5.2.1 Sarapuí clay site

Figure 5.2 shows the geotechnical profile of Sarapuí clay based on geotechnical investigation carried out near the trial embankments sites. An 11-m-thick, very soft clay layer, overlying a fine-to-coarse sand layer comprises the geotechnical profile. The liquid limit (w_L), plastic limit (w_P), measured without oven drying, and natural water content (w_n) profiles are also presented. Sarapuí clay water content is slightly higher than liquid limit, and the sensitivity of this clay is 4.4.

Sarapuí clay is lightly overconsolidated due to aging and water-table fluctuation. The initial voids ratio (e₀) decreases with depth from 4.9 to 2.46 and natural unit weight (γn) varies from 12.5 to 14.5 kN/m³. Below the crust, the overconsolidation ratio (OCR) of the clayey deposit varies from 2.0 to 1.3. Sarapuí clay is a very compressible soil; C_c (compression index) varies from 1.3 to 3.2. It has an average C_c/(1 + e₀) relationship of 0.41, as shown in Fig. 5.2.

Conventional and special oedometer tests were carried out to obtain coefficient of horizontal consolidation (c_h) values. Tests included inflow and outflow types of radial drainage tests, which are vertical drainage tests on samples at 90° with the horizontal plane (Lacerda et al., 1977, 1995). The representative value of c_h obtained at a normally consolidated range was 2.4 × 10^- 8 m²/s and it was found that the c_h/c_v relationship lies between 1.0 and 2.0, which is typical for very soft clays.

5.2.2 Trial embankment II

Embankment II was built 35 m wide and 315 m long and divided in seven instrumented subareas, where different kinds of drains were installed, as described in Table 5.1. Sections A and G, without drains, are not analyzed here.

Loading was applied in two main stages and the final height was about h = 3.6 m. The first stage of loading was applied during 1981, in two steps. The first step (h = 0.7 m) occurred over a month and the second, 200 days after, lasted approximately 2.5 months (h = 2.0 m). The second stage, in 1986, was applied only from section B to G, in one week, until final height occurred after approximately 3300 days of monitoring. Data records of the central settlement plates are shown in Fig. 5.3.

The lowest c_h value was obtained at a closed mandrel section, as expected, however, the monitoring of settlement plates showed that although sand-jetted drains presented initially high strain rates, the rate dropped with time and, in general, the strain rates of prefabricated drain sections were higher than those of the sand drains sections.

5.2.3 Coefficients of consolidation

The analysis presented here concentrates on the second loading stage, for which the clay layer was on a normally consolidated condition (Almeida and Ferreira, 1992).

Geometric data used in the consolidation solutions are given in Table 5.2. The main parameters used for the analysis of radial drainage are:

• d_w = diameter of the drain in the case of sand drains and equivalent diameter of the drain in the case of prefabricated drains

• d_s = diameter of the smeared zone and s = d_s/d_w

• k_h = horizontal coefficient of permeability of the undisturbed soil

• k′_h = horizontal coefficient of permeability in the smeared zone

The ratio s = d_s/d_w used for the analysis of radial drainage was based on literature recommendations (Hansbo et al., 1981; Jamiolkowski et al., 1983, 1985; Holtz et al., 1987; Hansbo, 1987). Values adopted were s = 2 for displacement type sand drains, s = 1 for jetted sand drains (also known as low displacement drains), and s = 1.5 for prefabricated drains. The ratio k_h/k′_h = 3, also used for the analysis of radial drainage, was based on the experimental data available for the Sarapuí clay. Further details about the parameters adopted in the analyses can be obtained in Almeida and Ferreira (1992).

Table 5.2 shows the average coefficient of vertical and horizontal consolidation obtained from laboratory tests, in situ tests, and back-calculated from settlement data using the Asaoka (1978) method.

Calculations of c_h presented in this section assumed just radial drainage, as preliminary calculations of combined drainage resulted in a c_h that was 10% smaller than when it was not considered.

The analysis of c_h values in Table 5.2 shows that the c_h values are fairly close. The smallest values are those from the laboratory c_h (lab) and pore pressure data c_h (u), and the greatest values are those from settlement data c_h (s) and piezocone data c_h (piez). The ratio c_h (s)/c_h (lab) is around 2.5, which is expected and usually attributed to the macrostructure of the deposit (e.g., small lenses and draining layers). In the present case, the top crust, however, is the stiffer and more permeable layer.

Comparison between c_h (s) and c_h (u) indicates that the former is about three times higher than the latter. The nonlinearity between effective stress and void ratios appears to be the reason for this discrepancy (Almeida and Ferreira, 1992). Comparison between c_h (lab) and c_h (piez) shows that laboratory value is smaller than the field value from piezocone for normally consolidated (nc) conditions. The data suggest that secondary consolidation had negligible influence on c_h (s) computations. However the influence of the secondary consolidation on the coefficient of vertical consolidation c_v was found to be quite substantial (Almeida and Ferreira, 1992).

5.3.1 Site description

The site under study consists of very soft fluvial–marine gray clay with a top layer of peat. Figure 5.4 shows data of water content and Atterberg limits. It is observed that water content decreases from around 500% close to the ground level to around 100% at the bottom of the clay deposit. The maximum thickness of the compressible soil is 12 m. A sandy soil of alluvial origin is found under the compressible soil, under which is the residual soil. Figure 5.4 also shows the variation in OCR and CR = C_c/(1 + e₀) with the depth. The average value of the overconsolidation ratio (OCR) is 1.5 below 2.0 m depth and the average compression ratio (CR) of the clay deposit is 0.52. A detailed discussion on the properties of this deposit was presented by Almeida (1998).

Settlements were measured by using 20 settlement plates installed before the embankment was built. Figure 5.5 shows the typical cross section of the completed embankment. The 0.60-m-thick drainage blanket provided support for the equipment used to install a prefabricated vertical drain with dimensions 0.1 m × 0.3 m, the equivalent drain diameter d_w = 6.6 cm in a 1.70 m triangular grid. Figure 5.6 shows settlement–time curves obtained from settlement plates. The embankment thickness has also been recorded for each settlement measured, as shown in Fig. 5.6. The last construction stage should have been performed when the embankment achieved 80% consolidation, but a delay occurred due to nontechnical issues.

Pore pressure observations showed that the drainage blanket and the underlying soil acted as drainage interfaces, thus vertical and radial consolidation were considered when analyzing settlement data to obtain coefficients of horizontal consolidation.

5.3.2 Coefficients of consolidation

Coefficients of consolidation at each settlement plate for the period up to 720 days were obtained using the Asaoka (1978) method modified by Magnan and Deroy (1980) for combined radial and vertical drainage, for which a ratio c_h/c_v = 1.5 was assumed, following local experience. Other parameters used were s = d_s/d_w = 1.5 and k_h/k′_h = 2. Values of c_h calculated up to 720 days for the 17 plates ranged from 3.7 to 10.3 × 10^- 8 m²/s and the average value was 6.8 × 10^- 8 m²/s. The majority of U values computed for this period were found to be close to 90% (Almeida et al., 2000).

The results of the coefficients of horizontal consolidation obtained from settlement data are compared here with those obtained from piezocone dissipation tests. The Houlsby and Teh (1988) method was used to obtain the c_h (nc) values for the normally consolidated band, from the piezocone tests, using the soil rigidity rate I_r = 93 obtained by Oliveira (1997) and the ratio C_s/C_c = 0.10. The results of c_h, obtained for 4 boreholes and 10 different depths was between 2.4 and 13.7 × 10^- 8 m²/s (Almeida et al., 2000). Special radial drainage oedometer tests (Coelho, 1997; Almeida et al., 2000) have also been computed. Figure 5.7 and Table 5.3 show a good agreement between c_h values obtained from the settlement analysis with the dissipation tests using the piezocone.

5.3.3 Shear strength

Two vane test campaigns were performed in the soft clay, the first in early 1996 before building the embankment and the second at the end of November 1997. Two verticals were carried out in each test campaign and all were performed (Nascimento, 1998) using the same electric vane equipment and at the same location of the site. The results of both campaigns are shown in Fig. 5.8.

To compare results from both vane test campaigns, taking into account the settlements that have occurred and the differences in depth in both test programs, the depth data was normalized with reference to the thickness of the layer at the time of the tests, thus the S_u data were reduced to the same normalized depths. Values of average undrained resistance S_u were computed for each normalized depth for the two test programs, which permit the calculation of the increased resistance ΔS_u given in Table 5.4.

Table 5.4

Analysis of the gain in strength

Normalized depth (m)	ΔSu (kPa)	ΔSu/Δσ′v
0.35	6.50 $\pm$ 1.31	0.25
0.40	8.85 $\pm$ 2.14	0.34
0.45	12.25 $\pm$ 0.67	0.47
0.50	11.86 $\pm$ 2.28	0.46
0.55	8.32 $\pm$ 2.06	0.32
0.60	1.65 $\pm$ 2.16	0.06
0.65	1.44 $\pm$ 2.51	0.05
0.70	1.39 $\pm$ 2.40	0.05
0.75	4.26 $\pm$ 2.91	0.17

In the design of embankments built in stages, it is common to compute the increased strength ΔS_u as a result of the increase in effective stress Δσ′_v. A conservative value ΔS_u/Δσ′_v = 0.22 may be adopted for the normalized increase in resistance. The ΔS_u/Δσ′_v values obtained for the present case are shown in Table 5.4. The computed value of Δσ′_v took into account the embankment submersion due to the settlement (measured nearby the vane tests) and also the degree of consolidation U = 90% estimated at the time when the vane test was carried out (Almeida et al., 2000). More than half the top part of the clay layer, with higher initial water content, presented ΔS_u/Δσ′_v above 0.22. However, extremely low ΔS_u/Δσ′_v values were obtained for the bottom half of the clay layer, with small water content. However, the average gain in strength of the whole clay layer is close to ΔS_u/Δσ′_v = 0.22. Therefore, the vertical drains have contributed to the improvement in the clay layer by increasing its resistance, although the magnitude of this increase has varied along the clay layer.

A test embankment has been constructed on well-studied Sarapuí soft clay site in which the rate of consolidation was accelerated using different types of vertical drains. Field observations have been made for about 10 years, and settlement and pore pressure data were back-analyzed and coefficients of horizontal consolidation calculated considering smear effects. Back-calculated c_h values from settlement and pore pressure data have shown a reasonable agreement with laboratory and in situ values.

Since vertical drains accelerate primary compression, the influence of secondary consolidation appears to be smaller on back-calculated c_h than on back-calculated c_v (Almeida and Ferreira, 1992).

It was observed that strain rates of prefabricated drain sections were higher than those of sand drain sections.

5.4.2 Barra da Tijuca

A settlement study was performed on the behavior of Barra da Tijuca very soft clay with prefabricated vertical drains. The settlements covered a period of up to two years, when the average degree of consolidation reached around 90%.

The coefficient of horizontal consolidation c_h obtained from settlement plates took into account the smear in the radial consolidation from the drain installation, as well as the vertical consolidation in parallel to the radial consolidation. Values of c_h were measured in special oedometer tests with drains installed in the sample, and in dissipation tests using the piezocone to compare with field values. The three sets of values were found to be quite close, with little dispersion of the laboratory values, followed by the field readings and lastly the piezocone data. The generally good agreement obtained suggests that during the period of analysis, the secondary consolidation was negligible compared to the primary consolidation.

The increase in clay resistance, assessed by the vane tests performed before and after the construction of the embankment, showed a higher gain in S_u for the top half of the clay layer with high water content and, oddly enough, a negligible gain for the bottom half of the clay layer with less water content.