Analysis of Deep Excavation in Boston Blue Clay with Hardening Soil Model, Limit-Equilibrium and Winkler Analysis Methods
The basement of the Stata Center building on the MIT campus required a 12.8 m deep excavation covering a large open-plan site. The excavation base was underlain by more than 25 m of medium to lightly overconsolidated Boston Blue Clay and was supported by a floating, perimeter diaphragm wall braced with internal corner struts, rakers, and tieback anchors.
To better control deformations and impact on adjacent structures, the project involved a complex sequence of berms, access ramps, and phased construction of the concrete mat foundation. Lateral and vertical deformations were monitored throughout the project construction. The project was examined by Orazalin, Whittle, and Olsen, who performed a 3D-FEM analysis back in 2011, and 2015.
In this page, the project was reanalyzed with the 2D finite element module of DeepEX. The original paper utilized a Mohr-Coulomb elasticity model for all soils except for the clay. In this reanalysis, the Boston-Blue-Clay was modelled with hardening soil properties, whereas all other soils utilized essentially the same elasticity properties as in the original work. The model was also analyzed with limit-equilibrium as well as non-linear Winkler elastoplastic analysis methods. Figure 1 presents a typical section through the excavation along a monitored north-south section. The following images present critical stages in the construction staging from DeepEX.
Figure 1: Typical north-south cross section for Stata basement excavation (Orazalin, et. al.)
Figure 2.1: Cantilever excavation
Figure 2.2: 1st tieback and excavation
Figure 2.3: 2nd tieback and excavation
Figure 2.4: 3rd tieback and excavation
Figure 2.5: 1st raker and excavation
Figure 2.6: 2nd raker and final excavation
Figures 3.1 to 3.3 demonstrate that monitored lateral wall displacements were comparable to the finite element results.
Figure 3.1: South wall bending moments and lateral displacements on the final stage
Figure 3.2: North wall bending moments and lateral displacements on the final stage
Figure 3.3: Reporter lateral wall displacements at two inclinometer locations vs. original 3D FEM model
Through the analysis, the magnitude of the assumed external building surcharge played a significant role is the obtained displacements. In the LEM method, we utilized the FHWA apparent earth pressure diagram with the CALTRANS beam analysis approach, assuming zero wall friction on the driving soil side. The support reactions from this method were generally similar to results from the FEM analysis. The LEM model on the north wall produced more similar bending moment magnitudes vs. the south wall. An important though observation was that the anchors on the south wall had very small free lengths that were well within the zone of movements. It is thus very likely that the ground anchors did not fully restrain the wall which in turn resulted in greater bending moments.
Figure 4: LEM support reactions and FHWA pressures
In the Winkler-Based non-linear analysis (NL) bending moments and displacements were more comparable for the raker section. In the tieback supported section the analysis produced similar anchor reactions to the FEM model, but the displacements were much smaller. The Winkler analysis cannot capture the basal movements and the movement of the anchors that are within the active zone of soil movement. Also, to make the NL analysis converge we had to assume wall adhesion in the clay. In FEM analysis, wall friction and adhesion are inherently assumed otherwise an FEM deep excavation solution cannot converge. In a separate model where ground anchor free length was increased, the bending wall moments dropped by about 30 percent.
Figure 5: Winkler-NL reactions, wall bending moments, and lateral displacements
Figure 6: Reanalysis of south wall with increased tieback free length
The current 2D finite element analysis was able to better capture surface settlements vs. the original 3D FEM with the MC model. The measured settlements were larger near the wall, which likely reflect some influence from construction installation (Figure 7).
Figure 7: 2D FEM settlement results (gray line) vs. measured data (original 3D FEM in dotted blue and red lines) \
All analysis methods will generally produce reasonably consistent support reactions. When basal stability movements are significant, the finite element analysis produces greater displacements that cannot be easily modeled for anchored systems with the limit-equilibrium and the non-linear analysis methods. In controlling wall displacements, our investigation confirmed that ground anchor free length needs to extend significantly to more stable soil zones. In such cases, it can prove significantly difficult to restrain wall displacements under 25mm without ground improvement.