Professional Documents
Culture Documents
CYCLIC TESTING
ABSTRACT
A raise and spillway realignment are proposed for Scoggins Dam, an existing
embankment dam located in northwestern Oregon. An extensive geotechnical laboratory
investigation was performed to evaluate the foundation material properties during static
and seismic loading. Specimens were tested from four borings along the proposed
spillway alignment, and two borings at the existing right abutment. The borings
encountered weak sandstone, considered to be a transition geomaterial, which
necessitated the development of a novel testing approach to determine the modulus
degradation and damping ratio for numerical modeling. The paper details the
development and results of the approach for determining weak rock cyclic strength
parameters.
INTRODUCTION
Scoggins Dam is located about 30 miles west of Portland, Oregon and impounds Henry
Hagg Reservoir. Dam construction started in 1972 and was completed in 1975. The dam
is a zoned earthfill embankment with the spillway located near the left abutment of the
dam. The spillway and dam are founded on quaternary alluvial deposits overlying the
Spencer Formation comprising primarily sandstone with occasional siltstone and
claystone interbeds.
This study was performed in support of the Tualatin Basin Dam Safety and Water Supply
Joint Project, a joint project between Clean Water Services and Reclamation. The
proposed dam raise would include modifying the current spillway alignment and entrance
channel, and possibly moving a bridge located on SW West Shore Drive in an upstream
reach of the reservoir. Funding for the borings and testing described in this paper was
provided by Clean Water Services.
Four borings were advanced along the proposed spillway alignment to investigate the
foundation conditions. The borings encountered overburden soils underlain by weak
sandstone bedrock. While the rock encountered had a relatively high Rock Quality
Designation (RQD) the sandstone was soft and degraded rapidly upon exposure to air.
Selected sandstone specimens were preserved with wax and sent back to Reclamation’s
Concrete, Geotechnical, and Structural Laboratory (CGSL) at the Technical Service
1
Bureau of Reclamation, Civil (Geotechnical) Engineer, Technical Service Center, Denver, Colorado,
elindenbach@usbr.gov
2
Bureau of Reclamation, Civil (Geotechnical) Engineer, Technical Service Center, Denver, Colorado,
rbearce@usbr.gov
While this testing aided in understanding the static engineering properties of the bedrock,
there was interest in investigating the dynamic characteristics for use in finite element
modeling of the structure when subjected to earthquake loading. A novel approach was
developed to evaluate the mechanical behavior under cyclic loading. As the static UCS
was needed to develop input parameters for the cyclic analysis, those results are briefly
reviewed as well.
PREVIOUS WORK
Weak rock, such as shale, claystone, etc., is encountered at many construction sites and
underlies many existing structures. In this paper, weak rock is assumed to represent the
transition between cohesive soil and hard rock, with this referring to both inherently weak
and weathered materials. As indicated by Santi and Doyle (1997), weak rocks are
unweathered to weathered rocks, with low compressive strength or are intensely
fractured. Weak rock can be difficult to sample and test in a laboratory, as typical
techniques for soil or hard rock may not be applicable (Nickmann, 2006). This difficulty
results in special techniques being employed to determine the engineering properties of
interest.
A number of authors have investigated the behavior of hard rock when subjected to cyclic
loading. A brief literature review is given here to provide context for this paper.
Haimson (1973) looked at the strength degradation of rock when subjected to cyclic
uniaxial and triaxial, compressive and tensile loading. The testing focused on developing
a relationship for the number of cycles to failure at various stress states by using load-
controlled testing. It was found that fatigue strengths of the rock were significantly lower
than the intact strength, with the recommendation that a fatigue strength be considered in
the design of underground structures. The author additionally noted that cyclic loading
typically results in a decrease in the Young’s modulus and an increase in Poisson’s ratio.
Bagde and Petroš (2005) investigated the fatigue behavior of sandstone subjected to
cyclic uniaxial compression with varying loading frequencies. It was observed that both
the frequency and amplitude of loading influenced the dynamic behavior of the sandstone
specimens. Higher loading frequencies resulted in a reduction of the fatigue strength and
the dynamic axial stiffness, while the dynamic modulus increased with loading frequency
and decreased with loading amplitude. In this instance, dynamic modulus is defined over
a specific range of the overall stress-strain curve, which includes several dynamic cycles.
The dynamic axial stiffness is the slope of the stress-strain curve found in the
corresponding single-cycle peak-valley data generated during dynamic loading. The
reader is referred to this paper for a detailed literature review of the cyclic behavior of
rock.
Vaneghi et al. (2018) studied the changes in mechanical properties of sandstone and
granodiorite when subjected to cyclic uniaxial compressive loading. The goal of the
research was to understand the effects of loading amplitude and stress on the engineering
properties, and to investigate the difference in response of soft rock and hard rock when
subjected to cyclic loading. The research found that the fatigue life decreased with
increased maximum stress level, and decreased with increasing loading amplitude, with
the loading amplitude having a larger effect. Stress level refers to the mean stress of the
load-controlled cyclic testing during a specific cyclic stage. Of particular note is that the
response of soft rock was found to be different than that of hard rock. The authors noted
that the fatigue strength of hard/brittle rocks is less than that of soft/ductile rocks, i.e.
softer rock undergoes less strength reduction during cyclic loading.
There remains a significant gap in the understanding of how weak rocks will behave
under cyclic loading; specifically, how to develop modulus reduction and damping ratio
curves for weak rocks to be used in a finite element analysis of a structure. This research
seeks to bridge the gap between cyclic soil and rock testing to better characterize
foundation materials subjected to seismic loading for use in numerical modeling.
Static UCS tests were performed on eight sandstone specimens obtained from the
proposed spillway alignment borings. All specimens tested were wrapped in the field to
preserve moisture and tested per ASTM D7012 (ASTM, 2014). The specimens were cut
in the dry using a diamond-bladed circular saw. The specimens were too fragile for
surface grinding; the ends of each specimen were carefully smoothed with fine-grit
sandpaper and not checked for dimensional tolerances. All specimens were tested with a
spherical platen.
Specimens were tested on a MTS 815 servo-hydraulic stiff testing machine which was
controlled by lateral-circumferential strain as measured during testing. Clip-on gauges
measured axial and lateral strain during the tests while an internal load cell measured the
applied load. MTS TestSuite software was used to control the test.
Table 1 summarizes the results from the static uniaxial testing program and provides soil-
type physical properties for context. Soil physical properties tests were performed on top
Tangent 20-50%
SP-17-1 54.1 SM 1,085 47 23 1.18E+05 1.46E+05 1.45E+05 0.29 0.23 0.24 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-1 75.7 SM 1,075 58 19 1.23E+05 1.45E+05 9.41E+04 0.31 0.45 0.24 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-2 44.3 SM 380 23 17 2.79E+04 2.81E+04 2.61E+04 0.45 0.45 0.45 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-2 80.9 SM 470 41 12 3.00E+04 3.07E+04 2.72E+04 0.38 0.43 0.32 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-3 56.7 s(CL) 933 32 29 7.40E+04 7.71E+04 6.25E+04 0.14 0.20 0.12 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-3 74.5 SM 2,034 42 49 1.83E+05 1.99E+05 1.31E+05 0.22 0.37 0.17 Average 40-70%
Secant 0-50%
Tangent 20-50%
SP-17-4 44.6 SM 420 24 17 2.20E+05 5.54E+04 3.41E+04 0.27 0.48 0.25 Average 40-70%
Secant 0-50%
The Tangent, Average and Secant Young’s Modulus is defined by linearizing the stress
strain curve between 20% and 50%, 40% and 70%, and 0% to 50% of the stress at failure,
respectively. Test results indicate that the strength generally increased with depth in each
boring but was found to be quite variable along the alignment. In SP-17-1 and SP-17-2, a
consistent decrease in the ratio between UCS and the approximate in-situ vertical stress
(σ v ) with depth can be observed, while this trend was not noted in the SP-17-3.
The samples all appeared to be of good quality based on a visual inspection, with
variability in test results likely being a function of a heterogeneous rock mass.
Unfortunately, it is difficult to assess the sample quality using either soil mechanics
(consolidation results) or rock mechanics (petrographic analysis or interpretation of the
stress-strain curve) as the material tested is a ductile transition material. Future work will
attempt to incorporate consolidation results to assess sample quality. Figure 1 shows the
results from a typical UCS test.
The authors performed a literature review of previous cyclic testing of rock. It was noted
that while many authors had tested cohesive soils or relatively hard rock under cyclic
conditions, there was not a procedure for testing weak rock. A program was developed in
which the rock specimens were tested under a hybrid technique. The method used for the
testing is detailed in the following section. It should be noted that the anticipated
confining pressure based on overburden thickness is relatively small compared to the
uniaxial compressive strength of the rock; therefore, it was assumed that applying a
confining pressure would have minimal influence on the rock strength and that cyclic
UCS was representative of in-situ.
Cyclic UCS tests were performed under axial strain control at three different strain levels,
followed by post-cyclic monotonic compression to failure. The procedure followed was
adapted from ASTM D3999 for cyclic triaxial testing of soil (ASTM, 2011), where a
specimen is consolidated to a specific stress state and modulus and damping properties
are calculated during strain controlled cyclic loading with constant cell pressure. As the
rock specimens tested as part of this work were not subjected to a confining pressure, the
use of the term “initial stress” is used in lieu of “consolidation stress” to describe the
initial axial stress (σ i ) applied to the specimen prior to cyclic loading. All calculations
for modulus and damping ratio followed ASTM D3999, with all cyclic loading
performed at 1 Hz using a sinusoidal waveform.
The σ i for the specimens was selected to be one-half of the crack initiation stress of the
closest proximity static UCS specimen. The crack initiation stress is the stress at which
micro-cracks begin to form in a specimen and was found using the crack volumetric
strain method (Ghazvinian et al., in press). While it is not clear if the crack volumetric
strain method is applicable to the ductile rocks tested as part of this program, the selected
initial stresses were all about 30% of the uniaxial compressive strength of the static
specimens ensuring that the specimen will undergo damage during the cyclic portion of
the test. The ratio between initial stress to the approximate in-situ vertical stress ranged
from about 3 to 14. The authors initially attempted to use the crack initiation stress as σ i ,
but several specimens failed at this load likely due to variability between the monotonic
and cyclic specimens.
The cyclic uniaxial compressive strength test procedure developed for this project used
the following steps:
Note that the axial strain was recalculated for each stage based on the specimen length
after the application of the initial stress.
Cyclic testing was performed on the same equipment used for static testing. A MTS
TestSuite protocol was used to control the testing equipment, with data post-processed
using a MatLab routine.
Four specimens (one from each boring) were tested under strain-controlled cyclic loading
to determine the degraded elastic modulus and the damping ratio, along with post-cyclic
monotonic strength and deformation characteristics. As the first cycle data is often
variable due to machine compliance and gage hysteresis, the second cycle data was
selected as more representative of the actual behavior. The small-strain modulus (found
from ultrasonic pulse velocity testing) of the static UCS specimen used to determine σ i is
also provided as E 0 and is connected to the degraded modulus curve with a dashed line.
Small-strain modulus of the cyclic specimens were not determined directly. Note that the
post-cyclic monotonic testing started at σ i .
Table 2 presents the cyclic testing results, including modulus at the second cycle and
damping ratio at the second cycle for each strain level.
Specimen Description Stage 1 (0.05% strain) Stage 2 (0.1% strain) Stage 3 (0.2% strain)
Depth of Young's Damping Young's Damping Young's Damping
UCS σi σi/σv E0 Modulus Coefficient Modulus Coefficient Modulus Coefficient
Drill Hole Specimen (2 cycle) (2 cycle) (2 cycle) (2 cycle) (2 cycle) (2 cycle)
(ft) lbf/in2 - lbf/in2 lbf/in2 % lbf/in2 % lbf/in2 %
SP-17-1 62.4 319 6 1.05E+06 1.83E+05 0.08 1.66E+05 0.13 1.44E+05 0.18
SP-17-2 76.3 128 3 4.18E+05 8.93E+04 0.09 7.69E+04 0.14 6.55E+04 0.17
SP-17-3 64.6 520 14 4.40E+05 2.06E+05 0.05 1.88E+05 0.10 1.59E+05 0.16
SP-17-4 47.9 170 7 3.93E+05 6.89E+04 0.06 5.82E+04 0.10 4.59E+04 0.16
Mean values for Sandstone 1.37E+05 0.07 1.22E+05 0.12 1.04E+05 0.17
Median values for Sandstone 1.36E+05 0.07 1.21E+05 0.12 1.05E+05 0.17
Coefficient of Variation 50% 26% 53% 18% 54% 6%
In general, the Young’s modulus decreased during continued cyclic loading and with
increasing axial strain, while the damping ratio increased. The decrease in modulus and
increase in damping ratio was generally small, indicating that the damage caused by
applied strains and loading frequency was minimal. Table 3 provides the post-cyclic
monotonic test results.
Young’s modulus for the post-cyclic monotonic portion was calculated using the linear
portion of the stress strain curve immediately after loading was initiated. As with the
static specimens, cyclic and post-cyclic monotonic tests results were highly variable. The
post cyclic UCS of the specimens typically fell within the range of the static UCS
specimens detailed earlier.
Figure 2 shows stress and strain versus time for one stage during a cyclic test, while
Figure 3 presents the data from all three cycles stages of one test.
Figure 2 - Cyclic test data from SP-17-2 at 76.3 ft from the ± 0.05% axial strain
cycle. Note consistent machine control during cycling.
The cyclic loops shown in Figure 3 appear to show a strain-hardening effect, with the
loops tending to have a concave up shape. This shape becomes more pronounced with
increasing strain amplitude. The peak stress reached decreases with increasing cycles,
with each subsequent loop becoming flatter overall. The loop flattening is indicative of a
general reduction in modulus during cyclic testing, confirming the findings of previous
researchers. Of note is that the loops all tend to approach the same slope at the largest
positive strain (peak stress point) for both the loading and unloading curves. This result
indicates that under strain-controlled cyclic loading the peak stress for each loop and
overall modulus is decreasing with increasing cycles, but the specimen stiffness is
relatively unchanged at the peak stress point immediately prior to the strain reversal.
Figure 4 provides the normalized modulus versus strain for all the specimens tested.
The degradation in modulus with increasing strain is apparent with the specimens all
displaying a similar trend. Figure 5 shows the damping ratio versus strain for all
specimens tested.
Figure 5 – Damping ratio versus strain for the four specimens tested.
The damping ratio versus strain trends are nearly linear in all four cases. Of note is that
the damping ratio for all specimens tends to converge with increasing strain, with the
specimens all approaching 0.16 to 0.18% at the highest strains.
Figure 6 indicates that the specimen is still within the linear elastic range after cyclic
loading. The initial stress is approximately 21% of the stress at failure for this specimen.
The initial stress to stress at failure ratio varied from 21% to 94% for the specimens
tested.
CONCLUSIONS
A method to develop modulus degradation and damping ratio curves for weak rock
(transition geomaterials) was developed and used to test four specimens. The proposed
method first applies a σ i , with subsequent cyclic loading performed in strain control using
a sinusoidal wave form. The cyclic loading is performed in stages with each stage
representing a doubling of the strain amplitude used in the previous stage. Monotonic
loading is performed once the cyclic stages are completed.
The peak stress of each cyclic loop decreased with increasing cycles. The concave up
shape of the stress-strain cyclic loops indicate that the specimens underwent strain
stiffening up to the peak stress point, with the loops all appearing to approach the same
slope or modulus immediately before and after the strain reversal.
RECOMMENDATIONS
Future work is needed to constrain a number of uncertainties associated with this testing.
This work is intended to be a starting point for characterizing the mechanical properties
of weak rock when subjected to cyclic loading. Future work could include: using this
procedure with different strength rocks, varying the frequency, amplitude, and duration of
loading, incorporating ultrasonic pulse velocity measurements during testing, performing
the tests under confined conditions, varying the initial stress, and improving the current
equipment to apply a tensile load thereby generating a full stress reversal.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Randy Kuzniakowski, Bryan Simpson, Tara
Schenk-McFarland and Clean Water Services for their support in this project. Thank you
to Bobby Rinehart for peer review, and Ashley Johnson for help with specimen
preparation. An additional thank you to MTS for the customer support as the testing
procedure was developed.
REFERENCES
ASTM Standard D3999-11e1 (2011). “Standard Test Method for the Determination of
the Modulus and Damping Properties of Soils Using the Cyclic Triaxail Apparatus,”
ASTM International, West Conshohocken, PA.
ASTM Standard D7012-14 (2014).“Standard Test Method for Compressive Strength and
Elastic Moduli of Intact Rock Core Specimens Under Varying States of Stress and
Temperatures”, ASTM International, West Conshohocken, PA.
Bagde, M.N., and Petroš, V. (2005). “Fatigue properties of intact sandstone samples
subjected to dynamic uniaxial cyclic loading.” Internation Journal of Rock Mechanics
and Mining Sciences 42, pp. 237-250.
Haimson, B.C. (1973). Aspects of Mechanical Behavior of Rock Under Static and Cyclic
Loading, Part B: Mechanical Behavior of Rock Under Cyclic Loading. Prepared for
Advanced Research Project Agency, ARPA Order No. 1579, Amendment 3.
Liang, W., Zhang, C., Gao, H., Yang, X., Xu, S., and Zhao, Y. (2012). “Experiments on
mechanical properties of salt rocks under cyclic loading.” Journal of Rock Mechanics and
Geotechnical Engineering 4(1), pp. 54-61.
Vaneghi, R.G., Federosi, B., Okoth, A.D., and Kuek, B. (2018). “Strength degredation of
sandstone and granodiorite under uniaxial cyclic loading.” Journal of Rock Mechanics
and Geotechnical Engineering 10, pp. 117-126.