Professional Documents
Culture Documents
Guy Lund1
Brad Dawson2
Mark Foster3
ABSTRACT
Hume Dam is located near Albury/Wodonga, Australia and was constructed between
1919 and 1936. The reservoir was enlarged in the 1960s to its current capacity of
3,038,000 megaliters (2.5 106 acre-feet). It is the main regulating reservoir on the
River Murray System and supplies irrigation water and hydro-electric power. State
Water Corporation (State Water) on behalf of the Murray Darling Basin Authority
(MDBA) currently manages the dam and reservoir.
The main dam consists of an embankment dam with a concrete core wall and a gated
concrete gravity spillway. Spillway discharges flow through the gates, over the ogee
gravity section, and into the river through a discharge channel. The flow is trained with
large concrete training wall on both the right and left side of the discharge channel. The
left, southern training wall (STW) is located between the spillway channel and the main
embankment, and retains the embankment fill as well as containing the spillway
discharges. The height of the STW varies from approximately 50 meters (165 feet) near
the crest of the embankment dam to 18 meters (60 feet) at the downstream end, and is the
subject of this paper.
Modifications have been performed on the STW over the last few decades to improve
stability due to the increased loads caused by severe deformation of the embankment.
The modifications have included installation of sub-vertical post-tensioned tendons and
horizontal post-tensioned anchors. However, continued embankment deformation has
resulted in the need for additional rehabilitation. In addition, it is understood that the
critical loading condition is due to the safety evaluation earthquake (SEE), and a
significant portion of the load is dependent on the combined behavior of the embankment
fill and the mass concrete wall.
The finite element method of analysis was used to analyze the soil/structure interaction
and the behavior of the STW for both static and dynamic loads. This paper summarizes
the finite element model, parameter assumptions, and sensitivity studies used to verify the
behavior of the model with the actual STW, and the results used to develop the design
modification.
Hume Dam
1531
PROJECT DESCRIPTION
The main dam consists of an embankment dam with a concrete gravity spillway. The
spillway consists of a gated ogee section, flip-bucket type stilling basin, and mass
concrete, gravity training walls. The north (right) and south (left) spillway training walls
extend downstream from the axis of the dam past the concrete apron to retain the
embankment fill and train spillway discharges. The southern training wall (STW) is
located on the left side of the spillway looking downstream and is photographed in Figure
1.
Figure 1. Photograph of the Southern Training Wall and Hume Dam, Australia.
Modifications to improve stability were performed in the 1980s and included installation
of sub-vertical post-tensioned tendons through the full height of the STW and into the
foundation rock. These tendons are spaced at approximately 2 meter (6.5 feet) centers
and were designed to have a working load (equal to 65 percent of the minimum breaking
load) of 2,900 kN (12.9 kips).
Additional modifications were performed to the STW between 1995 and 2004 to
accommodate the increased load due to a rockfill berm constructed on the downstream
slope of the embankment dam. These modifications included installation of large
diameter, horizontal anchors that extend from the upper part of the STW to a concrete
deadman wall located approximately 80 meters (262 feet) within the embankment/berm.
A mechanically stabilized earth (MSE) wall was installed to retain the rockfill berm
along the crest of the STW.
1532
Hume Dam
1533
Structural Capacity. The results from the analysis were used to evaluate the potential
for overstressing of the concrete (i.e., crushing or cracking), and the moment capacity
of the STW with the sub-vertical post-tension anchors.
1534
Overturning. The results from the finite element model were used to assure that
moment equilibrium, or rotational stability, was maintained throughout the structure.
Sliding Stability. The results from the finite element model were used to evaluate the
potential for sliding failure along different planes through the structure.
Structural Capacity
The structural capacity of the STW was evaluated based on the allowable strength of the
mass concrete, and the bending moment capacity at select horizontal sections within the
STW.
Concrete Strength
The overstressing evaluation compares the computed stress from the finite element
analysis to the estimated strength of the mass concrete. If the computed stress is greater
than the allowable strength, then the concrete is expected to crack (tension) or crush
(compression). If cracking or crushing of the concrete develops, then there is an
increased potential for deformation (movement). On-the-other-hand, without cracking or
crushing of the concrete then movement of a rigid body (STW) is very unlikely.
Bending Moment Demand Capacity Ratio (DCR)
The sub-vertical post-tensioned anchors in the STW can be thought of as a type of
reinforcement. The capacity of a reinforced section can be evaluated based on the axialflexural interaction diagram.
The U.S. Army Corps of Engineers use the demand capacity ratio (DCR) method to
evaluate the ability of structures to support the dynamic earthquake loads [U.S. Army
Corps of Engineers, EM 1110-2-6051, Engineering and Design Time History
Dynamic Analysis of Concrete Hydraulic Structures, December 2003.]. The method
determines if the behavior of the reinforced concrete structure is linear, or non-linear. If
the behavior is linear, then it can be assumed that when the load is removed the structure
will return to the initial state. For reinforced concrete hydraulic structures, the maximum
bending moment DCR is taken as 2.0, and the shear DCR is taken as 1.0 [USACE, EM
1110-2-6051].
The flexural capacity of the STW at Hume
Dam was determined using an interaction
diagram, as shown on Figure 4. The
relationship of axial vs. flexural load was
computed assuming that the maximum
compressive strain in the concrete is
0.003, and the maximum tensile strain in
the tendons is the yield strain (i.e., yield
strength of the tendons was assumed to be
65 percent of the ultimate strength, or
1210 MPa).
Hume Dam
For dynamic loads, the computed moment should not exceed the moment capacity for
prolonged periods of time. The DCR
is allowed to exceed unity (1.0) for
short durations during the earthquake,
because studies have shown that short
term yielding of the steel will not result
in a brittle failure. The assumption is
considered valid for the STW at Hume
Dam, because the tendons are nonbonded, such that any potential
elongation will be applied over the full
length of the tendon, and not an
isolated point such as with reinforced
Figure 5. DCR Acceptance curve.
concrete.
The U.S. Army Corps of Engineers use the demand capacity ratio (DCR) method
evaluates the total duration of time that the DCR greater than 1.0. The time duration is
plotted on the acceptance curve, shown in Figure 5. If duration of time is greater than
1.0, but within the limits of the acceptance curve (see Figure 5), then the structure is
expected to behave linearly and further evaluation is not necessary. If the time duration
falls outside the limits of the acceptance curve, then the structure will behave nonlinearly, and further studies may need to be performed to correctly evaluate the behavior
of the structural system.
Overturning
The rotational stability (moment equilibrium) is satisfied if the summation of moments
from the analysis equal zero. The finite element model for the STW at Hume Dam used
non-linear contact elements to simulate the condition at the base of the STW. The
contact element can develop compression, but will separate rather than developing
tension. Therefore, if the analysis shows that the separation of the contact elements
stabilized (i.e. there is no further propagation of the crack), then moment equilibrium has
been satisfied.
Sliding Stability
The minimum factors of safety required to satisfy sliding stability for the usual, unusual,
and extreme load combinations are based on criteria published by the USACE [U.S. Army
Corps of Engineers, EM 1110-2-2100, Engineering and Design Stability Analysis of
Concrete Structures, December 2005]. The minimum factor of safety for the postseismic load combination is based on the FERC [Federal Energy Regulatory Commission
(FERC), Engineering Guidelines for the Evaluation of Hydropower Projects, Chapter 3
Gravity Dam, 2002.]. If the computed sliding factor of safety is greater than the
minimum value set in the criteria, then sliding instability if very unlikely.
1536
METHOD OF ANALYSIS
The studies for the Hume Dam STW were performed using a two-dimensional plane
strain, finite element model using the computer program ANSYS. The geometry of the
STW section at OS 31 is shown on Figure 6.
Hume Dam
1537
The behavior of the concrete, rock and soil were simulated using different material
models. The concrete and foundation rock were evaluated using linear elastic material
behavior. The weathered rock, alluvium and embankment were evaluated using the
Extended Drucker-Prager (EDP) material model, which is an expanded use of the MohrCoulomb failure criteria and is used for the evaluation of granular materials, such as rock
and soil. The parameters used to define the behavior of the EDP materials are based on
the internal friction angle and cohesive strength of the soil.
Dynamic Material Properties
The dynamic behavior of the STW was evaluated using a full transient analysis. Some of
the most critical material properties used in this type of studies typically includes the
stiffness, Poissons ratio, and damping. The stiffness and Poissons ratio for the concrete
material was not modified for the dynamic studies, primarily because the ratio of concrete
stiffness to embankment stiffness was so large.
The stiffness and Poissons ratio for the embankment materials were modified for the
dynamic evaluations. Typically, the Poissons ratio for soils under static loads will range
between 0.3 and 0.4 [Lambe, T. William, Robert V. Whitman, Soil Mechanics,
Massachusetts Institute of Technology, John Wiley & Sons, Inc. New York, 1969.]. The
assumed Poisson ratio for the embankment materials varied depending on moisture
content and loading conditions. The Poissons ratio for dry embankment soil was
assumed equal to 0.30 for both static and dynamic loads. The Poissons ratio for
saturated soils was assumed equal to 0.30 for static loading conditions and 0.45 for
dynamic loading condition. The higher ratio better simulated the relationship between
elastic and shear modulus during the seismic loads, due to increased pore pressures.
When a soil liquefies the Poissons ratio will approach 0.50; however, mathematical
limitations within the finite element code will not solve for a Poisson ratio of 0.50.
Therefore, for the liquefied saturated soils the Poissons ratio was set equal to 0.49.
The elastic property of the soil was simulated using various values for Youngs modulus.
Initial studies assumed elastic values of Youngs modulus for the alluvium, backfill, and
rockfill materials based on test data [URS, Hume Dam Remedial Works Geotechnical
Investigation, Volumes I and II, Report prepared for State Water, dated June 22, 2009
The stiffness of the materials was then computed based on the relationship with Poissons
ratio and shear modulus.
The material damping constants used in the analysis conservatively assumed 5 percent
damping for the concrete, 5 percent in the foundation rock, and 10 percent in the
weathered rock and embankment materials.
Sensitivity Studies
Several sensitivity studies were performed to verify properties for the embankment
materials in the finite element model. The sensitivity studies evaluated the effective
horizontal pressure coefficient of the embankment, and vertical settlements of the
1538
embankment. The results from the finite element analysis were compared to previous
studies using Slope/W and field investigation data, as shown Table 1.
Table 1. Horizontal Pressure Coefficients
Model Description
Effective Horizontal
Pressure Coefficient
0.61
ANSYS Model
0.62
Hume Dam
1539
structure. It also provides a basis for comparison of the behavior between the total usual,
extreme, and post-earthquake loading conditions.
Figure 8. Vertical stress contours for Load Step 2, original design load.
1540
The finite element model representation of the load on the STW is relatively accurate,
and the concrete strength is greater than assumed.
The structural capacity of the wall is greater than assumed in the finite element
model.
The material parameters used to simulate the behavior of the embankment is overly
conservative in the finite element analysis.
After review and discussion of the studies, it was concluded that both the twodimensional finite element model and soil parameters used to simulate the behavior of the
embankment are likely overly conservative. However, if the studies were to conclude
that the STW has adequate capacity for all assumed loading conditions, then the
conservative model would be acceptable and further refinement of the modeling
assumptions would not be necessary. At the end of the studies, the stability against
overturning and sliding for load step 2 was considered to be stable based on the historical
performance.
Load Step 5 Current Load Condition
The evaluation for load step 5 included the effects of gravity, embankment pressures,
hydrostatic pressures, post-tension anchors forces, and the added load due to the rockfill
berm. This is considered the current condition of the STW. The vertical stress results in
the STW are shown in Figure 9.
Hume Dam
1541
Figure 9. Vertical stress contours for Load Step 5, current loading condition.
As expected, that additional load due to the rockfill berm results in an increase in tensile
stress on the embankment face of the STW. As previously discussed, the installed subvertical post-tension tendons add moment capacity to the STW. The actual (demand)
moment on the STW was compared to the moment capacity, as computed using the
moment-interaction diagram shown on Figure 4. The results for the critical section of the
STW are shown in Figure 10. The demand capacity ratio (DCR) for concrete is defined
as the ratio of moment on the structure (demand), divided by the flexural capacity of the
post-tensioned section, and is equal to 0.72. Based on the results, the potential for failure
due to overstressing is very unlikely due to load step 5.
1542
The potential for instability due to overturning and sliding for load step 5 was considered
very unlikely, due to the historical performance for the STW.
Load Step 7 Final Design Load Combination (concrete buttress with decommission
horizontal anchors)
The evaluation for load step 7 includes the effect of gravity, embankment pressures,
hydrostatic pressures, post-tension anchors forces, rockfill berm, concrete buttress, and
proposed decommissioning of the horizontal post-tension anchor. The vertical stress
results are shown in Figure 11.
Figure 11. Vertical stress contours for Load Step 7, final loading condition
The results show that there is an increase in crest displacement towards the spillway of
5.5 millimeters due to the release of the horizontal anchor (see Figure 7).
The decommissioning of the horizontal anchor increases the load on the concrete
buttress, as shown by the compressive stress contours in Figure 11. The results from the
study showed that the concrete buttress will resists the movement of the STW as it
rebounds towards the spillway after the decommissioning of the horizontal anchor. The
stresses in the buttress are compressive, and below the allowable compressive limit of the
concrete, and the demand moment is less than the capacity as shown in Figure 12.
Hume Dam
1543
1544
The results from the 2,500-year seismic event indicate that the maximum amplitude
of crest deflection will be approximately 7 millimeters, and there will be
approximately 2.0 millimeters of permanent deformation towards the spillway.
The results from the finite element analysis show that the Coalinga earthquake event was
the most severe seismic load and that the most critical time was 11.58 seconds into the
earthquake. The vertical stress contours at time 11.58 seconds are shown on Figure 14.
The results show an isolated area of tensile stress develops on the embankment face of
the STW. The maximum stress is less than the assumed tensile strength of the concrete,
which suggests that the structure has adequate capacity for this assumed load. However,
if the concrete in the STW does not have any tensile strength (e.g. lift joints) then the
section would crack and the load would be transferred to sub-vertical post-tensioned
anchors.
Figure 14. Vertical stress results for the Coalinga seismic analysis.
The results from the analysis were also evaluated for the bending moment DCR. The
DCR results during the Coalinga earthquake event for selected elevations in the STW are
shown on Figure 15.
The critical section regarding moment demand versus capacity is at EL. 150, which
corresponds to the top of the spillway slab. The spillway slab effectively fixes the base
of the STW against significant rotation, which is why the maximum moment develops at
that elevation. The computed DCR is less than 1.0 for all of the sections of the STW
during the 10,000-year earthquake events. Based on these results, the STW will be
expected to behave as a linear system, and instability due to overstressing is very
unlikely.
Hume Dam
1545
The finite element method was successfully used to simulate the horizontal
embankment loads on the STW structure using the extended Druker-Prager material
model.
The behavior of the STW during the seismic loading condition was evaluated using
demand-capacity ratio, as defined by the USACE methodology. The results indicate
that the structural behavior will remain linear and the potential for development of a
1546
seepage path between the STW and embankment material due to instability is very
unlikely.
There was confidence in the results from the finite element model due to the
sensitivity studies, which were able to verify the computed displacement from the
model with the measured field deformations.
ACKNOWLEDGEMENTS
URS wishes to acknowledge State Water for the opportunity to work on this project.
URS also wished to acknowledge the participation of the Hume Dam Expert Review
Panel (ERP) for their expertise and reviewing of the study.
REFERENCES
1.
2.
FERC Guidelines. Federal Energy Regulatory Commission, Office of
Hydropower Licensing, "Engineering Guidelines for the Evaluation of Hydropower
Projects", Washington, D.C., October 1999, Chapter 11-Arch Dams.
3.
U.S. Army Corps of Engineers, EM 1110-2-2100, Engineering and Design
Stability Analysis of Concrete Structures, December 2005.
4.
Lambe, T. William, Robert V. Whitman, Soil Mechanics, Massachusetts Institute
of Technology, John Wiley & Sons, Inc. New York, 1969.
5.
6.
U.S. Army Corps of Engineers, EM 1110-2-6051, Engineering and Design
Time History Dynamic Analysis of Concrete Hydraulic Structures, December 2003.
7.
Federal Energy Regulatory Commission (FERC), Engineering Guidelines for the
Evaluation of Hydropower Projects, Chapter 3 Gravity Dam, 2002.
8.
Author: Sheldon Imaoka, ANSYS Revision 11.0, Memo Number STI:08/02,
March 15, 2008, Subject Sheldons ANSYS.NET Tips and Tricks: Drucker-Prager
Model.
Hume Dam
1547