Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

CPTu-BASED STATE CHARACTERIZATION OF TAILINGS LIQUEFACTION

SUSCEPTIBILITY

Christina Winckler, P.E., Senior Geotechnical Engineer1


Richard Davidson, P.E., Senior Principal, Vice President2
Lisa Yenne, P.E., Manager Mining Group, Vice President3
Joergen Pilz, Principal Advisor4

ABSTRACT

A new piezo cone penetration testing (CPTu)-based state characterization approach has
been developed to augment conventional liquefaction triggering analysis. CPTu
soundings through tailings were used to illustrate and compare results for these two
methods. Conventional triggering analysis is generally applicable to the upper 100 feet
below the ground surface, while the CPTu-based state characterization provides a
continuous record throughout the profile to greater depth. This new practical
methodology provides a supplemental screening tool to evaluate liquefaction
susceptibility, in addition to the factor of safety approach by triggering analysis.
The CPTu-based state characterization provides an independent “first principles”
approach to evaluating whether the materials would behave in a dilative or contractive
manner during static or seismic loading. The CPTu-based state characterization included
four different methods to evaluate the state of saturated non-plastic tailings. If all of these
methods, including conventional triggering analysis, point to liquefaction susceptibility
or alternatively dilatant behavior, then confidence in the characterization is increased.
Findings presented from a static liquefaction case history support this characterization
method.

INTRODUCTION

Tailings dams present unique challenges to geotechnical engineers because they comprise
recently placed, hydraulically deposited and highly interbedded, angular non-plastic
sands and silts. When saturated, tailings can be susceptible to earthquake-induced
strength reduction and potential liquefaction. A number of cases of static and post-
earthquake liquefaction failures of tailings dam have been observed over the years
(Davidson, et al., 2011). Therefore, it is essential for these important structures to be
evaluated with a full suite of empirical and behavior-based tools available to the
profession.

1
Christina Winckler, URS Corporation, 8181 East Tufts Ave., Denver, CO 80237, 303-740-3986,
Christina.Winckler@urs.com
2
Richard Davidson, URS Corporation, 8181 East Tufts Ave., Denver, CO 80237, 303-796-4655,
Richard.Davidson@urs.com
3
Lisa Yenne, URS Corporation, 8181 East Tufts Ave., Denver, CO 80237, 303-796-4626,
Lisa.Yenne@urs.com
4
Joergen Pilz, Rio Tinto T&I, 4700 Daybreak Parkway, South Jordan, UT 84095, 801-204-2343,
Joergen.Pilz@riotinto.com

USSD will insert footer text here 1


One of the most powerful characterization tools available is the piezocone penetration
test (CPTu), see Photo 1, which provides a continuous record of tip resistance (related to
density and strength), sleeve friction (a measure of remolded strength and controlled by
effective stress and friction), and pore pressure. Interpretation of CPTu parameters
provides an excellent indication of the variability of tailings properties, but more
importantly, characteristic signatures of each tailings deposition regime. Pore pressure
dissipation tests provide equilibrium pore pressures within the impoundment at the time
of the sounding. With at least three dissipation tests at various depths in the sounding, the
phreatic surface profile and percent of full hydrostatic conditions can be inferred.

Photo 1. CPTu Sounding Investigation in Tailings.

Much work has been conducted in the last twenty years using CPTu data for liquefaction
potential assessment in alluvial and lake sediments, and hydraulically deposited materials
such as tailings. Robertson (2010) provides a useful review of these developments.

A conventional earthquake (seismic) triggering analysis (e.g. Idriss and Boulanger, 2008)
can be used to assess the liquefaction triggering of saturated tailings using CPTu data.
The conventional earthquake triggering analysis typically consists of comparing the
calculated cyclic resistance ratio (CRR) from CPTu soundings to the cyclic stress ratio
(CSR) calculated from either 1- or 2-D site response analysis. This empirical approach
was initially developed from correlations to Standard Penetration Tests (SPT) and has
largely been adopted by building codes and traditionally been used to depths less than
100 ft for which the data correlations were derived. However, many tailings
impoundments are much higher than 100 ft.

A new CPTu-based state characterization provides a “first principles” approach to


tailings material behavior under both dynamic and static loading conditions, which
overcomes the empirical depth limitation. The characterization provides an independent

2 Dams and Extreme Events


assessment of whether the materials would behave in a dilative or contractive manner
during loading.

This paper examines the application of these methods for a typical tailings dam that was
constructed using the upstream method and where multiple CPTu soundings have been
advanced at various locations. Additionally, these methods have also been evaluated for a
case history that experienced static liquefaction.

CONVENTIONAL TRIGGERING ANALYSIS

CPTu soundings along a study section are utilized to evaluate the behavior at the interior,
below the crest, and near the downstream toe of a tailings dam. The factor of safety
against seismic liquefaction is computed as:

Factor of Safety = CRR/CSR (1)

The calculations for the CRR and CSR ratios are described in more detail below.

Cyclic Resistance Ratio (CRR)

Resistance of soils to cyclic loading is usually represented in terms of a cyclic resistance


ratio (CRR), which is defined as the ratio of cyclic shear stress that will cause cyclic
liquefaction (i.e. zero effective stress). Because of its reproducibility and continuous
record of data, the CPTu was used to estimate the CRR values.
The procedure by Idriss and Boulanger (2008) is used to calculate the CRR profiles for
CPTu soundings. For fines contents (material passing the number 200 sieve with a
particle size of 0.074 mm) greater than five percent, a fines correction is used to increase
the soil resistance against liquefaction. Tailings fines contents are estimated from co-
located soils borings and laboratory sampling. Typically, an earthquake magnitude
scaling factor is applied to adjust the CRR7.5 for magnitudes other than 7.5. CPTu
soundings have been advanced over a period of time along the study section. Due to the
varying age of the CPTu soundings and the differences in effective stress conditions from
when the CPTu was pushed and the current stress conditions, the stresses used to
normalize the cone data and calculate the overburden stress coefficient Kmay be
different. To account for the varying age of the CPTu soundings analyzed, the phreatic
surface and effective stresses (at the time the cone was pushed) were used to calculate the
normalized tip and sleeve friction in the CRR calculation, while the overburden
correction K was calculated from the recent phreatic and effective stress conditions. The
normalization procedure is key to using the CPTu at a range of depths.

USSD will insert footer text here 3


Cyclic Stress Ratio (CSR)

There are various 1- and 2-D analysis tools available to evaluate the site response of a
tailings dam to earthquake ground motions. For 2-D analysis, the QUAD4 (Hudson et al.,
1993) computer program is often used to perform equivalent-linear elastic, plain strain,
finite element analysis. The program analyzes the response of soil to vertically
propagating earthquake ground motions, which are represented by an acceleration time
history specified at the base of the finite element model. QUAD4 allows for a compliant
base, which models the ability of the material underlying the finite element mesh to
absorb the reflecting seismic energy by using an elastic half space. The properties of
elastic half-space are defined by its total unit weight and shear wave velocity. Boundary
nodes on the two vertical faces at the lateral extremes of the mesh and the base of the
mesh were connected to a compliant base with dashpots such that energy reflecting from
the ground surface can be transmitted back into the foundation half space. Material
properties needed for the site response analysis are: maximum shear modulus (Gmax),
shear modulus and material damping variation with strain, and Poisson’s ratio, which are
described in more detail below.

Maximum Shear Modulus. Maximum shear modulus (Gmax) represents the stiffness of a
material at very small strain levels. In granular materials, the maximum shear modulus
usually increases with increasing confining pressure and can be expressed as a function
of the mean effective confining stress (’m). The maximum shear modulus was compared
to in situ shear wave velocities for both foundation and tailings material. Best fit Gmax
relationships were developed for each material type. The following relationship was
derived for the tailings:

Gmax = (630 (’m/1000)0.88) 1000 (2)

Shear Modulus and Damping Variations with Strain. As strains increase in the soil mass
as a result of the intensity of earthquake shaking, the soils progressively start behaving
non-linearly and show a decrease in shear modulus and an increase in intergranular
material damping. The reduction in shear modulus with increasing shear strains are
typically normalized with respect to Gmax and expressed in a normalized modulus
(G/Gmax) reduction versus strain relationship, which is dependent on the material type.
The current state of practice employs confining stress dependent shear modulus reduction
and material damping relationships for granular non-plastic materials. A sensitivity
analysis of the effects of the shear modulus and material damping curve was also
performed. The study showed that there is a first order effect on site response and
deformation results due to the selected shear modulus and material damping curves used
for the non-plastic tailings.

Published damping curves are often based on natural materials. For tailings, site-specific
confining pressure dependent curves were developed by Professor Ken Stokoe and his
team at the University of Texas at Austin. Two tailings samples were obtained at depths
of approximately 20 and 150 feet for testing. The purpose was to evaluate if there is a
change in tailings behavior with depth. Sample specimens from 20 feet were classified as

4 Dams and Extreme Events


a SM to ML material with water contents between 18 and 32 percent and dry unit weights
between 86 and 90 pcf. Sample specimens from the 150 foot depth were classified as ML
material with water contents between 25 and 29 percent, and dry unit weights between 94
and 99 pcf.

The testing program included performing torsional shear (TS) and resonant column (RC)
testing on both samples. TS tests were performed for low strain measurements at
confining stresses of 8, 17, 35, and 140 psi. RC tests were performed for large strains up
to 0.1 percent at confining stresses of 8, 17, 35, and 140 psi. The shear modulus reduction
and material damping curves were generated for the tested confining stresses and
extrapolated to obtain curves for confining pressures of 4 and 70 psi.

The effect of confining stress was found to be similar in the shallow and deep tailings
samples, which resulting in the development of one set of shear modulus reduction and
material damping curves for the tested confining pressures. RC and 10th cycle TS data
from the testing program are shown on Figures 1 and 2. The shear modulus reduction
data shows good agreement between the TS and RC test data results. However, the RC
data show lower damping ratios at strains of 0.01 to 0.1 percent than the TS data. This is
because the RC is performed with higher equivalent number of cycles, approximately
1,000 cycles. The 10th cycle TS data was used to obtain the damping ratio relationships at
strains below 0.1 percent. The shear modulus reduction curves were developed using the
RC and 10th cycle TS data, as shown on Figure 1.

Figure 1. Confining Pressure Dependent Shear Modulus Reduction Curves and


Laboratory Test Data.

A mathematical model similar to that used in the work by Darendeli (2001) was
developed to fit the TS damping results for strains less than 0.1 percent. This

USSD will insert footer text here 5


mathematical model generated material damping curves that, at high strains, have a
damping ratio that is not representative of tailings material behavior. The curves were
therefore revised based on other tailings results obtained for damping strains greater than
0.1 percent, as shown on Figure 2.

Figure 2. Confining Pressure Dependent Damping Ratio Curves.

Poisson’s Ratio. For the 2-D site response analyses, a Poisson’s ratio of 0.48 was
assigned to submerged or saturated materials, while a Poisson’s ratio of 0.35 was used for
unsaturated materials.

CSR Results. The maximum induced shear stress (max) was calculated in the 2-D site
response analyses for each element in the finite element mesh. This max represents the
peak shear stress, which occurs only once throughout the time history and only for a
fraction of a second. In dam design practice, this peak value is typically reduced to
represent an average shear stress (ave) that is 65 percent of max (Seed and Idriss, 1982).
The computed ave is then divided by the corresponding ’v to estimate the cyclic stress
ratio (CSR = /’v). For this study, the 65 percent reduction was also adopted to estimate
the CSR. The calculated CSR value for the study section is shown on Figure 3.

6 Dams and Extreme Events


Figure 3. Calculated Cyclic Stress Ratios and CPTu Sounding Location for Study
Section.

Results

The calculated CRR values from the CPTu soundings ranged between 0.08 and 1.2,
resulting in a calculated factor of safety of less than 1.0 in some areas. However,
triggering analysis is generally applicable to the upper 100 feet of sediment below the
ground surface and does not take into account confining pressure effects. Therefore, to
evaluate the extent of potentially liquefiable tailings, the results from a new CPTu-based
state characterization method were used.

CPTu-BASED STATE CHARACTERIZATION

In recognition of the limitation of conventional triggering analysis to extend beyond a


depth of 100 feet, a CPTu-based state characterization can be used to evaluate potentially
liquefiable tailings under earthquake loading.

The CPTu-based state characterization includes four methods to evaluate the state of the
saturated non-plastic tailings. The four methods include:

 comparison between recorded static and dynamic pore pressures;


 development of a normalized pore pressure parameter (P-value);
 plotting of normalized CPTu data plots with state parameter lines; and
 plotting of state parameter difference data plots.

CPTu soundings presented on Figures 4 through 6, illustrate these methods. The CPTu
soundings CPT-B and CPT-D shown on Figure 3 were used to illustrate the results of
these methods. Each method is described in more detail below. This new practical
methodology is dependent on maintaining full saturation of the CPTu pore pressure
sensor during penetration.

Another important part of the characterization is sampling of the tailings materials to


establish index properties such as density, water content, plasticity and gradation.
Piezometer data is also important to confirm dissipation test results. Cyclic testing can
also be used to assess triggering potential with undisturbed or specially reconstituted
samples.

USSD will insert footer text here 7


Static and Dynamic Pore Pressures

Pore pressures obtained from the CPTu pore pressure dissipation tests are compared to
the dynamic pore pressures recorded during cone advancement to evaluate potentially
contractive or dilative material behavior as follows:
 Materials that show dynamic pore pressures greater than the static pore pressures
indicate potentially contractive behavior during shearing.
 Materials that show dynamic pore pressures less than static pore pressures
indicate potentially dilative behavior during shearing.
Typical dynamic and static pore pressures for dilative and contractive behavior are shown
on Figures 4 and 5.

Figure 4. Typical Dilative Material Behavior.

8 Dams and Extreme Events


Figure 5. Typical Contractive Material Behavior.

Normalized Pore Pressure Parameter

The pore pressure difference value (P-value) is calculated by subtracting the static from
the dynamic pore pressures (udyn-ustatic) and normalizing the difference to the vertical
effective stress, reflecting the effects of depth (P = u2/’vo). If the P-value is negative,
the dynamic pore pressures are less than the static pore pressures and the material shows
potentially dilative behavior during shearing, as shown on Figure 4. If the P-value is
positive, the dynamic pore pressures are greater than the static pore pressures and the
material shows potentially contractive behavior during shearing, as shown on Figure 5.

Normalized Material Properties and State Parameter

Normalized CPTu soil behavior type charts (SBTn chart) are created for selected CPTu
soundings following the Robertson (2010) method to evaluate material behavior. The
normalized CPTu data are plotted as normalized tip resistance (Qtn) versus normalized
friction ratio (Fr).

Been and Jefferies (1985) used critical state soil mechanics to develop the state parameter
(Ψ) concept and applied these concepts to CPTu results (Been et al. 18986). The
approximate boundary between dilative and contractive material behavior is provided by
the state parameter Ψ = -0.05 line plotted on the SBTn charts. The state parameter, Ψ, is
the difference between the in situ void ratio, e0 and the void ratio at critical state, ecs, at
the same mean effective stress, p’ (Been and Jefferies, 1985). Jefferies and Been (2006)

USSD will insert footer text here 9


provided a detailed description of the evaluation of the soil state using the CPTu. In a
general sense, soils denser than the critical state (Ψ < 0) will be dilative and will strain
harden in undrained shear, whereas soils looser than the critical state (Ψ > 0) will be
contractive and will strain soften in undrained shear. Jefferies and Been (2006) and
Shuttle and Cunning (2007) suggested that when a soil has a state parameter Ψ > -0.05,
strain softening and strength loss in undrained shear can be expected. Therefore, defining
a region on the CPTu SBTn chart that represents a state parameter of about -0.05 is
helpful as a screening technique to identify the susceptibility for flow liquefaction.

The SBTn chart and data from liquefaction case histories (Robertson, 2010) are shown on
Figure 6. The data show that most liquefaction failures have CPTu data that plots below
the Qtn,cs = 70 line. The plot includes one mine tailings case history No. 35. The Qtn,cs = 70
line is similar to the state parameter Ψ = -0.05 line. The state parameter line for Ψ = 0.00
and Ψ = -0.10 were also shown for this study to evaluate the sensitivity of the state
parameter, as shown on Figure 7. The state parameter for the non-plastic tailings material
was evaluated from current and historic laboratory testing and was found to be consistent
with the state parameter of Ψ = -0.05, as suggested by Robertson (2010). The CPTu
soundings are normalized based on the phreatic surface at the time of sounding
advancement.

Figure 6. Liquefaction Failure Case Histories (from Robertson 2010).

Generally, zones of dilative tailings plot above the state parameter Ψ = -0.05 line of the
SBTn chart and contractive tailings plot below the state parameter Ψ = -0.05 line, as
shown on Figure 7. Materials that plot in zones on 2 and 3 of the SBTn chart generally
indicate plastic or clay-like behavior and may not be susceptible to cyclic liquefaction.

10 Dams and Extreme Events


Figure 7. Normalized Cone Resistance versus Normalized Friction Ratio for Typical
Dilative Behavior (left graph) and Typical Contractive Behavior (right graph).

State Parameter Difference

The state parameter difference plot combines the evaluation of the normalized pore
pressure difference (P-value) and the normalized CPTu data soil behavior plots to provide
an additional tool to help characterize the material as potentially contractive or dilative.

Negative P-values indicate potentially dilative behavior and positive values indicate
potentially contractive behavior, as described above, and plotted on the y-axis of the state
parameter difference plot Figure 8. The perpendicular distance from the state parameter
Ψ = -0.05 line for each data point was calculated and plotted on the x-axis on this plot
(Figure 8). Recognizing that data plotting above the state parameter Ψ = -0.05 line on the
normalized CPTu soil behavior plot indicate dilative behavior; the horizontal distance
calculated to a point above this line was given a negative value to remain consistent with
the convention that negative values indicate dilative behavior. Therefore, a point that
plots below the state parameter Ψ = -0.05 line on the normalized CPTu soil behavior plot
was positive, indicating contractive behavior.

The state parameter difference plot was broken into four quadrants (Figure 8). Material
that showed dilative behavior on both the normalized pore pressure difference plot and
normalized CPTu soil behavior plot can be found in quadrant three, as shown on Figure
8, left graph. Material that showed potentially contractive behavior in both the
normalized pore pressure difference plot and on the normalized CPTu soil behavior plot
will plot in quadrant one, as shown on Figure 8, right graph.

USSD will insert footer text here 11


Figure 8. Normalized Dynamic Pore Pressure Difference versus State Parameter
Difference for Typical Dilative Behavior (left graph) and Typical Contractive Behavior
(right graph).

Results

Tailings that could either be classified as potentially dilative, indicating less susceptibility
to liquefaction, or potentially contractive with higher susceptibility to liquefaction, are
characterized using the results from all four CPTu-based state characterization methods
to evaluate the state of saturated non-plastic tailings. If all these methods point
consistently to contractive or dilative behavior, then a strong case can be made for this
behavior under seismic loading and confidence in the characterization is increased. If the
results are mixed, then a less definitive, more conservative characterization is warranted.

In general for this case, evaluated CPTu soundings at depth located within the interior of
the impoundment show dilative material behavior, transitioning to contractive behavior
just downslope of the crest (near CPT-C), and show contractive behavior towards the
downstream toe, as shown on Figure 9. In some CPTu soundings, contractive material
behavior was found at shallower depths and a dilative material behavior was found with
depth i.e. CPT-C. Static and cyclic laboratory testing confirmed this characterization.

Figure 9. Dilative and Contractive Material Boundary for Study Section.

12 Dams and Extreme Events


LIQUEFACTION CASE HISTORY

A static liquefaction case history has also been evaluated using the CPTu-based state
characterization. The red zones identified in Figures 10 and 11 correspond to materials
located within the liquefied zone.

Figure 10. CPTu Data from Case History.

USSD will insert footer text here 13


Figure 11. Normalized Cone Resistance versus Normalized Friction Ration (left graph)
and Normalized Dynamic Pore Pressure Difference versus State Parameter from Case
History (right graph).

The CPTu-based state characterization methods clearly indicate the CPTu data as
contractive and potentially liquefiable. This contractive characterization fits well with the
observed static liquefaction and other more conventional triggering methods.

CONCLUSION

The CPTu-based state characterization provides an independent “first principles”


approach to assess tailings material behavior under dynamic and static loading
conditions. The characterization provides an assessment of whether the materials would
behave in a dilative or contractive manner during loading. Materials that exhibit dilative
behavior are typically denser, have a lower dynamic pore pressure response, and are less
susceptible to liquefaction. Materials that show contractive behavior are typically looser,
have higher dynamic pore pressure response, and are more susceptible to liquefaction.
Findings from the static liquefaction case history support this characterization method.

For shallower tailings profiles, the conventional triggering method and the new CPTu-
based state characterization method are in general agreement. Both methods indicate
tailings susceptible to liquefaction under the design earthquake event for CPTu soundings
located near the downstream toe. However, for tailings profiles with greater depth, the
CPTu-based state characterization provides a continuous record throughout the profile,
and may show dilative material behavior indicating less susceptibility to liquefaction.
This comprehensive characterization with multiple methods pointing to a conclusion
provides a more reasonable extent of potential liquefaction within a tailings cross section.

One limitation of the method is that high quality CPTu data must be obtained, such as
maintaining full saturation of the CPTu pore pressure sensor during penetration.

14 Dams and Extreme Events


Secondly, sufficient time for dissipation tests to fully reach equilibrium is needed and a
number dissipation tests are required with depth to obtain an equilibrium pore pressure
profile. Piezometer measurements and long term trends from piezometers are also useful
when evaluating the material behavior with time. Projects should also include correlation
borings with other in situ testing, undisturbed sampling (e.g. Shelby tubes or piston tube
samples), and laboratory testing (index and engineering properties) to assist in the
interpretation of the CPTu results. Numerical modeling is also important to evaluate and
interpret the behavior of the tailings dam under seismic loading. Data can vary across the
site and with depth; therefore, it is important to focus on characteristic signatures of
depositional regimes and the preponderance of what the data are revealing.

ACKNOWLEDGEMENTS

The guidance of Emeritus Professors Norbert Morgenstern, I. M. Idriss, James K.


Mitchell, and Dr. Yoshi Moriwaki is gratefully acknowledged. The authors also
appreciate the support and laboratory testing performed by Professor Ken Stokoe and his
team at the University of Texas at Austin.

REFERENCES

Been, K. and Jefferies, M. G. (1985). A State Parameter for Sands. Geotechnique, 35(2),
99-112.

Been, K., Crooks, J. H. A., Becker, D. E., and Jefferies, M. G. (1986). The Cone
Penetration Test in Sands: Part 1. State Parameter Interpretation. Geotechnique, 36(2),
239-249.

Darendeli, Mehmet Baris (2001). Development of a new Family of Normalized Modulus


Reduction and Material Damping Curves. Doctor of Philosophy Dissertation. The
University of Texas at Austin. August 2001.

Davidson, R., Pilz, J., and Brown, B. (2011). The Challenges of Building Tailings Dams
in Seismic Regions ANCOLD 2011, October, Melbourne

Hudson, M., Idriss, I. M., and Beikae, M. (1994). User’s Manual for QUAD4M: A
Computer Program to Evaluate the Seismic Response of Soil Structures Using Finite
Element Procedures and Incorporating A Compliant Base. Center for Geotechnical
Modeling, Dept. of Civil & Environmental Engineering, University of California, Davis,
May.

Idriss, I. M, and Boulanger, R. W. (2008). Soil Liquefaction During Earthquakes.


Earthquake Engineering Research Institute. Monograph MNO-12, Oakland, CA.

Jefferies, M. G. and Been, K. (2006). Soil Liquefaction – A critical State Approach.


Taylor and Francis, London.

USSD will insert footer text here 15


Seed, H. B., and Idriss, I. M., (1982). Ground Motions and Soil Liquefaction During
Earthquakes. Earthquake Engineering Research Institute, Berkeley, California. Vol. 5 of
a series titled: Engineering Monographs on Earthquake Criteria, Structural Design, and
Strong Motion Records.

Shuttle, D. A. and Cunning, J. (2007). Liquefaction Potential of Silts from CPTu.


Canadian Geotechnical Journal, 44, 1-19.

Robertson, P. K. (2010). Evaluation of Flow Liquefaction and Liquefied Strength using


the CPT, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, June 2010.

16 Dams and Extreme Events

You might also like