Analysis Approach
The time-dependent consolidation settlement of clay is governed by the diffusion differential equation, which was recast in one dimension by Terzaghi (
1925) as the consolidation differential equation. Analytical solutions to the consolidation equation exist for simple cases, but those solutions do not consider time variations in loading, hydrostatic pressures, and coefficient of consolidation that are important for the present application. Accordingly, we solved the consolidation equation using a finite-difference recurrence approach. This approach, which adapts recurrence relations developed by Harr (
1966) as presented in Section 9.3.2 of Holtz et al. (
2011) to incorporate the aforementioned time-variable quantities, is described in Section
S1 in the Supplemental Materials. The 1D analyses assumed double drainage at the top and bottom of both the upper and lower OBC units (relatively granular soils at the interface of these units were taken as a drainage boundary).
Secondary compression was considered in the analysis both during primary consolidation () and following primary consolidation () in a manner that accounted for the effects of OCR on the secondary compression strain index () of individual layers. Section S1.5 in the Supplemental Materials describes how secondary compression was considered simultaneously with primary consolidation, and Section S1.6 describes how the overconsolidation effect on secondary compression rates was accounted for.
Secondary compression settlement for a layer of thickness
over time interval
was computed as
where
= secondary compression strain index at depth
;
= ratio of the reduced secondary compression strain index due to overconsolidation and the secondary compression strain index at depth
;
= elapsed time since the load was applied that induced primary consolidation; and
to
=time interval for which the secondary compression increment
is computed. During initial loading, the ratio
is taken as zero for
and is otherwise estimated as described in Section
S1.6. Because secondary compression is neglected for high OCRs during initial loading, analysis of secondary compression associated with recompression requires an assessment of the time when
, which is denoted
.
Using these procedures for analysis of secondary compression increments, the secondary compression settlement for the entire stratum was computed as
where
is from Eq. (
3).
Soil Properties
Material parameters used in the consolidation and secondary compression analyses are derived from the test data presented in the “Site Characteristics” section. The specific soil layering and baseline properties used in the analyses are given in Table
2. As shown in Figs.
3 and
16, baseline profiles for each property were established for the upper and lower OBC units based on the data trends. Variations of compressibility parameters (i.e.,
and
) and rate parameters (
for normally consolidated and overconsolidated conditions;
) were considered in formulating 12 alternative parameter sets (referred to as Runs 1–12), as shown in Table
3. The specific changes in compressibility parameter values shown in Table
3 are up to approximately one standard deviation (i.e.,
,
,
) based on the data scatter shown in Fig.
3. Two additional parameter sets were considered for higher and lower
profiles in the upper OBC, as shown in Fig.
16 (referred to as Runs 13 and 14). These alternative parameter sets were formulated in consideration of data scatter, parameter correlations (e.g.,
,
, and
were varied from baseline in a similar manner), and degree of realism given data trends (discussed further subsequently). Variations in the depth to groundwater over the time horizon of the initial construction (i.e., through 2007) were considered in additional runs and found not to significantly influence results; these results are not presented here for brevity.
Results
Using the approach described previously with the net foundation loading in Fig.
19 and the groundwater variations in Fig.
7, consolidation, secondary compression, and immediate settlements were computed for baseline soil properties. These are shown in Fig.
20, as well as the total predicted settlements and measured settlements (planar average and SM11 and 27 to indicate the range). Consolidation made up about 70% of the cumulative settlement as of early 2020, the balance being roughly equally divided among immediate settlements and secondary compression. Immediate settlements have been essentially constant since 2009. Secondary compression has been a significant portion of the continuing settlement since early 2019.
The overall settlement amount compares favorably to the planar average, including the relatively modest rate of settlement since early 2019. As described in the “Foundation Settlement” section, measured foundation settlements accelerated at several points in time coinciding with construction activities (excavation, dewatering) at adjacent sites. These effects were also evident from the simulations, most notably as downward inflections of consolidation settlements in mid-2012 and the start of 2015. The principal misfit of the 1D simulations from the planar average settlement was an underprediction of settlement rate starting in 2013, at the time of STC Zone 3–4 and SFE dewatering. This underprediction causes simulated settlements to fall below measurements from 2013 to 2018. Potential causes of underpredicted settlement rates in 2013 include too-slow consolidation in the 1D model resulting from 3D flow dissipating excess pore pressures and potentially shear-related movements from excavations (examined further in the next subsection).
In the “Performance Assessment” section, we inferred the relative contributions of primary consolidation and secondary compression on observed settlement rates. This issue is explored further by interpreting the excess pore water pressure distribution with depth from the 1D simulations to evaluate the average degree of consolidation (
), computed as
where
= stress change inducing consolidation at depth
(influenced by
and groundwater changes, as discussed in Section
S1); and
= excess pore pressure at depth
and time
. At the end of 2011, which immediately preceded the onset of major STC construction,
was approximately 81% for the upper OBC and 100% for the lower OBC. This suggests that primary consolidation may have contributed modestly to the settlement rate shown in Fig.
15. In April 2019 and May 2020, the 1D simulations indicated
and 98%, respectively, for the upper OBC and
for the lower OBC at both times, which is consistent with inferences of consolidation having been essentially completed from both observed settlement rates (Fig.
18) and measured pore pressures (Fig.
8).
The groundwater rebound of 2.1 m that occurred between 2018 and mid-2020 lowered
and has overconsolidated portions of the upper OBC, which previously experienced virgin compression. The effective OCR of this sublayer was 1.02 in May 2020 in the 1D analysis, which slows secondary compression by reducing
. This effect is included in the computed settlements shown in Fig.
20.
Fig.
20 also shows variations of total settlements across the 15 runs reflecting soil parametric variability. The variability among the settlement results was modest, which is somewhat by design—the variations shifted compressibility parameters and
in opposite directions relative to baseline (e.g., Run 1 had lower compressibility, higher
; Run 2 had higher compressibility, lower
), which was done to reflect correlations between these parameters. The results show that reasonable variations in soil compressibility parameters and
did not affect the ability of 1D methods to capture the main settlement features. On the other hand, uniform modifications up or down of preconsolidation pressures across the OBC profile (not included in the 15 variability runs) shifted down and up, respectively, computed settlements. Such results are not shown in Fig.
20.
As described previously, the
profile used in the analysis (Fig.
16) was selected to represent trends in laboratory data while also producing computed settlements that are consistent with measurements. Variations in the
profile were considered for the upper OBC as shown in Fig.
16. If these
profile changes were applied while maintaining all other properties at baseline levels, settlements would shift up and accelerate for lower
and shift down and decelerate for higher
, producing mismatches with observation. Instead, we modestly adjusted compressibility and rate parameters along with
, as shown in the last two rows of Table
3, to see if reasonable agreement with data could be obtained. For the lower
case (Run 13), even with reduced compressibility parameters (within the range of data in Fig.
3), calculated settlements exceeded observations and had too-high settlement rates in mid-2020 (indicating continuing consolidation). This indicates that the lower
profile cannot reasonably replicate field performance when used in 1D analyses. For the higher
case (Run 14), the calculated settlement approximately matched the lower bound observed settlement for the initial loading and postconstruction time period. However, the
profile was too high for significant virgin compression to occur, which produced mismatched settlement rates, especially from dewatering (since 2012).