Sunteți pe pagina 1din 8

A Behavior Model for Rockfill;

Application to Settlements Calculus of


"La Yesca" Dam
Carmelino ZEAa, Rigoberto RIVERAa, Jorge B. GUILLNb y Humberto MARENGOc
a
Professors, School of Engineering, UNAM
b
Graduated of Graduate Program in Engineering, UNAM
c
Professor, School of Engineering, UNAM and Director of Projects and Construction,
Federal Electricity Commission (FEC)
Abstract. The Juarez-Badillos Principle of Natural Proportionality establishes a
relationship between natural strain and deviator stress, however in this work that
principle is applied in slightly different way, since strain is associated with secant
module so that, by solving corresponding differential equation, it manages to unify
the two branches of pre-peak and post-peak stress-strain curves in a single
equation which depends on three parameters; the initial tangent modulus, the
deviator stress failure in post-peak branch and the third, called by the authors
"ductility exponent," which controls the "form" and allows to describe the
"descent" of the post-peak stress-strain curve. The proposed model has allowed
fitting the experimental stress-strain curves and secant modulus against axial
deformation of different rockfills that were employed in the construction of the
rockfill dam named "La Yesca". Taking into account the low strain level of
experimental tests due to devices limitations, in all tested materials was adopted a
value of = 2. The model was applied to settlements calculus of "La Yesca" dam
using a commercial finite element program with light modifications in order to
implement the proposed behavior model, obtaining good results.
Keywords. Behavior Models, Rockfill, Settlements, Dams, Instrumentation.

1. Introduction
Currently there are several constitutive models to describe stress-strain behavior of
granular soils (sands, gravels and rockfill), most of them based on concepts of elasticity,
plasticity and viscosity. One of them, commonly used in professional practice for its
simplicity and easy implementation in numerical modeling using the finite element
method, is that created for Duncan and Chan [1], which is an extension of the
Kondners hyperbolic model [2]. However, for rockfill there is relatively little
experimental information on stress-strain behavior, especially for strains on the branch
called post-peak, bigger (> 5%) than that associated with the peak strength, primarily
due to experimental limitations and generation of inhomogeneous deformations in test
specimens.
Juarez-Badillos Principle of Natural Proportionality (PNP) [3] provides a
relationship between natural strain and deviator stress with respect to an original state
stress-strain, by mean of one relatively simple differential equation, thereby an

equation is counted for the pre-peak branch and another equation is counted for the
post-peak branch; but in this paper the PNP is applied in a slightly different way than
its author established, as strain is associated with the secant modulus, Es. Thus, solving
the corresponding differential equation, the solution manages a unify equation for the
two branches of the pre-peak and post-peak stress-strain curve. The single behavior
equation depends on three parameters, the first two (also included in the model of
Duncan and Chan) are the initial tangent modulus (Eo) and the failure deviator stress in
post-peak branch (qf) and a third parameter that the authors have named "ductility
exponent" (); the latter plays the "form" and allows to describe the "down" of stressstrain curve.

2. The Ductility Model


The Principle of Natural Proportionality (PNP) allows a relationship between deviator
stress and natural strains on a soil specimen under tri-axial compression test, dividing
the phenomenon in two parts: the volumetric component and the deviator component.
In turn, deviator component is described by two equations: one used to describe the
first stress-strain branch called pre-peak and other used to describe the second stressstrain branch called post-peak. However, Zea et al [4 & 5] unified the two branches
using a single equation, making assumption that second branch is a result of a
progressive loss strength, thus they found the following expression that is called
General Ductility Equation (GDE):

1
a

(1)

where: q = Deviator stress applied on a soil specimen; natural axial strain, defined
as: = ln(1+c); where c = H/Ho (Cauchys axial strain).

pr * 1

qu
Eo

po

qu qr

1
E

(2)

1
qu

(3)

(4)

1
qu qr

(5)

where: qu = maximum strength that soil tends to prolonging the pre-peak branch
beyond its limits, when strain tends to infinity; pr* = characteristic strain to 50% of qu;
qr = residual strength that soil tends when strain tends to infinity; po* = characteristic
deformation to 50% of (qu- qr); = curvature parameter of stress-strain post-peak
branch; Eo = initial tangent modulus of stress-strain curve; E = modulus related with
post-peak phenomenon.
Note that if the material doesn't have a peak in stress-strain curve, then equation
(1) becomes Kondner's equation, which is:

a b

Eo qu

(6)

On the other hand, Zea et al [6] used the PNP a little different way than JurezBadillo, because instead of choosing deviator stress as independent variable, they
chosen secant modulus (which is equal to ratio between deviator stress and
corresponding natural axial strain), finding a relationship that leadings a unify both
stress-strain pre-peak and post-peak branches, called Special Ductility Equation (SDE),
which is expressed below:

Eo q r

(7)

In which is a dimensionless parameter named Ductility exponent. The strainstress curve that can be obtained with eq. (7), is shown later in the figure 3. In that
figure, the three parameters of the SDE can be observed: E0, qr and the descent of the
curve because of the action of .
Note similarities of the equation (7) with the Kondner's equation (6); the main
difference is the parameter. It is observed that for = 1 (totally ductile soil),
equation (7) becomes Kondner's equation. It is possible to simulate a soil with fragile
failure using values of greater than one.
The proposed model (represented with the expressions GDE and SDE) may have
the disadvantage those tangent modules to stress-strain curve, Et, after peak could be
negatives and then numerical model leading to an error. In this paper, as solution to this
problem, initial secant modules to stress-strain curve, Eso, are used, it which has already
been proposed by other researchers like Zeevaert [7]; from equation (7):

E so

1
1

Eo q r

(8)

As its name indicates, GDE (which involves 5 parameters) of course is more


general and more exactly than SDE (which involves only 3 parameters), but in this
work for ease SDE was used to describing mechanical behavior of rockfill and to
analyzing the settlements of "La Yesca" dam.
3. Aplication to Hidroelectric La Yesca Case
3.1. Description of rockfill
Experimental data about triaxial tests on rockfill are low, compared with clays and
sands, mainly; due to size of samples required to ensure their representativeness. The

above demands not conventional and costly testing equipment, which is justified only
in works of great magnitude as was the case of "La Yesca" dam, a concrete face
rockfill dam with 207 m high, one of the largest of its kind on the American Continent,
located in west of Mexico. From experimental results of this project, three triaxial
compression tests were carried out at different confining stress and at different relative
densities. The tested materials were: alluvial gravel (3B), fluidal dacite rockfill (3C)
and porphyritic dacite rockfill (T). Series of consolidated-drained triaxial compression
tests (CD) were conducted on samples of 30 cm in diameter and 70 cm in height, using
a large triaxial chamber [8]. The maximum particle size was 38 mm, specific gravity of
2.63 and void ratio around 0.30 corresponding to a state of dense material.
Figure 1 shows the strainstress experimental curves of fluidal dacite rockfill, as
an example of results obtained in laboratory tests at three confining stress of 98.1,
294.3 and 490.5 kPa ; also shows the fitting obtained with the SDE (Eq. 7). The other
fittings could be consulting in Ref. [6].
3000

e0=0.31
2500

e0=0.31

q (kPa)

2000

1500

1000

3 (kPa)
98.1
294.3
490.5
Theoretical curves

e0=0.30
500

0
0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.050

Figure 1. Strainstress curves of fluidal dacite rockfill.

The adjustment parameters of the SDE obtained for the alluvial gravel, fluidal
dacite rockfill and porphyritic dacite rockfill [6 & 9] are shown in Table 1. For higher
easiness, a value equal 2.0 for the parameter was handled.
Initial tangent modulus, E0, and residual strength, qr, were plotted in a double
logarithmic scale leading to a straight line in both cases (potential adjustment).
Residual strength was plotted multiplied by 100 in order to appreciate both parameters
for all materials on the same graph as is shown in Figure 2.
1.0E+06

Parameter value, (kPa)

y = 2655.1x0.7591

y = 6213.1x0.5611

y = 1716.8x0.7603
1.0E+05

y = 33.914x

y = 23.72x
1.0E+04
Eo, Alluvial Gravel
Eo, Fluidal Dacite
Eo, Porphyritic Dacite
qr, Alluvial Gravel
qr, Fluidal Dacite
qr, Porphyritic Dacite

1.0E+03
10

y = 32.62x

100

1000

Initial confining stress, 3 (kPa)

10000

Figure 2. E0 and qr (multiplied by 100) parameters plotted against initial confining stress 3 (for all the
materials used in La Yesca dam).

Table 1. Fit parameters for all the materials used on La Yesca dam
Material
Alluvial
gravel
Fluidal
dacite
rockfill
Porphyritic
dacite
rockfill

3 (kPa)

*
0.0168
0.0238
0.0228
0.0267
0.0291
0.0281
0.0267
0.0355
0.0398

E0(kPa)
300 000
600 000
920 000
56 000
130 000
190 000
200 000
380 000
430 000

490.5
1471.5
1962.0
98.1
294.3
490.5
490.5
1471.5
1962.0

2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0

qr(kPa)
85.0
340.0
480.0
40.0
110.0
150.0
165.0
480.0
680.0

3.2. Finite element deformations analysis


In order to programming the SDE (Eq. 7) a computer program named as
FEDAM84 [10] was used, in which minor modifications were made. The Finite
Element uses the parameter known as "Elastic Modulus, E" to determine the stressstrain state at a point within the medium being analyzed; however, in the case of
materials such as rockfill, E is not known a priori, because their behavior is not linear
to usual levels of stress. In order to determine real value of E, successive iterations
are used. Figure 3 shows how "real" secant modulus was calculated in the SDE case: In
the first iteration within the adapted FEADAM84 program, initial tangent modulus,
E0, is used, as a function of initial confinement stress, 3. For the second iteration a
tangent modulus, Et, is used, as a function of deviator stress obtained in the first
iteration and assuming the parameter equals 1 (similar to use Kondners equation):

Et 1 q / qr Eo
2

(9)

Note that in Eq. (9) Eo and qr are function of 3


In the third iteration the secant modulus value is used, as a function of deviator
strain previously obtained in the second iteration and with the actual value of (in this
case 2.0) using the Eq. (8).
The Table 2 shows the numeric values of the parameters used by modified
FEADAM84 program. The information was complemented by the parameters of the
SDE.
1

Et

Deviator Stress (1 3)

qr

E0
Es
Successive
Iterations

Natural axial deformation (a)

Figure 3. It shows scheme of operation of iterations for the modified FEADAM program using the SDE.

Table 2. Material parameters of La Yesca Dam (including those of SDE)


Material

kN/m3

Kur

Rf

Kb

Ko

Alluvial Gravel
Porphyritic Dacite
Fluidal Dacite

20.47
19.55
19.75

872.8
818.5
567.5

104.7
982.2
681.0

0.76
0.56
0.76

0.04
0.07
0.06

693.7
466.3
713.6

0.71
0.60
0.42

46
45
49

0.28
0.29
0.25

2.0
2.0
2.0

A finite element mesh with 980 elements and 1053 nodes was created. The bottom
line of the dam was defined as a restricted vertical and horizontal displacements border,
and all other nodes were left free. The mesh forms 21 layers of construction.
Figure 4 shows the results of analysis for accumulated settlements under
construction of the La Yesca dam, analyzed with the model of ductility and visualized
with the Surfer 10 program.A settlement of about 83 cm was obtained in the central
area of the dam with the SDE.

Figure 4. Accumulated settlements (m) under construction of the La Yesca dam, obtained using the SDE.

In the case of filling stage of reservoir, five levels were proposed, with a filling
sequence: 50, 100, 140, 170 and 200 m elevation of the reservoir. The results of the last
stage can be seen graphically in Figure 5. Displacements of the order of 17 cm were
obtained in the concrete face of the dam.

Figure 5.Displacements (m) in the vertical direction obtained using the SDE, filling at 200 m [3].

4. Field measurements
Many instruments were installed on the embankment body [11]: a) Electrical
piezometers in the bottom, immediately behind the plinth, on the riverbed; b)
Inclinometers in the materials T and 3C; c) Topographic reference lines on surface; d)
Gauging weir at the bottom of the curtain; e) Hydraulic levels of settlement (HLS); f)

Groups of six pressure cells, and six strain gauges on three parallel sections to the river,
on platforms at different elevations.
The instrumentation in the concrete face dam aimed to determine the magnitude
and evolution of movements that pose a risk to its integrity and functionality; for this
reason were installed the following instrumentation: a) inclined inclinometers
embedded in slabs; b) unidirectional strain gauges on the vertical joints between slabs;
c) bidirectional strain gauges on the board of parapet slabs and perimeter board and
plinth; d) strain gauges rosettes on the perimeter-plinth board; e) topographical
references and f) meter-gauge type joint plate in the slabs and vertical joints with
parapet. In particular, the meter seal between the curb and the slab, at the maximum
dam section, horizontal construction joint and the elevation where the double curvature
of the slab was predicted, clinometers, and load cells were installed.
Figure 6 shows contours of settlements (cm) and moduli of deformability
determined with hydraulic levels at La Yesca dam for the end of construction process.
The settlements measured in the body of the embankment due to own weight in terms
of the HLS installed, a peak value of about 97 cm was recorded in material T, for the
maximum cross section (L-21), approximately at mid height of the embankment dam.
This value is quite similar to that found with the vertical inclinometers of about 92 cm.
NAME EL.578.00

EL. 579.00

EL. 576.00

LT2-M13

LT2-M12

V-82

V-83

V-84

V-85

V-86

V-87

V-88

V-89
540.00

NAMINO EL.518.00

N
CO

CR

E
ET

C
FA

1,4:

LT2-M9

0.5
:1

No influence of the reservoir on


the concrete face.

:1
1.4
1
0.5:

ELEVATIONS (meters above sea level)

NAM0 EL.575.00

Settlements at june 21th of 2012


Magnitudes in cm

513.73
1
1

V-53

V-54

V-55

V-56

V-57

V-58

V-59

V-60

V-61

500.00
497.18

1,4

:1

EL.473.00

476.84
1

V-23

1.8

V-24

V-25

V-26

V-28

V-27

V-29

V-30

V-33

5:1

463.00
457.17

V-31

V-2

V-3

V-32
1

3B

1,4:

V-1

CASETA C-9

V-34

V-4

V-5

V-6

3C

V-7

V-8

V-9

V-10

434.64

V-12

V-11

V-13

425.00

1
1

PLINTH

El. 409,00

1,8:

IC-6

0+300

0+250

0+200

0+150

0+100

0+050

0+000

0-050

0-100

0-150

0-200

0-250

0-300

0-330

IC-8

Figure 6. Contours of settlements (cm) and modules of deformability determined with hydraulic levels at
La Yesca Dam for the end of construction process.
NAME EL.578.00
EL. 576.00

EL.473.00

CO

1
1,4:

CE

:1

FA

TE

0.5

NAMINO EL.518.00

EL. 579.00

0.5:
1
1

RE
NC

5:1
1.8

3B

1,4:

3C

0+350

0+300

0+200

0+150

0+100

0+050

0+000

0-050

0-100

0-150

0-200

0-250

0-300

0+250

PLINTH

0-330

ELEVATIONS (meters above sea level)

NAM0 EL.575.00

Reservior level 558.14


Settlements after first filling, in cm
from apr 2, 2012 to oct 27, 2014

Figure 7.Contours of settlements (cm) in the maximum section (L-21) of the rockfill during the filling of the
reservoir and the two years of operation.

Figure 7 shows contours of the settlements (cm) measured of the embankment


during the filling of the reservoir and the two years of operation. It also notes that the
settlements in the 3B and T materials are not very different due to their similar

characteristics deformability and increased confinement of the material T. The


calculated values of deformability moduli at the end of construction for 3B material, T
and 3C, were 174 MPa, 164 MPa and 92 MPa, respectively. The influence of increased
deformability of the material 3C and push the reservoir is only appreciated in the upper
third of the rockfill, where settlements reach just over 24 cm.

5. Conclusion
The proposed Ductility Model allowed fitting the experimental stress-strain curves and
secant modulus against strain for different types of rockfill that were employed in the
settlements analysis in the construction stage of La Yesca dam. Taking into account
the low strain level of experimental tests due to devices limitations, in all tested
materials was adopted a value of = 2.0. The behavioral model was implemented on a
commercial finite element program and used to calculate the settlement of La Yesca
dam. The construction stage results were compared with field measurements, finding
good agreement between the estimated and the measured settlements. Given the few
number of parameters and, moreover, their easy laboratory determination, this model
has a great performance in engineering practice to make deformation analysis on
engineering formed or supported on granular materials, such as earth and rockfill dams.

References
[1] J. M. Duncan, C. Y. Chang; Nonlinear analysis of stress and strain in soils, Journal of the Soil Mechanics
and Foundation Division, ASCE, 96 (SM5) (1970), 1629-1653.
[2] . R. L. Kondner, Hyperbolic Stress-Strain response: cohesive soils, Journal of the Soil Mechanics and
Foundation Division, ASCE, 89 (No. SM1) (1963): Proc. Paper 3429, 115-143
[3] E. Jurez-Badillo, General volumetric constitutive equation for geomaterials, Special Volume on
Constitutive Laws of Soils, XI International Conference on Soil Mechanics and Foundations
Engineering (1985), San Francisco, USA., 131-135.
[4] C. Zea and R. Rivera; One Expression Based on Kondners Equation to Simulate Deviator Stress versus
Strain Behavior, Proceedings of the XXIV National Meeting of Soil Mechanics and Geotechnical
Engineering (2008), Aguascalientes, Mexico.
[5] J. B. G. A. Mercado. and E. Alvarez; A Ductility Model Applied to Over Consolidate Clays.
Undergraduate Thesis (In Spanish). Facultad de Estudios Superiores Aragn, UNAM (2015). Mexico.
[6] C. Zea, R. Rivera, J. B. Guilln, H. Marengo; Description of mechanical behavior of rockfill using a
ductility model, Proceedings of the XXVII National Meeting of Soil Mechanics and Geotechnical
Engineering (2014), Puerto Vallarta, Mexico.
[7] L. Zeevaert; Foundation Engineering For Difficult Subsoil Conditions, Ed. Van Nostrand Reinhold
Company, USA (1983).
[8] . R. J. Marsal, D. Resndiz, Presas de Tierra y Enrocamiento, Ed. Limusa, Mxico D.F., 1975.
[9] . J. B. Guillen, Analysis of La Yesca dam settlements in its construction stage with a ductility model,
based on the Principle of Natural Proportionality, thesis of master in engineering, UNAM (2014)
Mexico.
[10] J. M. Duncan, R. B. Seed, K. S. Wong, Y. Osawa; FEADAM84: A Computer Program for Finite
Element Analysis of Dams, Geotechnical Engineering Report No. SU/GT/84-03, Deparment of Civil
Engineering(1984), Stanford University
[11] H. Marengo and R. Rivera. Some relevant aspects of geotechnical design for La Yesca dam:
description of its behavior during the construction stage and first filling after one year of built. Studies
on modern technology of rock-fill dam construction and hydropower development (2013), Chincold
publication, Beijing China, November, p.p: 693-718.

S-ar putea să vă placă și