i.e., dVt= dVp, and the change in porosity d(j> can be given by d = (}-0)dV, / Vt = (1 - K (3) where cv. is the volumetric strain. 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection 2.2 Update of Hydraulic Variables The geo-mechanically induced d has an immediate effect on the fluid flow behavior, including the intrinsic permeability and the capillary pressure. The Kozeny-Carman Equation suggests that there should be a linear relation between the hydraulic conductivity k (or the intrinsic permeability K,k = Kyw /¡xw , where yw is the unit weight of water and ¡uw is the dynamic viscosity of water) and e3/( 1 +e) (or 1 -(j>)2, where e is the void ratio and is the porosity) of porous materials. Furthermore, according to Chapuis and Aubertin [22], the Kozeny-Carman Equation predicts the hydraulic conductivity of most soils (for 10"1 to 10"11 m/s) fairly well and can be applied to the soils encountered in this study. Hence, for a given soil, the ratio of the intrinsic permeability Kn(n) at the time step n to the initial intrinsic permeability Kq(i/>q) can be expressed as follows: 1-A)2 K0 $/(Wo)2 (4) where (¡>n is the porosity at the time step n. Therefore, the intrinsic permeability Kn at the time step n can be defined with Eq. (5). This relationship has also been used by Taylor to estimate the unknown hydraulic conductivity at a porosity from a known hydraulic conductivity at another porosity for the same soil [23]. porosity o of each element. Hereafter, the coupling process of TOUGH2-FLAC30 starts. The initial porosity o from the FLAC3D element is mapped to the TOUGH2 element, which is used to update the intrinsic permeability according to Eq.5. TOUGH2 is firstly executed for the first time-step, and the capillary pressure is corrected using Eq.6. When convergence is reached at the end of this time-step, the pore water pressure pi, the pore air pressure pg, and the liquid saturation Sj of each element in TOUGH2 are obtained and are mapped to the FLAC3D nodes using a weighted distance interpolation. Then FLAC3D runs under loads of average pore pressure for each node (which can be calculated using Eq.l), until an equilibrium state is reached. At this point the change in the porosity at the FLAC3D element can be calculated according to the strain increment (by Eq.3). Afterwards, the updated porosity at the FLAC3D element is sent back to the TOUGH2 element by interpolation. Utilizing the updated porosity, the intrinsic permeability is updated again according to Eq.5, and TOUGH2 is executed for the next time-step. And then the above coupling processes are repeated until the specified simulation time (Sum-time) is reached. 3 EXPERIMENT AND NUMERICAL SIMULATION OF THE IN-SITU AIR-FLOW TEST IN ESSEN K„ = K„ 4 1zA w, (5) n J According to Rutqvist and Tsang [15], with respect to porous media, the capillary pressure is scaled with the intrinsic permeability and porosity according to a function by Leverett [24]. PcL„ = P C I h (6) 4KJt where pcLn is the corrected capillary pressure at the time step n and pcn is the calculated capillary pressure dependent on the liquid saturation at the time step n. 2.3 Coupling Procedure The loosely coupling process between TOUGH2 and FLAC3D is typically developed according to the procedure in Fig.l. Some input data files for TOUGH2 and FLAC3D, such as the soil properties, mesh, boundary and initial conditions, must be initially prepared. Then TOUGH2 is executed for a sufficient simulation time to obtain an initial steady state, and FLAC3D is also operated under gravitational loads to establish initial equilibrium stress gradients and outputs the initial 3.1 Description of the ln-situ Air-Flow Test An in-situ, air-flow test was carried out by Kramer and Semprich [1] to investigate the air permeability of the Essen soil, determine the extent of the airflow field, and analyze the effect of compressed airflow on the deformation of the soil skeleton and in particular on the surface displacement. Fig. 2 shows a schematic diagram of the experimental setup for the in-situ, air-flow test. According to the description of the related experiment [1], the soil profile in Essen can be divided into the following four distinct layers: a fill layer, a thick silt layer, a thin permeable sand layer and a thick layer of marl that is rather weathered in the upper region and presents pronounced joints. The groundwater table is approximately 4.75 m below the ground surface and is located in the silt layer. Compressed air can be introduced into the ground through a 1.5-m diameter borehole in the injection well. A thin steel pipe can be installed inside the borehole, the lower part of the pipe being perforated for air permeating into soils and the top of the pipe being connected to an air compressor (which was used to control the air pressure) [1]. In test IB under consideration here, the 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection TOUGH2 FLAC3D Input data files (properties, mesh, Input data files (properties, mesh, boundary conditions, and initial boundary conditions, and initial conditions) conditions) * t > Run TOUGH2 to steady conditions establish initial Run FLAC3d to establish initial equilibrium stress gradients; Output (j>0 y r r Set up TOUGH2-FLAC3d coupling simulation Map (j)n | from FLAC3D elements to TOUGH2 elements, update Kn , (n = 1,2, 3,......) RunTOUGH2: time step n : tn_x-tn (n = 1, 2, 3,...) Input: (j)n_x, Kn_x; Update pcLn ; Output: pgn, p,„, Sln Map p , pln, from TOUGH2 elements to FLAC3D nodes Run FLAC3D: time step «: tn ] - tn (n= 1,2,3,......) Input: pgn, pln, Sln; Output: n VI VI J Figure 1. Flow chart for a loosely coupled algorithm. compressed air was introduced 18.0-21.0 m below the ground surface, and only the first applied pressure level Ap= 160 kPa over 27 h was considered for simplicity. The rate of air injection was monitored by a flow meter during the experiment. Several piezometers were installed near the borehole to measure the variation of the pore pressure at different depths and distances from the borehole. The land uplift was measured by geodetic leveling installed at different distances from the borehole. 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection >■>■3 m Marl 7.im Silt Sand Gravel I.arae diameter borehole 1.5 m Flaw meter g Air pressure gauge J.O m Fill Figure 2. In-situ, air-flow test in Essen (after Kramer and Semprich [1]). 3.2 Setting up of the Numerical Model 3.2.1 Model Domain and Soil Parameters In this study, the numerical model of the in-situ, air-flow test was established using the above coupling algorithm. Taking advantage of the axial symmetry of this problem, only one-fourth of the domain was simulated. The measured results show that the land uplift reached only 2 mm at approximately 20 m distance from the borehole. The extent of the model was just 40 m in the horizontal direction in the simulation by Ôttl [25] and 100 m by Chinkulkijniwat et al. [2]. Therefore, the model domain was 100 m long in both the transverse direction (X) and the longitudinal direction (Y), and was 25 m deep in the vertical direction (Z). The mesh generation in TOUGH2 was the same as in FLAC3D, consisting of 18,259 hexahedron elements and 20,480 nodes, and is shown in Fig. 3. The domain in the vertical direction was discretized to take into account the soil layers, the groundwater table and the air-injection region. The mesh size in the horizontal directions was relatively fine near the borehole and expanded with the distance from the borehole. 7.0 m 4.0 m Borehole f L_ Air injection zone Fill Sût Sand Marl i it sz jroundwal errai ft 100 Y (m) 0 X (m) Figure 3. Meshes in the model of TOUGH2 and FLAC30. 100 Acta Geotechnica Slovenica, 2017/2 7. Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection In the numerical simulation, the capillary pressure and the relative liquid and gas permeability dependent on the liquid saturation were described by the van Genuchten model [26] and the van Genuchten-Mualem model [26-27], respectively. The expressions of these two models are described in Eq. (7), Eq. (8) and Eq. (9) pc=-Po[(srm-ifA (~pmaxSls) (^>0) (8) (9) where S = (5, - Slr) / (1 - Slr - Sgr) and Sgr is the residual gas saturation. The hydraulic and geo-mechanical parameters for four soil layers in Essen, taken from Ottl [25], are shown in Table 1, including the solid density ps, Young's modulus E, Poisson's ratio v, the initial porosity 0q> the initial intrinsic permeability K0,p0, Sir, S/s and A. Notice that the lower part of the silt layer is characterized by a weaker stiffness. Additionally, due to the rather small deformation occurring in the soil (a maximum value of the measured heave is approximately 4 mm) in the field test, linear elastic behavior was assumed for the whole domain. This assumption of linear elastic behavior has also been used in other simulations of the airflow test in Essen [2, 25]. All the processes involved in the numerical simulation were assumed to occur isother-mally at 10°C. 3.2.2 Boundary and Initial Conditions In TOUGH2, the primary variables are P p (the pressure of the phase (the liquid phase (/) or gas phase ( (g/cm3) E (Mpa) V (/>0 (kN/m2) Sis Sir A K0 (10~12 m2) Fill 2.72 20 0.33 0.36 4.0 0.2 1.0 0.8 4.95 Silt 2.90 12.47 0.35 30.0 0.2 1.0 0.5 0.495 (9.24) (0.37) Sand 2.72 21.22 0.32 0.36 4.0 0.05 1.0 0.65 9.90 Marl 2.79 14.33 0.40 0.33 12.0 0.15 1.0 0.6 2.48 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection the simulations. Then, the coupled hydraulic-mechanical process, considering the interactions between the liquid, gas and solid phases, was repeated until the injection duration of 27 h was reached. Hereafter, the material property and the volume of the elements within the injection zone were reset as the property of the marl and their actual volume, respectively, in order to investigate the variations of multiphase flow and soil deformation after the air injection stopped. The post-injection phase lasted for 100 hours. Notably, the experimental data were not recorded after the air injection stopped in the in-situ, air-flow test. 3.3 Analysis of the Experimental and Simulated Results 3.3.1 Analysis of the Discrepancy between the Simulated and Measured Results The main objective of this field test is to investigate the magnitude and scope of the land uplift as well as the air loss rate induced by compressed air injection. Fig. 4a shows the distribution of the simulated and measured land uplift at different times. After 27 hours, the simulated land uplift was much higher than the measured value at a distance from the borehole, which might be because the high permeability of the sand layer promoted radial propagation of the overpressure in the numerical simulation. Actually, according to Kramer and Semprich [1] and Ottl [25], the thin sand layer contains a considerable portion of silt in the field, reducing the permeability of the sand layer. Additionally, average values of the permeability for the marl layer were used in the above simulation, although the marl corresponds to silt containing high portions of sand and clay according to Kramer and Semprich [1], Therefore, the possibility of high anisotropy between the horizontal and vertical soil permeability should be recognized. Fig. 4b compares the behavior over time of the measured and simulated air-injection rates during the test. The mass flux of air (unit: kg/s) injected into the soil interior through the injection zone was calculated using the numerical model and the air density under the normal conditions of atmospheric pressure and 10°C was determined by the ideal gas law, which permitted a calculation of the volume flux of air (m3/min) injected into the soil. The simulated rate was substantially the same as the measured results during the middle period, whereas there were large differences between them at the initial and final stages. In the field test, the applied air pressure was increased stepwise to 160 kPa by the compressor, as shown in Fig. 5. Before the air pressure reached 160 kPa, the matric suction in the marl layer could not surpass the air entry value of the marl, and no air penetrated into the soil. However, the applied air pressure was set Time (h) Figure 4. (a) Measured and simulated land uplift at different times and (b) temporal evolution of the simulated and measured air-injection rate during compressed air injection. to 160 kPa at the beginning of the numerical simulation, so the difference between the measured and simulated results was introduced at the initial stage. Considering the low permeability of the sand layer, the anisotropy between the horizontal and vertical soil permeability of the marl layer, and the implementation process of the applied air pressure, the values of permeability in the sand and marl layer in the numerical simulations were adjusted as follows: Kx = Ky= K: = 0.9xl0"12 m2 for the sand layer and Kx = Ky = 5.58xl0"12 m2, Kz = 3.38x 10"12 m2 for the marl layer, and the applied pressure a. - 200 0 100 0- 0.0 Aj»i>1:«i air pre sutc c \ -Jl* lliird level \ 1 lie tecanri level Air lor«< / jhef —i rsl level J ioo i 3.9.87 4.9.87 5.987 l ime (date) 6.9.87 5.0 0.0 Figure 5. Applied air pressure and air loss versus time for test IB (after Ottl [25]). (b )H -Is 'J 2-0 § B £ -I u 0.5 00 ■ Mtaarad nc suls a 27 h ■-Sinn laded rc =u»s a 27 li - - ~ ftalcitnureHl results Sinmbial ma I Its 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection X (m) (b)«-I " Z i_s G J ii> S * «5 k> •3 0.0 - - - - Measured rcsuks -Simulated rcsuks Figure 6. Measured and adjusted simulated land uplift at different times and (b) temporal evolution of the measured and adjusted simulated air-injection rate during compressed air injection. Ap in the numerical simulations was set as shown in Fig. 5, as follows: Ap = 61 kPa from 0 s to 3360 s, Ap = 78 kPa from 3360 s to 6720 s, Ap = 119 kPa from 6720 s to 10,080 s and Ap = 160 kPa from 10,080 s to 27 hours. The comparison of the adjusted simulated results and the measured results is shown in Fig. 6. There was good agreement between the adjusted simulated and measured land uplift after 27 hours (Fig. 6a). As shown in Fig. 6b, before the air pressure reached 160 kPa (the first 2.8 hours), both the simulated and measured injection rate remained zero. Hereafter, there was a slight discrepancy between the simulated and measured results from 2.8 h to 13 h, whereas they were substantially consistent from 13 h to 21 h. After 21 h, because the air flowed to the location where fractures can be present, the measured rate increased significantly. However, the medium in the numerical model was assumed to be porous media, thereby causing the difference between the simulated and measured rates. In general, the adjusted simulated results agreed well with the measured values, and the adjusted scheme was applied in the following simulation analysis. unsaturated. Moreover, because the applied air pressure was slightly greater than the maximum groundwater pressure within the air-injection regions, a small area below the injection regions also became unsaturated. The unsaturated zone was concentrated mainly above the air-injection regions. Close to the ground surface, the gas saturation near the borehole decreased because the airflow pushing caused the water to flow upwards. Fig. 7b shows the distribution of gas saturation and air flow after 75 hours (48 hours after compressed air injection stopped). As shown, after the compressed-air injection stopped, the air still flowed upwards and laterally due to the density difference and the pressure gradient, but the magnitude of the air velocity was much smaller at 75 hours than it was at 27 hours. The zone from the bottom to 10 m recovered water saturated because the air in the voids of the injection regions was displaced by the adjacent groundwater, whereas due to the continuous upward airflow, the size of the unsaturated zone from 10 m to 15 m remained substantially unchanged, and from 15 m to the water table, the unsaturated zone expanded into the shallow unsaturated zone. 20 —--'>- ' LT-- "'..^"'O-—- */ Sill Ssutd Lb Wf LI l'y MuJ (b)» 10 20 30 40 50 60 70 80 90 lOO X (m) 20 N to - llll -JLl^_- v ^'^JLMH — : r Marl 10 20 30 40 50 X (m) 70 »0 90 103 Figure 7. Distribution of simulated gas saturation and airflow at (a) 27 h and (b) 75 h (The blue arrows refer to the magnitude and direction of airflow velocity). 3.3.2 Analysis of Seepage and Soil Deformation Fig. 7a shows the distribution of the simulated gas saturation and airflow after 27 hours. It was expected that the injecting air would permeate into soils due to the pressure gradient and the adjacent regions became Fig. 8 shows the distribution of the simulated average pore-pressure equivalence at different times. After 27 hours, the average pore pressure near and above the injection zone increased significantly, and the • Mefefurtd results at 27 h -Simulated remits ai 27 h ---Simulated results ai 75 h 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection amplification decreased with the distance from the borehole. The affected zone extended to approximately 70 m horizontally. After 75 hours, due to the airflow dissipation, the average pore pressure near and above the injection zone decreased substantially. From the bottom to approximately 7 m, the average pore pressure recovered to its initial steady value. From 7 m to 16 m, the average pore pressure also decreased to its initial values near the borehole, while it still deviated from the initial values at a distance from the borehole. The reason might be that the gas saturation in this region reached a relatively large value within 10 m horizontally (Fig. 7b), so that it was easy for the airflow to move, but the overpressure and airflow 10 m away dissipated with relative difficulty due to its small gas saturation. Above the elevation of 16 m, the average pore pressure did not decrease to its initial value due to airflow pushing (Fig. 7b). The measured and simulated pore air pressure at the four observed points of A, B, C and D (Fig. 3) are shown in Table 2, and the measured and simulated results were substantially consistent. Fig. 9 shows the distribution of the simulated vertical effective stress at different times. After 27 hours, since the total stress acting on the soil layers remained almost unchanged (the mass of the injected compressed air was very small relative to the weight of the whole soil layers) and the stress overtaken by the fluids in the soil voids increased (Fig. 8), the stress overtaken by the solid skeleton decreased; i.e., the effective stress near and above the injection zones decreased. This release of the vertical effective stress caused an expansion of the soil skeleton in the corresponding zone. The magnitude of the porosity near and above the injection zones therefore increased, and the affected zone of the porosity increment reached approximately 32 m horizontally (Fig. 10a). After 75 hours, the vertical effective stress decreased to its initial values in most areas, while it still appeared to deviate from 12 m to 20 m vertically. Therefore, both the magnitude and scope of the porosity increments also decreased at 75 hours (Fig. 10b), and the maximum value appeared from 10 m to 20 m horizontally due to the distribution of the average pore pressure and vertical effective stress (Figs. 8b and 9). (a)» - Iratul st«dv »tali---27 hours 20 O 15 ---_ ; - ■ - -a------- -ito- 0 10 20 JO 40 50 <0 70 80 90 100 X (m) (b) N • Imtnl steady stite 75 hours -1-■- 10 20 30 —i—-»-i-1-1-r——i— 40 $0 60 70 80 90 100 X (m) Figure 8. Distribution of simulated average pore pressure at different times at (a) initiation and at 27 h and (b) initiation and at 75 h (unit: kPa). Table 2. Locations with the measured and simulated pore air pressure for four observation points. Observation points A B C D Horizontal distance from borehole (m) 2 2 2 6 Depth below ground surface (m) 6 10 13 12.5 Measured pore water pressure (kPa) 27 70 95 90 Simulated pore water pressure (kPa) 30.5 76.5 101 95 25 20 N 10 - Imaal steady sut«---27 hour» .........75 hours —ixr- -m- 0 10 20 Î0 40 50 «0 70 90 90 100 X (m) Figure 9. Distribution of simulated vertical effective stress at different times (unit: kPa). SO 20 30 40 X (m) 10 jo j« 40 X (m) Figure 10. Distribution of the simulated porosity increment at (a) 27 h and (b) 75 h. Because the porosity near the ground surface changed little, the land uplift was caused primarily by the expansion of deep deformed soils. As shown in Fig. 6a, after 27 hours, the maximum land uplift occurred near the borehole. The land uplift decreased with the distance from the borehole, consistent with the distribution of 10. Acta Geotechnica Slovenica, 2017/2 Z. Yongge et al.: Impacts of different factors on seepage and land uplift due to compressed-air injection the porosity increment within the soil. After 75 hours, the land uplift decreased due to overpressure dissipation and the flat segment within approximately 15 m from the borehole was relevant to the distribution of the porosity increment at 75 h. Fig. 11 shows the behavior over time of the maximum land uplift. During compressed air injection, the land uplift remained zero before 2.8 hours due to no air penetrating into the soil. Later, it increased rapidly and reached a maximum value at 27 hours. After the injection stop, the heave decreased rapidly at first and then gradually went to zero at approximately 110 hours. 25 20 0 10 20 X 40 50 60 TO ® «0 100 110 120 Time Phase (base case) and 1.5 Phase ■ Notably, Phase was the air entry pressure of the marl layer in the base case (Table 1). The other simulation conditions were the same as those for the base case. <0 «0 X (m) h for bw cast h for bisc am h for 0 JK^ li for 05^ li for IJK^, h for l JK^ -2' h for b*s* w* - - - 75 h for bw aw -2'hforOJK^ - - - TJhiorOJK^ -Î' li for I JK^ ---75 h for 1 SK^ <0 60 X (m) Bust due 0 5K. -15Kw. 0 10 20 JO 40 50 «0 W SO 90 100 110 120 Time (h) Figure 18. (a) Land uplift at different times and (b) changes in the maximum land uplift with time for different air entry pressures. -i— 10 20 "I 30 •10 -r- 50 —I— «0 -I— so 90 100 110 120 Time