You are on page 1of 24

BEARABLE MAXIMUM SEISMIC ACTION FOR EXISTING

CONCRETE DAMS

M, Colombo1, M. Domaneschi2, A. Ghisi1, S. Griffini3


1
DICA Politecnico di Milano, Milan, Italy
2
DISEG, Politecnico di Torino, Turin, Italy
3
Studio Griffini S.r.l., Milan, Italy

SUMMARY: The new Italian standard requires the evaluation of the bearable maximum seismic
action for existing concrete dams, in addition to the safety assessment with respect to the expected
seismic actions at the site. Various available guidelines and research results are considered, to
show the intrinsic innovation of this peculiarity. Different critical conditions allow the estimation
of the bearable maximum seismic action at increasing earthquake intensity. In this work, the
bearable maximum seismic action is investigated through numerical, three-dimensional, finite
element simulations in Abaqus for two existing concrete dams, namely an arch-gravity and a
buttress-gravity dam. The bearable maximum seismic action is interpreted differently according
to the dam typology. With reference to the first case study, the failure mechanism that drives the
estimation is related to the overcoming of the limit stress conditions in wide parts of the bulk,
stiffness degradation in the dam global nonlinear response and displacement discontinuity in
structural joints. Instead, when the buttress-gravity dam is analyzed, sliding in the construction
and foundation joints represents the critical mechanism.

KEYWORDS: concrete dams; standards; finite elements; maximum earthquake; sliding; rock-
mass

1 Introduction
Performance and safety assessment of existing structures against natural hazards is identified as a
major objective in the civil engineering community, that is as part of a comprehensive decision-
making process about possible actions that can be undertaken after potentially damaging events
[Li et al., 2016; Nagarajaiah and Erazo, 2016; Mori and Spina, 2015]. More than half of the large
concrete dams are more than fifty years old and are subjected to repeating hazards, as earthquakes
and overflows, which can affect their strength and robustness.

Corresponding author: Martina Colombo, DICA, Politecnico di Milano, Italy.


Email: martina.colombo@polimi.it
To promptly identify any unusual behavior in existing dams, ambient vibration and laboratory tests
are performed in parallel with the effective tools coming from numerical techniques [Phansr et al.,
2010; Sevim et al., 2013; Türker and Bayraktar, 2014].
The seismic safety verifications usually required by standards for large concrete dams are
evaluated at prescribed seismic actions. In the past a deterministic approach was usually followed
[D.M. 24 marzo, 1982; Reservoir Act, 1975]. Instead, in the last years a semi-probabilistic concept
prevailed, see e.g. [Canadian Dam Association, 2013]. By following the last approach, risks are
not studied according to a rigorous stochastic process analysis, but they are implicitly treated by
safety coefficients. The next section is devoted to summarize seismic safety verifications
approaches from a selection of available International standards.

1.1 Review of International guidelines on concrete dams

Swiss guidelines [Darbre, 2004; Officé Fedéral des Eaux et de la Géologie, 2003] define seismic
risk as the combination of the occurrence probability of an earthquake and the consequences of
dam failure. The same risk level is typically accepted for all the dams: when the failure
consequences are very high for the downstream population, a lower occurrence probability is then
expected. However, there is not an indication on an earthquake level at the dam site leading to the
failure of the system itself.
In Japan guidelines [Shimamoto et al., 2007] the Maximum Scenario Earthquake (MSE) is fixed
as the largest earthquake motion at the site and the required seismic performance of the dam system
is defined. The basic dam seismic design in Japan, before the recent publication of guidelines in
2005, was mainly conducted by using a pseudo-static approach. Despite this simplification, it is
worth noting that no earthquake damages, severe enough to affect people lives and properties in
the lower areas, have been experienced up to now.
UK guidelines [Charles et al., 1991] point out that the evaluation of the seismic risk requires to
consider mainly three issues: i) the probability of occurrence of an earthquake larger than the MSE,
ii) the vulnerability of the dam, and iii) the consequences of the failure (e.g. release of reservoir
water). Moreover, in the same document an accurate list of historical seismic events in the world
at dam sites and their effects is presented: in several cases, dams were able to sustain seismic loads
larger than the design ones.
Reference from France [Comité francais des barrages et reservoirs, 2012], focuses on concrete
gravity dams. A stability verification with respect to sliding becomes largely preponderant with
respect to tensional limit states. The rationale for the seismic verification remains the same as
above.
US Army Corps of Engineers [USACE, 1994; 1995; 1999; 2003] has written a complex of
standards compatible with risk-based planning and design methods. Earthquakes loads comply
with the requirements of [USACE, 1994] and, consistently with standards previously presented,
MSE are defined for a subsequent seismic safety verification. The same approach is followed by
the Federal Energy Regulatory Commission (2007) and the Bureau of Reclamation (1977)
guidelines. In [Bureau of Reclamation, 1999] the problem of existing concrete dams is deepened
with special reference to geotechnical aspects. The capacity of the structure to resist to applied
loads is assessed by satisfying minimum global safety factors (lowest level approach).
As a conclusion from this literature survey, it appears that a scenario earthquake is usually adopted
to test the limit state of the dam system for subsequent safety verifications. The next section
introduces the evaluation of the bearable maximum seismic action for existing dams, as originally
required by the new Italian standard.

1.2 Bearable maximum seismic action

The Italian standard [D.M. 26 giugno, 2014] specifies, for existing large concrete dams, that the
maximum bearing capacity has to be estimated. It is an unusual and innovative requirement with
respect to the aforementioned guidelines, where the reference seismic motion was carefully
reconstructed according to the logic of the largest earthquake motion at the site. The matter here
becomes the estimation of the maximum bearable earthquake, as the unknown, for the dam. The
current Italian standard does not add any other prescription: at present, there is no further action or
requirement to be satisfied after such bearable limit estimation. Neither any indication is given for
the modelling. This silence can be interpreted as an implicit recognition both of the singularity of
each dam system and of the uncertainties in the problem. The evaluation of the seismic capacity
for the dam should allow estimating an (innovative) safety margin of the structure, moving the
problem from the seismic hazard characterization at the site to the structural fragility. Such
perspective, even though prescriptions on modelling and site investigations are not yet indicated,
could be also useful for the national authority for tracing the general condition of the existing dams,
built often more than fifty years ago, in times when earthquake engineering was less developed.
The assessment of the bearable maximum earthquake addresses a concept not far from the
framework of the potential failure mode analysis, see e.g. [Hariri-Ardebili & Saouma, 2016], and,
more in general, of the performance based seismic engineering, see e.g. [Bertero & Bertero, 2002].
An interesting recent contribution [Hariri-Ardebili et al., 2016] focuses both on linear and
nonlinear approaches for concrete dams. In particular, it discusses the more appropriate intensity
measure, i.e. the excitation input parameter, to be compared with an engineering demand
parameter or a damage state for the evaluation of the failure mode. Fragility analyses have been
also performed within probabilistic framework for concrete gravity dams in [Tekie & Ellingwood,
2003].
This work is focused on the maximum seismic bearable assessment for two concrete existing dams
on the Italian territory: an arch-gravity dam and a buttress-gravity dam. The perspective of the
practical engineer, who needs a reasonable cost in terms of time and money for the computational
resources, without renouncing at safety and reliability, is adopted. The estimation of the bearing
capacity through a pseudo-static approach, simplified but consistent with the new Italian standard
and some International guidelines, even if for low to moderate seismic areas [Charles et al., 1991],
is proposed. This represents the first step, and the only one here investigated, of a sequence, under
the assumption that a paradigm of increasing complexity should be followed for such a difficult
issue. In both the case studies here considered, a relevant failure mechanism is investigated through
three-dimensional finite element (FE) analyses, performed in [ABAQUS, 2013]. The investigation
is finally extended from the concrete bulk to the stability of foundation by consolidated procedures
in innovative numerical schemes. In the following first the case of the arch-gravity dam is
presented, then the buttress-gravity dam is analyzed.

2 Maximum bearable earthquake for the reference arch-gravity dam

A large, existing concrete, arch-gravity dam, whose maximum height H is about 130 meters,
maximum base thickness t is about 20 meters and crest-arch development is about 250 meters, is
considered. The portion of the rock foundation included in the model measures 900 x 900 meters.
Moreover, the rock foundation itself develops about 280 meters farther below the pulvino. These
foundation dimensions have been chosen with parametric studies indicating that a further increase
would not modify the response in the region of interest [USACE, 2003]. Figure 1 shows the solid
and a detail of FE models adopted for the dam-foundation system with the main section view as
well.

(a) (b) (c)


Figure 1 - (a) solid and (b) finite element models of the arch-gravity dam-foundation system. (c) Dam
main section.

The rock foundation base is fixed, while the lateral surfaces are free, i.e. not restrained at all [Léger
& Boughoufalah, 1989].
For this dam typology, a pseudo-static analysis with a nonlinear FE model is preliminarily adopted,
under some assumptions. Two nonlinear artificial joints with unilateral contact and Coulombian
friction have been included: the first, at the interface between the dam body and the pulvino, the
second one at the interface between the pulvino base and the upper surface of the rock foundation.
The first joint models a relevant design feature, which is present in many Italian dams. The second
one is introduced to account for the no-tension rock foundation behavior: a discontinuity surface
could therefore develop.
Furthermore, since dam concrete can be affected to stiffness degradation due to extensive micro-
cracking during the service life, it is reasonable to assign to such material an elastic-plastic
constitutive law. The Drucker-Prager model is then chosen, including a “cap” in the compression
regime. It demonstrates able to provide a greater realism to the calculation, smoothing the elastic
tensile peaks and simulating the micro-cracking through plastic deformations.
The “cap” Drucker-Prager law is defined in the meridional plane pressure vs deviatoric stress (p-
t) [ABAQUS, 2013]. A perfectly plastic response has been assumed. The associated parameters
have been fixed as:
• d = 5.15 MPa (intersection of the elastic-plastic yield surface with the t-axis);
• β = 53.80° (the slope of the yield surface);
• R = 7.92 (“cap” eccentricity);
• α = 0.025 (radius of the transition surface);
• pb = 100 MPa (intersection of the “cap” yield surface with the hydrostatic axis p).
Concrete mass density has been fixed to 2510 kg/m3, the isotropic elastic modulus to 30 GPa and
the Poisson’s ratio to 0.15.
The isotropic rock mass parameters have been identified by geological characterization (see Figure
1): for the most superficial rock foundation the elastic modulus is equal to 8.5 GPa and the
Poisson’s ratio to 0.2; for the deepen strata the elastic modulus is equal to 16 GPa and the Poisson’s
ratio to 0.2. Consolidating injections under the pulvino improve the superficial rock elastic
modulus to 14 GPa for a total depth of about 10 m.

2.1 Model validation

The FE model has been prepared and validated by following the procedure described in [Colombo
et al, 2016]. The methodology is based on numerical analyses to assess the structural response
including the effects of seasonal loading conditions, represented by hydrostatic pressure on the
upstream dam surface and thermal variations as recorded by a thermometers network. The
validation procedure consists in the comparison between the monitored and numerical crest
displacements and temperatures.

2.2 Results

As anticipated above, the new Italian guidelines [D.M. 26 giugno, 2014] require for existing dams
the evaluation of the bearable maximum seismic action. This task can be fulfilled by considering
the developed numerical model of the structure, which should be able to reproduce not only the
service life conditions of the dam, but also the critical condition leading to failure. In this light, for
three-dimensional concrete arch dams where complex failure mechanisms are expected, the
nonlinear behavior in terms of constitutive material and mechanical friction joint has to be
included. For the gravity-arch dam object of this study, such nonlinear domain issues have been
implemented from the beginning.
Different critical conditions up to failure should be considered at the same time:
1. stiffness degradation in the external force-displacement global response at significant positions
(e.g. crest displacement of the highest vertical section);
2. overcoming of the stress limits in large part of the bulk in the dam body or in the pulvino;
3. significant relative displacements in the perimeter joint between the dam body and the pulvino,
or between the pulvino and the rock foundation.
Obviously, the sustainability of the estimated bearable maximum seismic action for the dam-
pulvino system (previous points 1-3) by the rock foundation system has been checked in terms of
global bearing capacity. This aspect is discussed in the following section.
The bearable maximum seismic action has been investigated at the most significant system
configuration for the seismic risk evaluation, with reservoir at the maximum level.
It is also worth reminding that the seismic action is defined in the new Italian guidelines [D.M. 26
giugno, 2014] by three orthogonal components with appropriate concomitant multiplicative
coefficients, which reduce by a factor equal to 0.3 the intensity of seismic action in two main
directions. However, for nonlinear analyses, it allows to consider only one of the two main
horizontal seismic components associated to the vertical one. The signs of these seismic forces are
arbitrary; therefore, it is necessary to consider the most unfavorable combinations of such forces.
For example, the upward direction of the vertical earthquake component usually generates a
lightening effect with a non-negligible role. Furthermore, a downwards directed vertical
component could be critical for maximum compression stresses. Another critical condition, for
arch dams, consists in equivalent static forces acting in downstream-upstream direction.
Pseudo-static nonlinear analyses have been developed at linearly increasing seismic action,
subsequently to the application of the self-weight, thermal and maximum level hydrostatic loads.
Each analysis considers the linearly increasing seismic condition in a main direction in the
horizontal plane (upstream-downstream +X, downstream-upstream –X and right-left side +Y),
while an additional component in vertical direction +Z is also considered at the design collapse
limit state. The vertical component has been applied so as to light the self-weight stability effects,
thus the most critical conditions have been traced. Table 1 summarizes the pseudo-static analyses
adopted for the estimation of the bearable maximum seismic action. The seismic coefficients in
the horizontal vertical direction have been linearly amplified, maintaining the same ratio as
prescribed at the design collapse limit state with the concomitant multiplicative coefficient.
Table 1 - Pseudo-static nonlinear analyses for the estimation of the bearable maximum seismic action
Analysis Direction Seismic coefficient [m/s2]
P1 +X 10
P2 +Y 10
P3 -X 10
P4 +X and +Z 10 and 1
P5 +Y and +Z 10 and 1.93
P6 -X and +Z 1 and 1

Figure 2 depicts the global response curve in terms of crest displacement at the main transversal
cross section of the arch-gravity dam with reference to the analyses in Table 1. The vertical red
dashed line indicates the design collapse limit state. Along the horizontal axis, three analysis steps
are represented. Step time from 0 to 1 (not shown) corresponds to self-weight (with staged
construction), step time from 1 to 2 refers to other static (hydrostatic and thermal) loads, step time
from 2 to 3 is the seismic action. The seismic coefficient therefore changes between 0 and 1 during
the last step. The following comments arise.
• By comparing P1 and P3, the maximum reservoir condition has a stability effect in terms
of global response: stiffness degradation is reached at higher loading level in P3 with
respect to P1.
• Condition P2 shows a low stiffness degradation, thus the transversal seismic action can be
considered of lower risk in the maximum level condition of the reservoir.
• Comparing P4 with P1, the role of the vertical component is evident: it reduces
significantly the bearable maximum seismic action.
• Comparing P5 with P2, the effect of the vertical component has a minor role and both the
curves can be assumed equivalent in large part.
• The last comment is mostly sound by comparing P6 and P3: the responses are equivalent.
The intensity of the maximum bearable earthquake, indicated with light-blue dotted line in the
same Figure 2, is estimated at 58% of the final load corresponding to a seismic coefficient equal
to 5.8 m/s2 for a horizontal earthquake. The reasons for this limit can be traced considering the
overall situation of the system in three successive time steps.
With reference to the analysis P4 (earthquake applied along the upstream-downstream direction
and contemporary vertical component), Figure 3 shows the progress of the maximum tensile
stresses in the deformed configuration (amplification factor: 200), left side of the dam with respect
to the main cross transversal section. For clarity, only a small fraction of the rock foundation is
shown in the same figures, enough to highlight the absence of significant relative displacements
between the dam, pulvino and rock foundation.
At the same 58% of the final load in the seismic analyses, the displacement-load multiplier curve
(Figure 2) is modestly moved from linearity. Maximum tensile stresses are within the prescribed
limits of concrete. Furthermore, sliding upstream-downstream at the perimeter joint between dam
and pulvino (main transversal cross section) consists in approximately 0.6 cm. It is believed that
this configuration is still acceptable for safety requirements of the dam.

Figure 2 - Crest response in terms of displacement for the pseudo-static analysis


of the arch-gravity dam.

At 59% of the final load in the seismic analyses, the displacement-load multiplier curve is moved
away from the linearity. Maximum tensile stresses overcome prescribed limits of the bulk material
in a limited area next to the main section. There has been a significant sliding in upstream-
downstream direction at the perimeter joint between dam and pulvino (approximately 2.3 cm). It
is doubtful to say that this configuration is acceptable. In other words, an increase of only 1% of
the load multiplier (from 58% to 59%) corresponds to an increase of sliding at the joint equal to
almost four times the value observed in the previous step. At 62% of the load, the displacement-
load curve multiplier has considerably moved away from the linearity, and the maximum tensile
stresses overcome the regulatory limits in a region not excessively extended. A sensitive relative
displacement in upstream-downstream direction at the perimeter joint between dam and pulvino
is verified, about 6 cm. In subsequent steps of analysis, not reported here, the relative
displacements at the perimeter joint increase and extend. Maximum tensile stresses also extend
within the dam body over the limits. These phenomena, along with the trend of displacement-load
multiplier curve, confirm that the dam behavior overcome the safety margins, according to the
adopted model.
(a) (b) (c)
Figure 3 - Contour plot of the maximum principal stress corresponding to (a) 58%, (b) 59% and (c)
62% of the load (upstream-downstream and vertical seismic forces).

2.3 Rock mass verifications

Addressing the objective of the maximum bearable earthquake, it is important to establish that the
rock mass in the foundation does not enter into a crisis before the dam concrete bulk. Historically,
some disasters were in fact induced in the dam foundation [Charles et al., 1991; Colombo et al.,
2017].
For the verification of the dam cases considered in this paper, a paradigm of increasing complexity
is followed also for the foundations. In the first step, the local stress state is controlled in
accordance with the modified Hoek-Brown (H-B) failure criterion [Hoek, 1990].
Since safety conditions are not fully reached, a subsequent step is carried out for assessing the
possibility of a global failure according to the Wyllie limit equilibrium state [Wyllie, 1992]. In the
Appendices A and B, the H-B criterion and the Wyllie approach, respectively, are detailed.
If the geological conditions of the rock mass are more complex, i.e. the material is non-
homogeneous or strong nonlinearities is observed, then these methods are an over-simplification
and several hypotheses on the material behavior and on possible dissipative kinematisms have to
be explored. These aspects are beyond the scope of this paper.
Following the Willie approach, for the arch-gravity dam the bearable maximum seismic action
just determined (corresponding to a seismic coefficient of 5.8 m/s2 for horizontal earthquake) is
verified compatible with respect to the safety margin of the rock foundation.
Figure 4 shows pressures in the rock foundation for the P1-P6 analyses at 58% of the final load.
They are compatible with the regulatory limits: they are, even locally, lower than the bearing
capacity values as evaluated by Wyllie method (10.37 MPa). Therefore, the bearable maximum
seismic action for the dam at the maximum level of the reservoir is mainly driven by the response
of the body dam rather than by the load-bearing capacity of the rock foundation.
(a) (b)

(c) (d)

(e) (f)

Figure 4 - Contact pressures on the pulvino-foundation rock interface at the bearable maximum
seismic action as evaluated in the dam-pulvino system: (a)-(f) = P1-P6 analyses.

3. Maximum bearable earthquake for the reference buttress-gravity dam

As a second application, an existing, large concrete buttress-gravity dam, about 100 m high and
with a crest length of more than 500 m, is considered. The dam is characterized both by an
upstream reservoir and by a downstream reservoir. It presents a sequence of concrete buttresses.
Figure 5 shows the highest one with a total height of 100 m and a base of about 60 m. A three-
dimensional FE analysis of a central slice of the dam-foundation system, represented in Figure 6,
is carried out. Such slice incorporates the two central highest buttresses with the pertinent rock
foundation. The lateral surfaces that belong to the planes bounding the considered slice are
constrained in the direction of the lateral abutments and unconstrained in the upstream-
downstream and vertical directions. The lower boundary of the rock foundation is assumed fixed.
Model preparation and validation is fully described already in [Colombo et al., 2016]. From
preliminary analyses of this study, the most critical failure mechanism for the estimation of the
maximum bearable earthquake is related to the sliding conditions at uplift construction joints,
accordingly with existent literature (e.g. [Zhu et al., 2010]). In such conditions, it is verified that
the stress state in the concrete bulk does not exceed the material elastic limit. Furthermore, as
described in the previous Section, the rock foundation state has finally to be controlled.
Concrete mass density is fixed to 2500 kg/m3, the isotropic elastic modulus has an average value
of 33 GPa and the Poisson’s ratio is equal to 0.2.
The rock mass parameters are identified by geological characterization. The isotropic elastic
modulus is equal to 10 GPa and the Poisson’s ratio to 0.2.
(a) (b)
Figure 5 - (a) solid and (b) finite element models of the highest buttress.

(a)
(b)
Figure 6 - (a) solid and (b) finite element models of the considered buttress gravity dam central
module and of the corresponding portion of rock foundation.

3.1 Sliding conditions

Surfaces of potential sliding are usually identified as the (i) uplift construction joints, or the at the
dam-rock interface, and (ii) within the rock mass. For the first case, the bilinear sliding verification
approach is employed (Appendix C), as allowed by the new Italian guidelines [D.M. 26 giugno,
2014]. Instead, for the second case, the sliding strength is defined according to the Mohr-Coulomb
criterion, whose parameters are obtained from the H-B formulation [Hoek, 1990], as described in
Appendix D, with m = 1.122, s = 0.002, 𝜎𝑐 = 90000 kN/m2. Figure 7(a) shows the rock mass
characterization, related to the zone of the central buttresses, together with the bilinear strength
curve for the construction joints. Accordingly to Figure 7(a), the strength domain for the
construction joints is always smaller than the rock mass one within the range 0-4000 kN/m2. FE
simulations confirmed stress states lower than such limit. Therefore, from the comparison between
the rock mass and the concrete strengths, critical kinematic movements are not expected in the
rock mass previously to those potentially verified in the concrete joints.
Figure 7(b) depicts the sliding surfaces at the relevant uplift construction joints herein considered.

3.2 Results

The determination of the sliding critical conditions along the construction joints and the dam-rock
interface section is the first step of the proposed procedure for determining the bearable maximum
seismic action. Such analyses are performed on the base of rigid body equilibrium, using the
bilinear shear strength criterion in the concrete bulk, as explained in Appendix C. For sliding
surfaces including rock mass (section #1 in Figure 7(b)), the shear strength is assessed as described
in Appendix D, by employing the H-B criterion.
The assessment of the maximum bearable seismic action is carried out by linearly amplifying the
reference acceleration components in both vertical and horizontal directions. Following the same
approach as for the arch-gravity dam, both the seismic forces and the hydrodynamic overpressures
result amplified by the same factor. In this way, the accelerations deduced by the guidelines [D.M.
26 giugno, 2014] reference spectra are increased until reaching the limit value in one of the
horizontal concrete joints, corresponding to the potential sliding surface.

(a) (b)
Figure 7 - (a) Hoek-Brown failure domain for the rock mass and bilinear curve (see Appendix C) for
the concrete joints (in green). (b) Sections chosen for the sliding verifications in the buttress-gravity dam.

In Tables 3-5, the sliding conditions are reported for the five potential sliding surfaces depicted in
Figure 7(b). The following symbols are employed: A is the cross section area of the sliding surface;
agh and agv are the reference horizontal and vertical (reduced according to the standard by 30%)
accelerations for the pseudo-static analyses, respectively; vol is the volume of the rigid body over
the considered sliding surface; HEd is the sum of the applied horizontal forces (horizontal seismic
action and horizontal component of hydrostatic load and hydrodynamic overpressure); N is the
sum of the applied vertical forces (self-weight, vertical seismic action, uplift pressure and vertical
component of the hydrostatic pressure and hydrodynamic overpressure); σn' is the effective normal
stress, defined as N/A; τres is the shear stress strength for the Mohr-Coulomb approach; h, θ and
Φi are H-B parameters for the rock mass as used at the sliding surface #i (so evaluated only for
section #1); HRd is defined as τresA/ γr, where γr = 1.15 is a standard safety factor.
After preliminary analyses, three extreme loading combinations are considered:
• maximum upstream reservoir level and downstream empty reservoir (Table 2);
• maximum upstream reservoir level and downstream average reservoir level (Table 3);
• maximum upstream and downstream reservoir level (Table 4).
It is worth noting that the symbol “-“ means that the sections 2-5 are within the concrete dam body
and the H-B parameters are not defined due to they are related only to the rock.
The seismic amplification factor is increased until the ratio HRd/HEd reaches the critical value of 1.
For all three considered cases this condition is verified at section #2, with different magnification
values: 1.63, 1.60 and 1.54, respectively. Therefore, the minimum amplification factor leading to
sliding in the construction joints #2 for seismic loading is equal to 1.54.

Table 2 - Actions and sliding forces: maximum upstream reservoir level and downstream empty
reservoir with an amplification factor of 1.63. The critical section is #2.
Sliding surface
Parameter Unit 1 2 3 4 5
2
A [m ] 1000 608 353 235 184
agh/g [-] 0.529 0.529 0.529 0.529 0.529
0.3agv/g [-] 0.113 0.113 0.113 0.113 0.113
3 3 3 3 3
vol [m ] 3.5 10 2.2 10 1.1 10 0.5 10 0.17 103
HEd [kN] 8.2 105 7.1 105 3.3 105 1.3 105 0.3 105
5 5 5 5
N [kN] 6.9 10 4.4 10 2.1 10 0.97 10 0.3 105
σ’n [kN/m2] 691 726 611 417 154
2
τres [kN/m ] 1295 1335 1203 980 401
h [-] 1.04 - - - -
θ [°] 53.14 - - - -
Φi [°] 54.62 - - - -
HRd [kN] 1.3 106 0.81 106 0.42 106 0.23 106 0.07 106
HRd/HEd [-] 1.37 1.00 1.12 1.55 2.08
Table 3 - Actions and sliding forces: maximum upstream reservoir level and downstream average
reservoir level with an amplification factor of 1.60. The critical section is #2.
Sliding surface
Parameter Unit 1 2 3 4 5
A [m2] 1000 608 353 235 184
agh/g [-] 0.520 0.520 0.520 0.520 0.520
0.3agv/g [-] 0.111 0.111 0.111 0.111 0.111
3 3 3 3 3
vol [m ] 3.5 10 2.2 10 1.1 10 0.5 10 0.17 103
HEd [kN] 8.0 105 7.0 105 3.3 105 1.3 105 0.3 105
N [kN] 6.8 105 4.3 105 2.2 105 1.0 105 0.3 105
2
σ’n [kN/m ] 687 705 613 418 154
τres [kN/m2] 1290 1311 1205 981 401
h [-] 1.04 - - - -
θ [°] 53.15 - - - -
Φi [°] 54.66 - - - -
6 6 6 6
HRd [kN] 1.3 10 0.79 10 0.43 10 0.23 10 0.07 106
HRd/HEd [-] 1.39 1.00 1.13 1.57 2.11

Table 4 - Actions and sliding forces: maximum upstream and downstream reservoir level with an
amplification factor of 1.54. The critical section is #2.
Sliding surface
Parameter Unit 1 2 3 4 5
A [m2] 1000 608 353 235 184
agh/g [-] 0.50 0.50 0.50 0.50 0.50
0.3agv/g [-] 0.106 0.106 0.106 0.106 0.106
vol [m3] 3.5 103 2.2 103 1.1 103 0.5 103 0.17 103
HEd [kN] 7.4 105 6.5 105 3.2 105 1.3 105 0.3 105
5 5 5 5
N [kN] 6.1 10 3.9 10 2.1 10 1.0 10 0.3 105
σ’n [kN/m2] 609 636 616 421 155
2
τres [kN/m ] 1201 1231 1208 984 404
h [-] 1.04 - - - -
θ [°] 53.46 - - - -
Φi [°] 55.41 - - - -
6 6 6 6
HRd [kN] 1.2 10 0.75 10 0.43 10 0.23 10 0.07 106
HRd/HEd [-] 1.40 1.00 1.16 1.60 2.19
3.3 Rock mass and concrete bulk verifications

In a second step, it is necessary to control the local stress state in the rock mass underneath the
buttress-gravity dam and in the concrete bulk. The seismic amplification factor obtained in Section
3.2 (equal to 1.54), leading to sliding at the section #2, should not be critical for the rock mass
foundation.
For the first assessment, the H-B criterion, as described in Appendix A, is followed. In particular,
the attention is focused on capacity/demand index (CDI) defined as √𝐽2,lim /𝛾𝑅 )/√𝐽2 .
It is worth noting from Figure 8 that the CDI satisfies the standard limit range, except for the
superficial areas. This result is related to the hypothesis of linearity for the dam-foundation system
model. In particular, between the rock mass and the buttresses, no joints are introduced.
3𝑠𝜎
The area characterized by a negative value of the CDI corresponds to 𝐼1 > 𝑐 , and the
𝑚
corresponding stress state is outside the safety domain, since it is beyond the vertex of the H-B
domain.
For sake of completeness, the Wyllie limit equilibrium state is checked (as in Section 2.3) at the
foundation portion under the dam body, where the H-B CDI overcomes, even if locally, the safety
domain. The mean pressure over the dam-rock foundation interface results always within the
safety limit for the Willie approach.
Finally, for the verification of the stress state in the concrete bulk, Figure 9 depicts the maximum
and the minimum principal stresses in the central slice of the dam body. Everywhere, the limit
values are satisfied.

(a) (b)
Figure 8 - Rock mass verification for Table 4 conditions (maximum upstream and downstream
reservoir levels, amplification factor 1.54). Maps of capacity/demand index in the rock foundation.
Isometric views (a) with and (b) w/o dam body.
(a) (b)

Figure 9 - Concrete bulk stress verification at the amplification factor 1.54:


(a) maximum and (b) minimum principal stresses.

4. Discussion

The approach followed in this work is based on a pseudo-static analysis, which is a rather
conventional and not completely satisfying method in case of the verification of dams subjected
to strong earthquakes. It has been here proposed in view of the following considerations. The level
of seismic load according to the best of the actual knowledge in the area where the dams have been
built can be considered from moderate to low. The requirement of the maximum bearable seismic
action evaluation is in addition to the regular verifications for existing dams; for these verifications,
instead, response spectrum dynamic analyses have been followed as shown in [Colombo et al.,
2017]. We recall again that no action is required to the owner or to the institutions after the
determination of this maximum bearable seismic action. The conjecture is that this datum could
provide an index to orient subsequent actions from the institutions, possibly generating a graded
list of dams needing of intervention. Nevertheless, in many cases concrete dams are unique
structures, depending not only on the typology but also on their actual geometry, materials and
boundary conditions in the rock foundation, and it is arguable to reach a common approach good
for all the cases. The seismic direction, the effect of local inhomogeneities in the ground
acceleration for structures so extended in (planar) length, the eventual role of the (full or empty)
basin, of the neighboring slopes are a non-conclusive list of possible conditions giving rise to
complications difficult to be completely captured. It appears reasonable to approach the problem
starting from very simple and understood conditions and next to complicate the framework. The
present work suggests a possible starting level for the two considered dam typologies, since the
models approximate mechanisms of failure well identifiable for practical engineers.
Of course, even in a logic of limiting the economic effort, the modelling should be further
deepened. In particular, it is natural to suggest, in a context of increasing complexity, the following
steps in the possible evaluation of the maximum bearable seismic action: (i) pseudo-static analyses
with inertia forces shaped according to the first fundamental cantilever mode for gravity dams and
according to the lowest modes for arch-gravity dams; (ii) time-dependent dynamic analysis with
(at least) seven artificial spectrum compatible accelerograms. Contextually, further mechanisms
of failure could be investigated, e.g. the local opening of artificial joints, whose danger could
become relevant for the downstream population in presence of water release.

4. Conclusions

The evaluation of the bearable maximum seismic action for existing concrete dams has been
recently introduced in the Italian standard. This work is focused on two existing concrete dams, an
arch-gravity and a buttress-gravity dam. A possible procedure on common dam typologies for
modeling and evaluating the new guidelines prescription is traced. The bearable maximum seismic
action is investigated differently with respect to the two dam case studies. For the arch-gravity
dam the failure mechanism that drives the estimation is related to the overcoming of the limit stress
conditions in wide parts of the bulk, stiffness degradation in the dam global nonlinear response
and displacement discontinuity in structural or foundation joints. When the buttress-gravity dam
is instead analyzed, a sliding mechanism drives the failure. The bearable maximum seismic action
is assessed at increasing earthquake intensity through three-dimensional pseudo-static finite
element analyses.
Finally, for the critical condition in terms of seismic action in the dam bodies, the rock foundation
stress state is checked compatible. For this aim, two different approaches are followed. At first a
local verification is pursued by implementing a user subroutine in a multi-purpose FE code.
Subsequently, whenever the local stress state is not completely verified, a global verification based
on limit equilibrium is performed. The simplified approach described represents a useful
evaluation, while finer models could be then invoked to study more in deep the standard
requirement.

References

ABAQUS 6.13 [2013] ABAQUS Documentation, Dassault Systèmes, Providence, RI, USA.
Bertero, R.D. and Bertero, V.V [2002] “Performance-based seismic engineering: the need for a reliable
conceptual comprehensive approach”, EARTHQ ENG STRUCT DYN, 31, 627-652.
Bureau of Reclamation, U.S. Department of the Interior [1977] Design of Arch Dams – Design Manual for
Concrete Arch Dams – A Water Resources Technical Publication, Denver, Colorado.
Bureau of Reclamation, U.S. Department of the Interior [1999] Foundation and Geotechnical Studies for
Existing Concrete Dams, Denver, Colorado.
Canadian Dam Association – Association Canadienne des Barrage [2013] Dam safety guidelines, Canada.
Charles, J.A., Abbiss, C.P., Gosschalk, E.M. and Hinks, J.L. [1991] “An engineering guide to seismic risk
to dams in the United Kingdom”, Building Research Establishment Report.
Choi, S.O., and Deb, D. [2005] “Supplementation of generalized Hoek-Brown yield surface through the
singularity adjustment in elastic-plastic analysis”, Geosystem Engineering, 8, 43–50.
Colombo, M., Domaneschi, M. and Ghisi A. [2016] “Existing concrete dams: loads definition and finite
element models validation”, Structural Monitoring and Maintenance, 3, 129-144.
Colombo, M., Domaneschi, M., Ghisi A., Griffini, S., Novati, H., Perego, U., Petrini, L. and Valgoi, P.
[2017] “Stress verifications of large concrete existing dams: comparison of two seismic Italian codes”,
International Journal of Earthquake Engineering, 1, 61-82
Comité francais des barrages et reservoirs [2012] Guidelines for the justification of the stability of gravity
dams, France.
D.M. 24 marzo 1982 [1982] Norme tecniche per la progettazione e la costruzione delle dighe di sbarramento
- Technical guidelines for the design and construction of dams, Italy.
D.M. 26 giugno 2014 [2014] Norme tecniche per la progettazione e la costruzione degli sbarramenti di
ritenuta (dighe e traverse) – Technical guidelines for the design and construction of dams, Italy.
Darbre, G. R. [2004] “Swiss Guidelines for the Earthquake Safety of Dams”, Proc. of 13th World
Conference on Earthquake Engineering, Vancouver, Canada.
Federal Energy Regulatory Commission [2007] Engineering Guidelines for the Evaluation of Hydropower
Projects, US.
Hariri-Ardebili, M.A. and Saouma, V.E. [2016] “Probabilistic seismic demand model and optimal intensity
measure for concrete dams”, STRUCT SAF, 59, 67-85.
Hariri-Ardebili, M.A., Saouma, V.E. and Porter, K.A. [2016] “Quantification of seismic potential failure
modes in concrete dams”, EARTHQ ENG STRUCT DYN, 45, 979-997.
Hoek, E. [1990] “Estimating Mohr-Coulomb friction and cohesion values from the Hoek-Brown failure
criterion. International Journal of Rock Mechanics and Mining”, Sciences & Geomechanics Abstracts, 27,
227–229.
Legér, P. and Boughoufalah, M. [1989] “Earthquake input mechanism for time-domain analysis of dam-
foundation systems”, ENG STRUCT, 11, 37–46.
Li, H.N., Li, D.S., Ren, L., Yi, T.H., Jia, Z.G. and Li, K.P. [2016] “Structural health monitoring of
innovative civil engineering structures in Mainland China”, Structural Monitoring and Maintenance, 3, 1-
32.
Mori, F. and Spina, D. [2015] “Vulnerability assessment of strategic buildings based on ambient vibrations
measurements”, Structural Monitoring and Maintenance, 2, 115-132.
Nagarajaiah, S. and Erazo, K. [2016] “Structural monitoring and identification of civil infrastructure in the
United States”, Structural Monitoring and Maintenance, 3, 51-69.
Officé Fedéral des Eaux et de la Géologie [2003] Sécurité des ouvrages d’accumulation - Documentation
de base pour la vérification des ouvrages d’accumulation aux séismes, Switzerland.
Phansri, B., Charoenwongmit, S., Warnitchai, P., Shin, D.H. and Park, K.H. [2010] “Numerical simulation
of shaking table test on concrete gravity dam using plastic damage model”, Structural Engineering and
Mechanics, 36(4).
Reservoirs Act [1975] United Kingdom.
Ruggeri, G. [2004] “Sliding safety of existing gravity dams - final report, Technical report”, Proc. of the
Meeting of European Club of ICOLD British Dam Society - 13th Biennal Conference, University of Kent,
Canterbury, UK.
Sevim, B., Altunisik, A.C. and Bayraktar, A. [2013] “Structural identification of concrete arch dams by
ambient vibration tests”, Advances in Concrete Construction, 1(3).
Shimamoto, K., Sasaki, T. and Kondo, M. [2007] “Trial implementation of new Japanese guidelines for
seismic performance evaluation of dams during large earthquakes”, Proc. of ICOLD 75th Annual Meeting,
Saint Petersburg, Russia.
Stone & Webster Engineering Corporation [1992] “Uplift pressures, shear strength and tensile strengths for
stability analysis of concrete gravity dams”, EPRI TR- 100345, Project 2917-05.
Tekie, P.B. and Ellingwood, B.R. [2003] “Seismic fragility assessment of concrete gravity dams”, .
EARTHQ ENG STRUCT DYN, 32, 2221–2240.
Türker, T. and Bayraktar, A. [2014] “Vibration based damage identification of concrete arch dams by finite
element model updating”, Computers and Concrete, 13(2).
US Army Corps of Engineers [1994] Arch Dam Design, EM 1110-2-2201, US.
US Army Corps of Engineers [1995] Earthquake Design And Evaluation For Civil Works Projects, ER
1110-2-1806, US.
US Army Corps of Engineers [1999] Response Spectra and Seismic Analysis for Concrete Hydraulic
Structures, EM 1110-2-6050, US.
US Army Corps of Engineers [2003] Time-History Dynamic Analysis of Concrete Hydraulic Structures,
EM 1110-2-6051, US.
Wyllie, D.C. [1992] Foundations on Rock: Engineering Practice, Chapman & Hall.
Zhu, H.H., Yin, J.H., Dong, J.H. and Zhang, L. [2010] “Physical modelling of sliding failure of concrete
gravity dam under overloading condition”, Geomechanics and Engineering, 2(2).
Appendix A

The generalized Hoek-Brown limit surface can be analytically described as


𝜎3 𝑎
𝜎1 − 𝜎3 − 𝜎𝑐 (𝑚𝑏 + 𝑠) = 0 (A.1)
𝜎𝑐
with σi (i=1,2,3) principal stresses following the structural convention, i.e. tensile stresses are
positive, and σ1 ≤ σ2 ≤ σ3; mb, s and a are model parameters; 𝜎𝑐 is the uniaxial compression yield
stress for the intact rock material. For the same intact rock, it is typically assumed s=1 and a=0.5,
while mb=mintact is obtained from triaxial tests. The H-B failure envelope in Eq. (A.1) for the
rock foundation is then regularized in [Choi &. Deb, 2005] to obtain a circle in the deviatoric plane.
Model parameters are then calibrated according to the in situ experimental geomechanics. The
general equation of the modified H-B limit surface is
𝐼1
α(𝐽2 )α + β√𝐽2 + 𝑚 + 𝑠 σc = 0, (A.2)
3
being 𝐼1 = 𝜎𝑖𝑗 𝛿𝑖𝑗 the first stress invariant, 𝛿𝑖𝑗 is the Kronecker’s delta; 𝐽2 the second deviatoric
1
stress invariant, i.e. 𝐽2 = 𝑠𝑖𝑗 𝑠𝑖𝑗 with 𝑠𝑖𝑗 = 𝜎𝑖𝑗 − 𝐼1 /3 𝛿𝑖𝑗 ; 𝛼 and 𝛽 are instead
2
1 𝑎−1
𝛼 = (√3)𝑎 (−𝜎𝑐 ) 𝑎 (A.3)
𝑚
𝛽= . (A.4)
√3
Operatively, during post-processing at each integration point in the rock foundation described in
the FE model, the limit value is obtained by solving Eq. (A.2) for √𝐽2 (it is assumed a=1/2):
𝐼
−𝛽+√𝛽2 −4𝛼(𝑚 31 +𝑠 σc )
. (A.5)
√𝐽2,lim = 2α

Then, the same quantity √𝐽2 in Eq. (A.5) is calculated for the stress state obtained in the FE
analysis under the considered load combination. The ratio (√𝐽2,lim /𝛾𝑅 )/√𝐽2 relates the limit value
scaled by a safety factor (𝛾𝑅 is equal to 2.3 for the Italian standard) with respect to the square root
of the deviatoric stress invariant for the stress state obtained from a certain load combination. This
ratio, therefore, is a dimensionless risk index stating the distance from the limit surface (in Eq.
(A.2)) of the actual stress state and it will be referred to as the “capacity/demand index” (CDI). It
is worth emphasizing, however, that this index is related to a global equilibrium state, not to a local
tensional criterion, and the present method could be excessively precautionary.
To make the comparison efficient, in the FE code [ABAQUS, 2013] a user-defined contour plot
variable UVARM equal to the ratio (√𝐽2,lim /𝛾𝑅 )/√𝐽2 is created through a coded user subroutine
at each integration point. Its value is plotted for the rock foundation domain (see e.g. Figure 8).
CDI higher than 1 are considered acceptable, values lower than 1 (including negative values) are
dangerous. Since the latter result represents a local violation of a stress criterion in a bulk
continuum, engineering judgment is however necessary to assess the actual risk: violations very
limited in space could be related e.g. to a crude representation of the rock surface or other
approximations of the FE model, and therefore should not be considered with alarm. If the
violation is meaningful, a more accurate method such as the one described in Appendix B, based
on a limit equilibrium, should be followed.

Appendix B

When the local tensional state does not satisfy the H-B criterion described in Appendix A, the
Wyllie approach [Wyllie, 1992], estimating the load bearing capacity for the rock foundation, is
invoked. In this method, a global equilibrium, not some local tensional state, is considered, under
the following hypotheses: (i) the rock mass obeys to the H-B criterion; (ii) plane strain conditions;
(iii) the rock mass as a continuum body, homogeneous and isotropic; (iv) the rock foundation
weight is neglected; (v) an overload qs could be applied nearby the foundation. Figure B.1 details
the calculation scheme adopted. The rock mass below the foundation is split into two zones, each
one with its own stress state. In zone I, i.e. at the foundation sides, the stress state can be written
as:
𝑞𝑠 𝑎
𝜎3𝐼 = 𝑞𝑠 , 𝜎1𝐼 = 𝜎𝑐 (𝑚𝑏 + 𝑠) + 𝑞𝑠 . (B.1a, B.1b)
𝜎𝑐

Zone II represents the region directly below the foundation, where the stresses are defined as:
𝑎
𝜎3𝐼𝐼
𝜎3𝐼𝐼 = 𝜎1𝐼 , 𝜎1𝐼𝐼 = 𝑞ult = 𝐶f1 [𝜎𝑐 (𝑚𝑏 + 𝑠) + 𝜎3𝐼𝐼 ]. (B.2a, B.2b)
𝜎𝑐
being 𝜎1𝑖 , 𝜎3𝑖 (i=I,II) the horizontal and vertical stresses, respectively; mb, s, a, σc are the H-B
parameters defined in Appendix A; qs is the lateral overload at the foundation base height; qult is
the load bearing capacity in terms of mean stress for the rock foundation system; Cf1 is the
foundation shape correction factor [Wyllie, 1992].
The design value for the foundation load bearing capacity is obtained as qult/γR (for γR see the
definition in Appendix A) and it is checked versus the local pressure mean value on the foundation
surface. The latter is here considered coincident with the interface between the dam/pulvino and
the rock.
It is worth emphasizing that, in case of the arch-gravity dam, the choice of the transversal
orientation of the foundation vertical sections is pivotal for the result, because of the three-
dimensionality of the geometry and of the seismic load orientation.
1

Figure B.1. Scheme adopted for the foundation load bearing estimated according to (Wyllie, 1992).

Appendix C

At given sliding surfaces, represented by concrete uplift joints or concrete-rock base interface, a
residual strength value is estimated through bilinear shear strength related to a Mohr-Coulomb
friction criterion. Thus, an admissible sliding stress for the construction joints (see [Stone &
Webster Engineering Corporation, 1992] and [Ruggeri, 2004] for details) can be defined as:
𝜏res = 𝜎𝑛′ tg𝜑 + 𝑐 (C.1)
where 𝜏res is the shear stress strength for the Mohr-Coulomb approach; 𝜎𝑛′ is the effective normal
stress, defined as the ratio between the resultant of the forces acting in a direction perpendicular to
the studied section 𝑁 and the area A itself; 𝜑 is the friction angle, assumed equal to 69° if 𝜎𝑛′ ≤

𝜎𝑛,limit ′
and 49° if 𝜎𝑛′ ≥ 𝜎𝑛,limit ′
, with 𝜎𝑛,limit = 300 kN/m2 ; c is the cohesion, equal to 0 kN/m2

if 𝜎𝑛′ ≤ 𝜎𝑛,limit and 500 kN/m2 if 𝜎𝑛′ ≥ 𝜎𝑛,limit

.

Appendix D

The H-B failure criterion is used for rock mass verification at the estimated maximum bearable
seismic conditions in the dam body. In particular, the Mohr-Coulomb coefficients, namely the
friction angle and the cohesion values, are evaluated through the H-B failure criterion following
the procedure described in [Choi & Deb, 2005]. In this way, an equivalence between the two
criteria is obtained. The following admissible shear stress is considered:
𝑚𝜎𝑐
𝜏res = (cot𝜑𝑡 − cos𝜑𝑡 ) (D.1)
8
where:
1
𝜑𝑡 = arctan ( ) (D.2)
√4ℎcos 2 𝜃 − 1

1 1
𝜃= (90 + arctan ( )) (D.3)
3 √ℎ3 − 1

16(𝑚𝜎𝑛′ +𝑠𝜎𝑐 ) 1
ℎ =1+ (90 + arctan ( )). (D.4)
3𝑚2 𝜎𝑐 √ℎ3 −1
MASSIMA AZIONE SISMICA SOPPORTABILE PER DIGHE
ESISTENTI IN CALCESTRUZZO

M, Colombo1, M. Domaneschi2, A. Ghisi1, S. Griffini3


1
DICA Politecnico di Milano, Milano, Italia
2
DISEG, Politecnico di Torino, Torino, Italia
3
Studio Griffini S.r.l., Milano, Italia

SOMMARIO: La nuova normativa italiana richiede la stima della masima azione sismica per
dighe esistenti in calcestruzzo, oltre alla valutazione della sicurezza nei confronti delle azioni
sismiche attese al sito. Sono disponibili diverse linee guida e risultati di ricerca, riportati all’inizio
di questo lavoro per mostrare l'innovazione di questa peculiarità. Diverse condizioni critiche
consentono di stimare l'azione sismica massima sopportabile aumentando l'intensità del
terremoto. In questo lavoro, l'azione sismica massima sopportabile viene studiata attraverso
simulazioni numeriche ad elementi finiti in Abaqus di due dighe esistenti in calcestruzzo: una diga
ad arco-gravità e una diga a speroni. L'azione sismica massima sopportabile viene interpretata
in modo diverso in base alla tipologia di diga considerata. Con riferimento al primo caso studio,
il meccanismo di rottura è legato al superamento della condizione di stato limite in diverse parti
del corpo diga, alla riduzione di rigidezza nella risposta globale nonlineare della struttura e alla
discontuinità degli spostamenti nei giunti. Nel caso di diga a speroni, invece, il meccanismo critico
è rappresentato dallo scorrimento nei giunti.

KEYWORDS: dighe in calcestruzzo; normative; elementi finiti; azione sismica massima;


scorrimento; roccia di fondazione.

You might also like