Multidisciplinary Center for Earthquake Engineering Research logo google logo
navigation bar

Liquefaction and Post-Liquefaction Dissipation/Densification Characteristics of Silty Soils

(Task E2-1)

S. Thevanayagam, State University of New York at Buffalo
Geoffrey R. Martin, University of Southern California

Abstract

Detailed understanding of the pre- and post-liquefaction behavior of silty soils, their pore pressure generation, dissipation, and densification characteristics are input information needed for detailed analysis of seismic performance sites improved by stone columns and development of appropriate modifications to this techniques so that it can be applied to silty soils with a high level of confidence. This paper presents results from Task-I study of this project focusing on pre and post-liquefaction behavior of silty soils. Undrained cyclic triaxial loading followed by dissipation tests were conducted on three silty soils. Pore pressure generation characteristics are evaluated and compared with that of sands. Pre- and post-liquefaction compressibility and coefficient of consolidation, and densification characteristics are summarized. Implications of these findings on performance of sites containing non-plastic silts and silty sands, during installation of stone columns and during earthquakes are discussed.

Introduction

Current ground improvement design guidelines for liquefaction mitigation using drainage/densification techniques (dynamic densification, vibro-stone columns, deep blasting, etc.) have been developed mainly based on research and field experience with sands containing little or no silt content. Soil sites, with low silt content, improved by these techniques have performed well during earthquakes. Experience indicates that sites with high silt content, mostly non-plastic, also liquefy due to seismic loading. Research experience also indicates that these soils may behave differently from clean sands and are low permeable. The knowledge gained from past three decades of research on clean sands does not directly translate to silty soils. At present the above ground improvement techniques do not work well or face uncertainties on their applicability for soils containing silt content in excess of about 15 to 20% (Andrus and Chung, 1995). These techniques must be further modified to facilitate their application to silty soils. Recent preliminary experience indicates that installation of supplementary wick drains helps to relieve pore pressures developed during dynamic compaction and stone column installation in silty soils (Andrews 1998, Dise et al. 1994, and Luehring et al. 2000) and help densification. In the case of vibro-stone columns, the reinforcement effect has also been identified as a possible contributing factor that can help to reduce the seismic shear stress intensity experienced by soils surrounding the stone columns (Baez and Martin 1995). When combined together they may offer an effective technique for remediation of silty soils. Detailed understanding of the pre- and post-liquefaction behavior of silty soils, their pore pressure generation, dissipation, and densification characteristics, and possible reinforcement effect by stone columns are input information needed to develop appropriate modifications to the above techniques so that they can be applied to silty soils with a high level of confidence. Based on this knowledge appropriate analytical tools must also be developed to aid the design process and develop effective design guidelines.

This paper presents results from the Task-I work conducted as a part of this research focusing on pre- and post-liquefaction behavior of silty soils. In particular it addresses cyclic resistance, pore pressure generation, dissipation, pre- and post-liquefaction compressibility and coefficient of consolidation, and post-liquefaction densification characteristics. A detailed report summarizing the phase-I work and summary of results can be found elsewhere (Thevanayagam et al. 2000b, 2000c, Thevanayagam 2000b).

Experimental Program

Laboratory tests were carried out on three non-plastic silty soils: (i) artificial soil mixes of a sand (Foundry sand #55) and a non-plastic silt (Sil-co-sil#40), which are commercially available from U.S.Silica Company; (ii) a natural silt from Newark area, NJ, (iii) a natural silt from San Fernando area, CA. Unfortunately the natural silt obtained from San Fernando area turned out to be clayey silt, which was not suitable for further study. Table 1 summarizes the index properties of these soils. Table 2 presents a brief summary of the experimental program.

Specimen Preparation and Consolidation – Cylindrical specimens having typical dimensions of 155 mm in height and 75 mm in diameter were prepared using Moist Tamping Method. Each specimen was prepared at a different target void ratio. A known weight of dry solids required to reach the target void ratio was weighed and mixed thoroughly with water at water content of about 5%. The soil was divided into four equal portions. Each portion was poured into a mold mounted on a triaxial cell, and tamped gently using a wooden rod until the height corresponding to the target void ratio was achieved. The specimen was then percolated with CO2 and saturated with deaired water using backpressure saturation. The backpressure was increased gradually while maintaining the effective confining pressure at 15 to 20 kPa. This process was continued until the B (=Δu/Δσc , u = sample pressure, σc = cell pressure) factor exceeded 0.95. Following saturation, the specimens were consolidated to an effective isotropic consolidation stress of 100 kPa.

In each stage the amount of water flowing into or out of the specimen was recorded. Final void ratio at the end of consolidation of the specimen was calculated using the dry weight of the solids, specific gravity of solids, and net volume of water introduced into the specimen. The specimen notations are as follows: os25-408 = Ottawa sand/silt mix at 25% silt content and e=0.408.

Table 1. Index Properties

Property Ottawa Sand/Silt Ratio by Weight Natural Silt
100/0 85/15 75/25 40/60 0/100 Newark, NJ San Fernando, CA
Gs 2.65 - - - 2.65 - -
emin 0.608 0.428 0.309 0.413 0.627 - -
emax 0.79 0.75 0.86 1.35 2.10 - -
D10 (mm) 0.16 0.018 0.0085 0.0027 0.0015 0.02 0.0009
D30 (mm) 0.22 0.19 0.15 0.01 0.006 0.03 0.018
D50 (mm) 0.25 0.235 0.23 0.0285 0.01 0.04 0.06
D60 (mm) 0.27 0.245 0.24 0.07 0.015 0.05 0.09
Cu 1.69 13.61 28.24 25.93 10.00 2.5 100
Cg 1.12 8.19 11.03 0.53 1.60 0.9 4

emin = minimum void ratios (ASTM D1557), emax = maximum void ratios (ASTM D4254 method C).

Cyclic Loading – Undrained cyclic loading was applied using a triaxial apparatus (GEOCOMP Inc., MA). The tests were conducted at a constant cyclic stress ratio (CSR = Δσ1/2σc', σ1 = axial stress, σc' = effective confining pressure) of 0.2 at a frequency of 0.2 Hz. Following consolidation phase, a small amount of water was removed from the triaxial cell (surrounding the specimen) while the cell pressure was maintained the same as the value at the end of consolidation. This was done to make room for cyclic movement of the axial loading piston into and out of the triaxial cell during the cyclic loading phase without adversely affecting the cell pressure. For safety purposes, the maximum axial strain allowable was set at 8%. The axial displacement, cell pressure, and sample pore water pressure were monitored using a built-in data acquisition system.

Table 2. Experimental Program

Soils Tested

Fines Content (%)

D50

e
Range

No. of Tests

Cyclic Triaxial

Post-Liq. Dissip.

Hyd. Conduct.

Ottawa Sand + Silt mix

0

0.25

0.645~0.782

8

Yes

Yes

Yes

15

0.235

0.567~0.641

4

25

0.23

0.443~0.470

5

60

0.0285

0.490~0.545

6

100

0.01

0.811~0.878

8

Natural Silt - NJ

89

0.04

0.675~0.785

3

Yes

Yes

Yes

Natural Silt - CA

54

0.06

0.599

1

Yes

Yes

Yes

Post-Liquefaction Pore Pressure Dissipation and Volume Change – Once the specimen liquefied, cyclic loading phase was terminated. Post-liquefaction pore pressure dissipation test was initiated immediately. The bottom end of the specimen was connected to a pressure controlled volume measuring burette. The top end of the specimen was connected to a pore pressure transducer with no drainage allowed from this end. This setup simulated a one-way drainage condition. The dissipation tests were done in three stages. In the first stage the pressure in the burette was set at a value such that the post-consolidation effective stress in the specimen was 25kPa. In the second and third stages, the burette pressure was set at values such that the final effective stresses were, 50 and 100 kPa, respectively. The pore pressure at the top of the specimen and outflow volume of water from the bottom of the specimen versus elapsed time were recorded in each stage. The duration of each stage ranged from 16 sec to more than 3 hours, depending on the silt content of the specimen.

Analysis

Pore Pressure Generation – Figure 1 shows the pore pressure ratio, (ru = shear induced pore pressure due to cyclic loading/ σc') versus normalized number of cycles (N/Nl, N = no. of cycles, Nl = no. of cycles to reach liquefaction [at 5% double amplitude axial strain]) to reach liquefaction for a few specimens, at different silt contents. Also shown in this figure is the best-fit curve for sands proposed by Seed et al. (1976). The data for sand and silty sand up to about 25% silt content follow the pattern found by Seed et al. The pattern for silt and sandy silt at high silt contents appears to deviate from the above trend. The generation rate for the silts is a little faster at the beginning compared to sands.

Figure 1. Pore pressure generation: (a) sands (silt<25%) and (b) silts (silt>25%)

Pre- and Post-Liquefaction Compressibility Pre-liquefaction virgin consolidation lines and post-liquefaction consolidation lines (e vs. σ3') were drawn from volume change data obtained during the tests. Figures 2a-b show example plots for sand and silt, respectively. The post-liquefaction consolidation line is nearly parallel to the virgin consolidation line. This indicates that during liquefaction the soil skeleton is completely remolded and behaves like a freshly deposited soil. The post-liquefaction compression line follows a new virgin consolidation line differing from a typical recompression line.

Figure 2. Pre- and post-liquefaction consolidation: (a) sand:OS00-782 and (b) silt: OS100-878

Figures 3a-b show the pre- and post-liquefaction volume compressibility data (mv) for sand, silty sand, and sandy silt specimens. Compressibility values of silt and silty sand are of the same order of magnitude as that of sands at the same effective stress. Also pre- and post-liquefaction compressibility values do not differ significantly from one another.

Figure 3. Pre- and post-liquefaction volume compressibility

Pre- and Post-Liquefaction Coefficient of Consolidation – Pre-liquefaction coefficient of consolidation (cv) values were calculated based on hydraulic conductivity and volume compressibility data for virgin loading. Post-liquefaction cv values were back calculated using pore pressure and volume change measurements obtained during post-liquefaction dissipation tests. Back calculations were done by fitting the measured pore pressure vs. time data at the closed-end of the specimen to the theoretical solution for pore pressure at that end based on Terzaghi’s one dimensional consolidation (Coduto 1999), given by:

where Hdr = length of longest drainage path, Tv = time factor, u = excess pore pressure, Zdr = nearest distance to the drainage end, and Δσ = change in total stress. Similarly, volume change data was also used to back calculate cv using Terzaghi’s one dimensional consolidation theory.

Figures 4a-b show typical plots for each case, respectively, for a 100% silt specimen (os100-838) at void ratio of 0.838. The back-calculated values in each case are in close agreement except for minor differences.

Figure 4.  Back calculation of cv using (a) pressure measurements, and (b) volume measurements (U(top) = deg. of consolidation at top of specimen, U(avg) = Average deg. of consolidation)

Figures 5a-b show the cv values as a function of effective confining stress. For the same specimen, at the same effective confining stress, the cv values for pre-liquefaction consolidation along virgin loading and post-liquefaction cv values are nearly the same for the same soil. The reason for this is the same as that identified for the prior observation for mv. The soil is completely remolded following liquefaction and it behaves as a freshly deposited soil. Its consolidation line parallels that of the pre-liquefaction virgin loading.

None of the specimens was subjected to dissipation tests before liquefaction occurred. Hence, no direct data exists to determine the applicable cv value during generation of pore pressure up to liquefaction (viz. before remolding of soil structure). It is thought that the cv values for such generation/dissipation stages would be similar to the value during unloading/reloading stage of the soil along the recompression line.

Further, for the same soil, cv is lower by more than one order of magnitude at an effective confining stress of 10 kPa compared to its value at 100 kPa (Figs.5a-b). This indicates the need for use of confining stress dependent cv values for post-liquefaction dissipation analyses.

Coupled analysis of pore pressure generation and dissipation requires use of appropriate values adjusted to the status of the soil.

Figure 5. Coefficient of consolidation (a) pre-liquefaction (normally consolidated) and (b) post-liquefaction

The cv values are smaller for silty soils by more than one order of magnitude compared to sand. It is predominantly affected by the soil’s permeability, which is affected by silt content. Hydraulic conductivity of the soil specimens ranged from 0.6 to 1.3x10-3 cm/s for the sand, 9x10-5 cm/s for 15% silt, 0.6 to 1.2x10-5 cm/s for 25% silt, and 3 to 5x10-6 cm/s for 60% and 100% silt soils. The cv values are affected in the same order.

The cv values for the natural silt (NJ) in Figs.5a-b are somewhat higher than those for artificial silt (Sil-co-sil#40). But the mv values (Fig.3) are nearly the same. The difference in cv is due to the difference in grain size and permeability for these two silts. Grain size (d50) for the natural silt is about 38mm versus 10mm for the artificial silt. Hydraulic conductivity values are of the order of 2x10-4 cm/s for the natural silt versus 3 to 5x10-6 cm/s for the artificial silt. The difference in hydraulic conductivity is reflected in the cv values for these two silts.

Post-Liquefaction Densification – Quantification of post-liquefaction densification is an important aspect in performance evaluation of liquefiable soil sites. At present, there is only limited data available on this subject (Lee and Albaisa, 1974, Pyke et al., 1975, Silver and Seed, 1971a,b, Tokimatsu and Seed, 1987). The data are primarily limited to clean sands. The data from the current study sheds further light on this subject.

Figures 6a-c show the post-liquefaction densification data for the soils tested. There is no single relationship for volumetric strain due to densification against void ratio for all soils (Fig.6a). When the data are split into two parts, one for sands and silty sands up to 25% silt content, and the other for sandy silts, and plotted against the equivalent intergranular and interfine void ratios (ec)eq {= [e+(1-b)fc]/[1-(1-b)fc], b=a coefficient, 0<b<1, e = global void ratio, fc = FC/100, FC = (silt) fine content in percentage, and b = a coefficient}and (ef)eq {= e/[fc+(1-fc)/Rdm, m = a coefficient, Rd = D50/d50, D50 = 50% passing size of sand portion, and d50 = 50% passing size of fines portion] (Thevanayagam, 2000), respectively, the data for each group fall in a narrow band (Figs. 6b and c).

Figure 6. Post-liquefaction densification

Cyclic Strength – Figure 7 shows the number of cycles to cause liquefaction (Nl ) versus e for all soils, versus (ec)eq for soils at silt content less than or equal to 25%, and versus (ef)eq for soils at silt content greater than 25%, respectively, at a cyclic stress ratio of 0.2. At the same e, cyclic strength of a silty sand decreases with an increase in silt content. This trend reverses beyond a transition silt content (about 20 to 30%), and the strength increases with further increase in silt content. However, when plotted against (ec)eq, silty sands up to 25% silt content fall in a narrow band. Similarly, sandy silts (beyond the transition) correlates well with (ef)eq. It is also interesting to note that the above respective equivalent intergranular void ratios have also been found to correlate well with the energy required to cause liquefaction (Thevanayagam et al. 2000a) as well.

(a)

(b)

(c)

Figure 7. Number of Cycles to Liquefaction Vs Global Void Ratio and Equivalent Void Ratio

Conclusion

Laboratory undrained cyclic tests followed by dissipation were conducted to study pre- and post-liquefaction characteristics of silty soils. Analysis of the data shows the following.

  1. Pore pressure generation characteristics (ru vs N/Nl) for sand and silty sand up to 25% silt content follow the same trend found for sand by Seed et al. (1976). The generation rate for silt and sandy silt (silt > 25%) is somewhat faster than that of sand.
  2. Soil skeleton is completely remolded during liquefaction and, as a result, post-liquefaction compression line (e vs. σc') almost parallels the pre-liquefaction virgin compression line. It behaves like a freshly deposited soil.
  3. For the same soil, post-liquefaction volume compressibility (mv) and coefficient of consolidation (cv) values are similar to those of the normally consolidated virgin soil at the same stress level.
  4. Coefficient of consolidation for silty soils is lower by more than one or two orders of magnitude and is primarily affected by the silt content (viz. permeability of the soil). For the same soil, coefficient of consolidation values of soils at low confining stresses (10 kPa) immediately following liquefaction are about an order of magnitude less than that at 100 kPa confining stress. It is significantly stress dependent.
  5. At the same void ratio, silty sand (silt content<25%) densifies more than clean sand following liquefaction. Post-liquefaction volumetric densification of sand and silty sand up to 25% silt content correlates well with equivalent intergranular void ratio, (ec)eq. Post-liquefaction volumetric densification of silty soils correlates well with equivalent interfine void ratio, (ef)eq.
  6. At the same void ratio, liquefaction resistance of a silty sand decreases with an increase in silt content. This trend reverses beyond a transition silt content (about 20 to 30%), and the resistance increases with further increase in silt content. When plotted against equivalent intergranular void ratio (ec)eq, cyclic strength of silty sand (up to 25% silt content) and clean sand fall in a narrow band. Similar observation is made when silty soils are compared against equivalent interfine void ratio (ef)eq.

Ground improvement schemes for liquefaction mitigation in silty soils based on densification/ drainage methods (dynamic compaction, vibro-stone columns, etc.) need to take into consideration of the differences in the above behavior characteristics of silts compared to sands. Primarily, permeability (and silt content) affects the dissipation characteristics of silty soils compared to sands. This requires much closer spacing of dynamic compaction grids or stone columns with provision of additional means such as supplementary wick drains to expedite dissipation of pore pressures developed during ground improvement operation in silty soils. Further work is ongoing towards developing numerical tools for analysis of such combined systems.

References

Andrews, D.C.A. (1998). “Liquefaction of silty soils: susceptibility, deformation, and remediation.” PhD Dissertation, Dept. of Civil Eng., USC, CA.

Andrus, R.D., and R.M. Chung (1995), “Ground improvement techniques for liquefaction remediation near existing lifelines.”, NISTIR report # 5714, Building and fire research laboratory, National Institute of Standards and Technology, Gaithersburg, MD 20899.

Baez, J.I., and G. Martin. (1995). “Permeability and shear wave velocity of vibro-replacement stone columns.” GSP 49, ASCE, NY, 66-81.

Coduto, D.P. (1999). Geotechnical Engineering, Prentice Hall, Inc., Upper Saddle River, NJ.

Dise, K., M.G. Stevens, and J.L. Von Thun. (1994). “Dynamic compaction to remediate liquefiable embankment foundation soils.” GSP No.45, ASCE, Reston, VA.

Lee, K.L., and A. Albaisa. (1974). “Earthquake induced settlements in saturated sands.” J. Geotech. Eng. Div., ASCE, 100(4), 387-406.

Luehring, R., B. Dewey, L. Mejia, M. Stevens, and J. Baez. (2000). “Liquefaction mitigation of a silty dam foundation using vibro-stone columns and drainage wicks - a test section case history at salmon lake dam.” No.00-0748, Unpublished report.

Pyke, R., H.B. Seed and C.K. Chan. (1975). “Settlement of sands under multi directional shaking.” J. Geotech. Eng. Div., ASCE, 101(4), 379-397.

Seed, H.B, P.P. Martin and J. Lysmer. (1976). “ Pore water pressure change during soil liquefaction.” J. Geotech. Eng. Div., ASCE, Vol.102(4), 323-346.

Seed, H.B., and J.R. Booker. (1977). “Stabilization of potentially liquefiable sand deposits using gravel drains.” J. Geotech. Eng. Div., ASCE, 103(7), 757-68.

Silver, M.L., and H.B. Seed. (1971a). “Deformation characteristics of sands under cyclic loading.” J. Soil Mech. and Found. Div., ASCE, 97(SM8), 1081-98.

Silver, M.L., and H.B. Seed. (1971b). “Volume changes in sands during cyclic loading.” J. Soil Mech. and Found. Div., ASCE, 97(SM9), 1171-82.

Singh, S. (1994). “Liquefaction characteristics of silts.” GSP. 44, S. Prakash and P. Dakoulas, eds., ASCE, Reston, VA, 105-116.  

Thevanayagam, S. (2000a). “Liquefaction potential and undrained fragility of silty soils.” Proc. 12th World Conf. Earthq. Eng., New Zealand.

Thevanayagam, S. (2000b) “Liquefaction in silty soils – Considerations for screening and retrofit strategies”, Proc. 2ndIntl. Workshop: Mitigation of Seis. Effects on Transp. Struct., Loh and Buckle, I., eds., NCREE, Taipei, Taiwan.

Thevanayagam, S., J. Liang, and T. Shenthan. (2000a). “Contact index and liquefaction potential of silty and gravely soils.” Proc. 14th ASCE Eng. Mech. Conf., Austin, Texas.

Thevanayagam, S., M. Fiorillo, and J. Liang (2000b). “Effect of non-plastic fines on undrained cyclic strength of silty sands.” GSP107, ASCE, Reston, VA, 77-91.

Thevanayagam et al. (2000c). “Ground remediation for silty soils using stone columns-Task 1: Liquefaction and post-liquefaction dissipation/densification characteristics of silty soils.”, Draft report, TEA-21 project No. 150-C104E, Oct. 2000, SUNY Buffalo.

Thevanayagam, S., Martin, G. R., Shenthan, T., and Liang, J. (2001) “Post-liquefaction pore pressure dissipation and densification in silty soils”, Proc. 4th Intl. Conf. On Soil Dynamics & Earthquake Engineering, San Diego, CA, S. Prakash, ed. (in press).

Tokimatsu, K., and H.B. Seed. (1987). “Evaluation of settlements in sands due to earthquake shaking.” J. Geotech. Eng. Div., ASCE, 113(8), 861-78.


  Contact Us  |  Acknowledgements   |  Disclaimer  |  Copyright© 2007 by the Research Foundation of the State of New York. All rights reserved.