Numerical Analysis on the Performance of the Underwater Excavation

Based on the Yongdingmen Station of Beijing Metro, the underwater excavation method for deep foundation pit was introduced. This study constructed a numerical analysis model to analyze the performance of surface settlement and lateral wall deﬂection in the process of underwater excavation. Results showed that this method was better to control the surface settlement and lateral wall deﬂection compared with other dewatering excavations. In detail, most of the surface settlement was caused during the dry excavation stage and dewatering excavation stage while the deﬂection caused by underwater excavation only accounted for about 10% of the total settlement. Besides, the maximum settlement occurred 0.25 ∼ 0.5 H e behind the retaining wall and the value was 0.04% H e . Similar to the result of the surface settlement, most of the lateral wall deﬂection had been completed before the underwater excavation, which only caused about 7% of the total deﬂection. The maximum wall deﬂection and its location were approximately 0.06% H e and 0.5 H e , respectively. Moreover, a series of 3D numerical analyses were studied on the design parameters of the underwater excavation method. This study can be used as a reference for general performance and structural design of foundation pits with underwater excavation.


Introduction
e excavation of the foundation pit was often accompanied by structural deflection [1] and surface settlement [2,3]. Among the causes of large deflection, groundwater was a common problem that directly determines the selection of excavation technology and retaining structure. To avoid the engineering disaster caused by the large deflection, it was necessary to adopt effective groundwater control measures and reliable water-stop structures [4,5]. According to the research, there were three methods to control the groundwater in the excavation process: first, setting multiple dewatering wells to control the groundwater level at a certain height before excavation, which was suitable for the excavation with small depth and low moisture content [6][7][8]; second, employing a water curtain to cut off the underground water directly, which generally requires the structure to have a large embedded depth. [9][10][11]; third, combining the water interception and grouting reinforcement treatment, which requires proven technology and engineering cost [5,12].
Influenced by the existing metro lines, the increase of excavation depth of the Beijing subway station will cause most or even all of the station structure to enter the confined water layer. Blindly using traditional dewatering excavation will bring multiple problems such as water resource loss and high cost of precipitation. On this basis, a new excavation method of underwater excavation has appeared in some engineering. Almaleh et al. [13] introduced the basic expansion project of the Crystal River with underwater excavation and underwater grouting. Archontidou-Argyri et al. [14] used the method of underwater excavation to mine cultural relics, which achieved the effect of controlling ground deflection. Hu [15] proposed the underwater excavation to balance the external water pressure and passive Earth pressure by using the recharge water in the Meizizhou air shaft project. Qu et al. [16] applied the nondrainage excavation method in an excavation of Shanghai to control the ground settlement.
It can be seen from the above excavations that this method can effectively protect groundwater resources and control soil deformation. However, due to the particularity of underwater excavation, the successful cases were limited. Most of the literature still focused on the summary of construction technology and there was no further study on the deformation characteristics and the structural design for this new excavation method. Based on a new well-documented excavation project of Yongdingmen station abbreviated as the YDM station, this paper established an underwater excavation model, in which a series of 3D finite element analyses were carried out to investigate the performance and mechanism of the underwater excavation combined with the field observation. Figure 1 shows the stratigraphic section of the YDM station studied in this article. e soil layer above the water level was mainly alternating silt sand (SM) and silty clay layers (CL), and below the water level was a huge thickness of pebble stratum (Cb). e excavation bottom was located in the pebble stratum filled with water and large-diameter pebbles.

Project Background.
e pebble diameter greater than 20 mm was 50%∼80% and often mixed with pebbles and boulders, which has a maximum diameter of 80 mm. erefore, the YDM excavation was unsuitable for traditional dewatering excavation due to the high permeability and large water inflow at the excavation bottom. Furthermore, there was no thickness bearing Earth layer below the excavation bottom. If the water outside the pit was cut off directly, a diaphragm wall with a depth of 60 m was required, which will increase the project difficulty and cost waste. To solve the above construction problems and protect the groundwater resources in Beijing, the underwater excavation method was proposed for YDM excavation. Figure 2 shows the top view of the geometric parameters of the YDM project and the layout of the deflection monitoring points. e diaphragm wall with 1.2 m thick was used as the retaining wall of the excavation, on which 12 inclinometers were installed at different locations. Several 1 m thick cross walls divided the foundation pit into 16 bays. Figure 3 shows the schematic graph of retaining structure, stratum, and excavation (section A-A), which was finished in eight steps using the traditional excavation method combined with the underwater excavation method. e cross walls were backfilled within concrete between GL-32 m and GL-44 m (GL referred to the ground surface level). e concrete slabs 1BF, 2BF, and 3BF were the beam-floor systems, 4.5 m in spacing for 1BF, and 9 m in spacing for 2BF, respectively. A 4meter-thick underwater concrete bottom seal was performed. e construction depth was 36.5 m, and the groundwater level was GL-19.0 m. e toes of diaphragm walls and cross walls were embedded 8 m in the Cb layer. e critical steps of the YDM project are shown in Figure 4. Figure 5 illustrates the YDM project excavation and three-dimensional finite element mesh used for analysis. A 10-noded tetrahedral element was applied to simulate the soil volume. Soil movements normal to the four vertical sides were restrained (roller boundary), whereas soil movements were restrained in all directions at the bottom of the geometry (fixed boundary) [17]. e distance between the diaphragm walls and the outer boundary of the mesh was larger than five times the final excavation depth (He) to minimize the boundary effect. e hardening soil model [18], abbreviated as HS model, was adopted for this study. is model must input 11 Table 1.

Establishment of the Finite Element Model.
e structural members such as diaphragm walls, cross walls, and concrete slabs employed were made of plate elements and simulated as linear elastic material as well as the concrete strut. Also, the interface friction angle between the structure and the soil was the same as that of the soil, and the mechanical behavior of the contact interface was set according to the Mohr-Coulomb criterion. Considering the possibility of the microcracks in concrete after structural deflection, the elastic modulus of the structure has been reduced by 20% [26,27]. e input parameters of the soil layer and structure are all shown in Table 2.

Ground Surface Settlement.
e settlement of 3Danalysis in 8 excavation stages is extracted in Figure 6. e results show that, along with the Earth excavation, the value and the influential range of the soil settlement gradually increased. e main increase in settlement occurred before the underwater excavation (step 6). During the underwater excavation stage, the surface settlement increased slowly. Besides, the surface settlement in step 4 was obviously increased due to the strong Earth permeability of GL-22 to GL-28. Figure 7 showed the 3D-analysis results and field observations of surface settlement (δ v ) at different distances (d) from the middle line of the diaphragm wall. e 3Danalysis results were fit well with the field observations under various steps. e error of two analysis results was approximately 1∼2 mm, which was an allowable error range of projects. However, because the excavations equipment used for underwater excavation and underwater concrete filling were placed at the edge of the excavation for a long time, the result of the observation was slightly larger than the 3D-analysis result within 10 m from the edge of the excavation and decreased faster beyond 10 m away from the excavation.
As shown in Figure 7, most of the settlement occurred between step 1 and step 4. e δ v of 3D-analysis and observations were 12.4 mm and 13.2 mm, which had already accounted for 86.7% and 81.9% of the final settlement (14.3 mm and 16.1 mm), respectively. During the underwater excavation (step 6), the δ v of 3D-analysis and observation were increased by 1.3 mm and 1.6 mm, which only accounted for 9% and 10% of the final settlement. is increment was far less than the increased δ v of 2.8 mm and 3.1 mm in the drainage excavation (step 4) under the condition of little difference in excavation depth. To better explain the surface settlement characteristics during underwater excavation, the research results of Hsieh [28] and this study are plotted in Figure 8. e settlement influence area in the figure Step 5 Step 6 Step 7 Step 8 Step 2 Step 1 Step 3 Step 4 Underwater excavation

Normal excavation
Note: Step 1-first beam for construction Step 3-third beam for construction Step 5-recharge Step 7-underwater sealing Step 2-second beam for construction Step 4-drainage excavation Step 6-underwater excavation Step 8-pit pump Figure 3: Excavation geometry and stratigraphy for YDM project.  maximum settlement δ vm was at the ratio of (d/H e ) � 0.5 and δ v close to the wall was 0.5 times of δ vm . In this study, however, the main settlement area was within (d/H e ) � 1 while the secondary settlement area was little. e maximum settlement (δ v /δ vm ) � 1 was at the ratio of (d/H e ) � 0.25 ∼ 0.5, and δ v close to the wall was 0.25 times of δ vm .
Kun [29] and Xu [30] counted the maximum settlement distribution of excavations with traditional dredging excavation, as shown in Figure 9. e measured data indicated that δ vm was distributed between 0.1% He and 1.0% He, which was far larger than that of the YDM project (0.04% He on average). e above analysis further proved that the underwater excavation method had better resistance to the surface settlement; that is, the settlement influence range and the maximum settlement were much small than the predicted value compared with the traditional dredging excavation. Figure 10 shows the wall deflection of the 3D-analysis at 8 construction steps. e results showed that the wall deflection increased with Earth excavation but rapidly decreased around the position of the cross wall. Most of the lateral wall deflection occurred before drainage excavation (step 4). During the whole process of underwater excavation, the wall deflection was only slightly increased in the area between adjacent cross walls, and the maximum deflection position was no longer moving towards the excavation bottom. Figure 11 shows the 3D-analysis results and field observations of surface settlement (δ h ) at the midline of the diaphragm wall on the long side. ere was an error of 3∼4 mm between the observations and the 3D-analysis results. e reason was that, in step 2, the excavation depth was up to 10 m and the free face of the diaphragm wall was unsupported for a long time, as well as the mechanical load and construction disturbance. Moreover, the water level in the pit did not recharge to a certain height for the first time in the practical excavation, for which δ h of the observations increased below the third beam compared with those of the 3D-analysis. Although the error existed, the deformation trend and the deformation increment of two results in each step were similar.     Similar to the surface settlement, most of the deformation occurred between step 1 and step 4. e maximum δ h of 3D-analysis and observations reached 16.1 mm and 19.1 mm in step 4, which had already accounted for 96.4% and 97.4% of the final deformation (16.7 mm and 19.6 mm), respectively. During the underwater excavation (step 6), the maximum δ h of 3D-analysis and observation were increased by 1.2 mm and 1.3 mm, which only accounted for 7.2% and 6.6% of the final deformation. Besides, δ h had a certain reduction during the water recharge (step 5) and bottom sealing (step 7). erefore, adopting the measures of water recharge and bottom sealing in time was necessary to resist wall deformation during underwater excavation. Figures 12 and 13 show the relationship between the maximum wall deformation (δ hm ) and maximum deformation position (H δm ) with the relation between the excavation depth (H e ). Owing to the underwater excavation methods and the installation of cross walls, δ hm of the YDM project was about 0.06% H e . is value was far less than the statistical results by Ou [31] and Kung [29], which showed δ hm were nearly 0.2% He or 0.5% He , respectively. Ou [31] concluded that the position of δ hm was located approximately near the excavation surface and was basically within the range of H e − 5m∼H e + 5m. Nevertheless, the corresponding relationship was H δm � 0.5H e in YDM project. It proved that the location of maximum lateral deflection does      is phenomenon could be explained as follows. During the underwater excavation stage, the water pressure can resist part of Earth pressure, and the stress release speed of the diaphragm wall was far less than that of the dry excavation. Besides, the lateral wall deflection was limited when it extends to the position of the cross wall, for which the resistance effect was much better than the upper support.

Water Recharge.
To illustrate the effect of water recharge during the underwater excavation, the final wall deformation with water recharge and without water recharge is as shown in Figure 14. In the recharge area (GL-19 m ∼ GL36 m), δ h increased rapidly below the water level and H δm moved to the excavation bottom without water recharge condition. Compared with the water recharge excavation, δ hm reached 36.1 mm and increased by 58%, which was far beyond the deformation limit in the specification. Furthermore, δ h without water recharge decreased rapidly near the cross wall, but there was still a deformation of 5 mm at the feet of the diaphragm wall.
It indicated that, under the condition of dewatering excavation (without water recharge), the water head difference on both sides of the wall increased with the Earth excavation, which results in large water and Earth pressure acting on the diaphragm wall. Moreover, seepage action occurred at the corner of the wall due to the strong permeability of the pebble layer, which will lead to an inclined failure of the diaphragm wall. erefore, recharging the water level in time and keeping equal water head on both   Numerical results (YDM) Figure 13: He versus H δm for YDM t and Ou [31].
Advances in Civil Engineering sides of the wall was a key step in underwater excavation to reduce wall deformation.

Insertion Ratio of Diaphragm
Wall. e diaphragmcross wall system was the retaining structures to ensure the implementation of the underwater excavation. Figure 15 shows the optimization target parameters of retaining wall structure. e insertion ratio of the diaphragm wall was the ratio of the embedded depth of the diaphragm wall (H dw ) to the final excavation depth (H e ), as shown in Figure 15(a). Keeping other parameters of the model unchanged, the insertion ratio of the diaphragm wall was adjusted to 0.07, 0.14, 0.21, 0.28, and 0.35, respectively, and lateral wall deflection was made as in Figure 16. e deformation curve was basically coincident when the insertion ratio was 0.14, 0.21, 0.28, and 0.35, but increased at the insertion ratio of 0.07. e deformation decreased with the increase of the insertion ratio until it reached a limit value of 0.14 in the YDM project. Furthermore, the deformation still maintained a small value even when the insertion ratio was 0.07. erefore, with the underwater excavation method, the embedded depth of the diaphragm wall can be appropriately reduced to speed up the construction progress and save the cost.

Embedded Ratio of the Cross Wall.
eoretically, the deflection of the diaphragm wall was the smallest at the time of (h cw /h dw ) � 1, but it wasted construction costs. erefore, a critical embedded depth ratio (h cw /h dw ) cr was defined to study the optimal cross wall embedded ratio. Any value beyond this critical ratio would no longer cause the deflection value to decrease.
By plotting the relationship between the maximum lateral deflection (δ h max ) and the ratio of (h cw /h dw ), the critical embedding depth for this project can be expressed as (h cw /h dw ) cr � 0.5, as shown in Figure 17. When (h cw /h dw ) cr was less than 0.5, the value of δ h max decreased as the ratio increased; when (h cw /h dw ) cr was greater than 0.5, the value of δ h max remained unchanged. is meant that if the project did not take into account other construction factors such as concrete filling, it was most reasonable to set the embedded depth of the cross wall as half of the embedded depth of the diaphragm.

Interval of the Cross Wall.
For the specified excavation size, the ratio of the excavation width B to the cross wall interval L can be used to study the cross wall interval. Figure

Layout of the Cross Wall.
Under the condition that the number of the bays was determined, the arrangement forms of 8 vertical bays and 8 horizontal bays are designed to study the effect of the layout of the cross wall (Figure 15(b)). Figures 19 and 20 plot the wall displacement cloud chart and

Advances in Civil Engineering 11
the deflection at the midline of the long side of the diaphragm wall in the two different cases, respectively. Figure 19 plots the increase of the deflection occurring mainly between two cross walls after the excavation was near the cross wall. e wall deformation of the lower part was larger and it developed deeper to the bottom under the condition of the horizontal layout of 8 bays. At the depth of −26 m, the maximum displacement under the vertical layout of the cross wall bay was 11.50 mm, which had a reduction of 37% compared with the wall deflection under the horizontal bays at the same location that was 18.25 mm, as shown in Figure 20. erefore, the vertical cross wall bays had a better counteraction on the lateral deformation. While ensuring the convenience of concrete casting and underwater excavation, the cross wall can be reduced appropriately. However, the most reasonable way to reduce the cross wall was to minimize the cross wall parallel to the long side of the diaphragm wall and to ensure the cross wall perpendicular to the diaphragm wall. When choosing the same space area, it was better to use vertical cross wall bays.

Conclusions
Based on field observations and 3D-analysis results of the YDM project with the underwater excavation method, the deformation behavior of the surface settlement and the lateral wall deflection were conducted. e main findings and conclusions of this study were summarized as follows: (1) By the hardening soil model, a 3D numerical analysis method for the underwater excavation model was established to analyze the surface settlement and the lateral wall deflection. (2) Most surface settlements mainly occurred in the stage of dry excavation and dewatering excavation, and the deflection caused by underwater excavation only accounted for about 10% of the total settlement. Meanwhile, the influence range of settlement was approximately equal to the excavation depth. e maximum value of settlement occurred 0.25∼0.5 H e behind the retaining wall and the maximum value was basically 0.04% H e , which was much less than the predicted value under other dewatering excavations.
(3) Most of the lateral wall deflection has been caused before the underwater excavation, and the deflection only accounted for about 7% of the total deflection in the underwater excavation stage. e maximum wall deflection was about 0.06% H e , which was far less than the deflection statistics of other similar excavations. Besides, the location of the maximum wall deflection was half of the excavation depth (0.5 H e ). (4) A novel retaining wall structure based on the underwater excavation method was optimized, and the wall deformation was suppressed with a much smaller embedded ratio of 0.07. In fact, setting the critical interval ratio of the cross wall, (B/L) cr � 8, the critical embedded depth ratio of the cross wall, (h cw /h dw ) cr � 0.5, and using the vertical cross wall bays were more advantageous to controlling wall deformation. ickness of the diaphragm wall, m t cw : ickness of the cross wall, m L:

Abbreviations
Interval of the cross wall, m c: Cohesion, kPa φ: Internal friction angle,°e : Void ratio C s : Swelling index.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

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