You are on page 1of 6

Numerical Methods in Geotechnical Engineering – Schweiger (ed.

)
© 2006 Taylor & Francis Group, London, ISBN 0-415-40822-9

Reservoir, dam and rock mass interaction modeling

D. Divac & D. Vuckovic


Institute for the development of water resources “Jaroslav Cerni”, Belgrade, Serbia and Montenegro

M. Zivkovic & S. Vulovic


Faculty of mechanical engineering, University of Kragujevac, Kragujevac, Serbia and Montenegro

ABSTRACT: The objective of this paper is to show the modeling of filtration and stress-strain processes, during
the building of dam and accumulation in naturally loaded structure, using the finite elements method. Analyzed
region is modeled by the 3D finite elements mesh, whereas real geometry and disposition of different geological
environments are completely respected. Through modeling of filtration, the fields of potential and fluid velocity are
obtained for determined boundary conditions and for accepted material properties as well as for field of filtra-
tion forces as a measure of interaction between fluid and soil. Achieved filtration forces are used as input for stress-
strain analysis. The elasto-plastic material model with critical state is applied for stress-strain analysis. The paper
presents developed methodology which gives a very good evaluation of global safety of structure as well as a pos-
sibility of the critical zones localization concerning static and filtration stability shown in analysis of FR
Macedonian St. Petka dam.

1 INTRODUCTION curtain and surrounding rock mass. Based on achieved


results critical zones from the aspect of soil stress and
High dams as engineering objects have a significant strain can be identified as well as the velocity of fluid
stress value in foundations and supports because of dam flow. Also, the cumulative indicators of dam safety
dead weight as well as the water pressure on dam and (safety factor in relation to appearance of instability,
rock mass. The analysis of filtration and stress–strain total flow through the dam, etc.) can be determined.
processes in the interaction of dam and rock mass as The finite elements method is applied respecting the
well as the analysis of stability of object-rock coupled real behavior of mediums using suitable nonlinear
system represents an important part of engineering material models. The program PAK (which is devel-
studies for projecting of high dams. It also helps to oped at Mechanical Faculty in Kragujevac in coopera-
select a certain type of dam for a particular area. Rock tion with the Institute “Jaroslav Cerni”) is used and
mass on dam has to be investigated and modeled by adapted for the purpose of solving the problems of fluid
defining certain characteristics of its filtration and flow interaction and deformation of rock mass.
stress-strain behavior. These characteristics are quanti-
fied by the laboratory tests and investigation in situ. 2 MODELING OF FILTRATION PROCESSES
According to this the material characteristics for filtra-
tion and stress-strain numerical analyses are defined. Filtration processes are analyzed according to the dif-
By the development of hardware and software tech- ferential equation of fluid flow through a porous
nology, numerical methods made great expansion and medium for stationary and non-stationary conditions,
found application in many fields of technique replac- which is derived based on Darcy’s law for the case of
ing traditional methods of modeling and offering great small velocities
possibilities in daily analytical and project work.
By modeling of filtration and stress-strain processes (1)
for various conditions, the fields of potential, gradients
and fluid velocity are obtained as well as the filtration where q ⫽ the Darcian velocity, k ⫽ hydraulic con-
forces which appear as the result of fluid and soil inter- ductivity. The gradient of potential i is
action. By modeling of stress-strain processes, the
fields of displacement, stress, elastic and plastic strains (2)
of soil are obtained. The fields of all relevant values
refer to the complete investigated space: dam, injection where f ⫽ total fluid potential or head.

721
The governing differential equations of incom- The incremental iterative equilibrium equation of
pressible fluid flow through a porous medium of vari- finite element, which can be derived from principle of
able permeability have the following form, Desai virtual work (9), has the following form:
(1984),
(10)
(3)
where:
where Q ⫽ volume source/sink and S ⫽ specific
storage. For the case of stationary fluid flow right side (11)
of Equation (3) is equal to zero.
The finite element equation of balance is
is the stiffness matrix and
(4)
(12)
where S is the profusion matrix

is the vector of internal forces.


(5) F is the vector of external forces and ⌬U(i)
t⫹⌬t s

is the vector of increment of displacements. The


equilibrium iterations continue until the equilibrium
K is the hydraulic conductivity matrix
Equation (10) is satisfied within a specified tolerance.
To determinate the tangent stiffness matrix, we
(6) need to find tangent constitutive matrix

and t⫹⌬t f is the flux vector


(13)

(7)
For the case of small strains, total strains are
When the procedure converge, we can calculate the
gradient of potential (2), Darcy’s velocity (3) and (14)
flow quantity in each node of the model using the
potential, where t⫹⌬tB(i⫺1) is strain-displacement matrix. The total
displacements are
(8)
(15)
Flow quantity on a contour is equal to the algebraic
sum of flows in nodes inside that contour. In the implicit numerical procedure for stress calcula-
tion we start with the assumption that the stress-strain
3 MODELING OF STRESS–STRAIN state is known at a material point and time t. The known
PROCESSES quatities are: stress t␴, strain t␧, plastic strain t␧p at time
t, and total strain t⫹⌬t ␧ at the end of load step. The task
Stress-strain processes are analyzed by elasto-plastic within the stress integration is to determine stress t⫹⌬t ␴
material model with the concept of critical state, and plastic strain t⫹⌬t ␧ at the end of time steps.
Kojic & Bathe (2005). This analysis can give field of The basic steps in concept of the implicit stress
displacement, total and plastic strains, stress and integration are as follows:
other relevant internal variables for given initial and 1. Express all unknown variables in terms of one gov-
boundary conditions. erning parameter ⌬e mp (increment of the mean plastic
The principle of virtual work has the following strain).
form: 2. Solve the governing equation (yield function)
f(⌬emp ) ⫽ 0
(9) 3. Calculate the unknown variables using ⌬em p
the
solution of the governing equation.

722
4 COMPLEX MODEL OF FILTRATION AND where its filtration coefficient has been varied within
STRESS-STRAIN PROCESSES a 0.5–3 Lu range.

Unique model for filtration and stress-strain processes


at St. Petka dam location (interaction of the reservoir, 5 THE PARAMETRIZATION OF MODEL
dam and rock mass) is developed. Modeling is done
for terrain within area of dam with basis dimension of The material characteristics for filtration and stress-
252 ⫻ 250 m, from elevation 230 asl to 400 asl. The strain analyses are determined by using experimental
3D elements mesh was developed for all calculations and numerical methods with verification on in situ.
(filtration and stress-strain) with 17099 nodes and Generally, methods for parameterization are heteroge-
14505 elements (Fig. 1). Model is based on real filtra- neous and depend on applied investigation methods as
tion and stress-strain characteristics of presented medi- well as on material model. The filtration coefficients
ums. Therefore the mesh was formed to comprise all for quasi homogenous areas are determined by interpo-
quasi-homogenous zones by filtration and stress-strain lation and extrapolation of values obtained by experi-
parameters (in total 41 zones). In filtration analysis, mental investigation of water discharge. The model
boundary conditions with regard to potential are comprises real position and configuration of defined
defined as follows: at the part of the contour that cor- quasi homogenous zones of natural rock mass and
responds to terrain surface below nominal water level – cracking zones as well as injection curtain with possi-
potential is equal to 364 asl; at the part of the contour bilities of variation of its dimension and water dis-
that corresponds to terrain surface below river level charge. An acceptance of filtration coefficient depends
potential is 318 asl. Contour conditions with regard to on investigation level and results interpretation.
discharge (flux) are set according to model limits Although the rock mass is discontinuous medium, for
(q ⫽ 0). the dimension problem it can be approximated by the
For stress-strain analysis, displacements for bound- quasi continuum. The constitutive model should
ary conditions are set up. In these analyses, following describe the most important rock mass features for
loads are used: dead weight, filtration forces and water elasto-plastic behavior comprising strengthening and
pressure at dam. softening. The model of mechanical behavior for rock
In order to determine the significance of certain mass as a discontinuous and anisotropic medium is
elements of technical solution, several versions in developed. The interaction of cracks and monolith is
regard to prime model were established. Technical incorporated in the constitutive relations. The concept
solution of injection curtain are varied – dimensions of constitutive model for cracking rock mass is based
and filtration coefficient. The model without the cur- on coupling monolith deformation and whole systems
tain is used as well as models with injection curtain respecting their real space location and frequentation.
The model for discontinuities which is based on theory
of plasticity with appliance of non associative yield is
developed. The Desai’s one surface yield condition and
concept of limit state are used, Desai (1984), Desai &
Siriwardane (1984). The methodological procedure and
software for data interpretation obtained by investiga-
tion is developed in order to obtain parameters of con-
stitutive model of rock mass. This method is based on
iterative procedure of simulation by application of finite
elements method and errors minimization. The numeri-
cal models are formed for simulation of more typical
tests. Therefore, the definition of quasi homogenous
zones and derived investigation of mechanical prop-
erties enable definition of certain values of mechani-
cal characteristics which are used as input data for
stress-strain modeling.

6 CRITERIA FOR ANALYSIS

6.1 Criteria for filtration


Criterions for analysis of results of filtration processes
model are: allowed accumulation losses and criterion
Figure 1. Model of partition place of dam St. Petka. connected to filtration dam stability defined by

723
allowed gradients. It should be noted that water losses forces are increased linearly until the dead weight
and stability are not independent, because forbidden load is constant.
filtration velocities can also produce progressive In the second case, safety factor is determined with
increase of water discharge, what would result in pro- regards to all loads at the same time (dead weight,
gressive increase of water losses, thus endangering hydrostatic water pressure and filtration forces). After
dam stability. determining initial stresses (step 1) and achieving nom-
Filtration stability at critical locations is investi- inal values for pressure and filtration forces (step 2), all
gated, that is, at contacts of various mediums, within loads are increased linearly.
a zone where underground water erupt, and at down-
stream under the dam. With regards to this, filtration
stability is defined by a safety factor for maximum 7 THE RESULTS OF FILTRATION
output gradients in relation to allowed ones (Fs ⫽
Idop/Imax; where Idop ⫽ allowed output gradient, Imax ⫽ The 3D sections and views are selected to describe
maximum output gradient). filtration analysis around the dam and through sur-
Criterions for allowed gradients are formed, through rounding rock mass. They show fields of potential,
application of results of numerous and of many years
investigations of local filtration deformation appear-
ance and of porous environment filtration collapse. In
regards of previous, allowed gradients criterions are
spatial variables, because they generally increase when
permeability decreases.

6.2 Criteria for stress-strain analysis


Main problem of applied method is criterion of dam
collapse. As very important, it has been supposed that
a numerical divergence coincide to physical instabil-
ity. This taken into account, criterion of dam collapse
depends on applied numerical method, on control of
numerical process and on a selection of tolerance
parameters.
Integral stability criterion for the complete structure
has been adopted and expressed through one number.
Safety factor is defined as the multiplication coefficient
for nominal load with which a balance of inner energy
can still be achieved. If load is further increased, proce-
Figure 2. Field of potential (kcurtain ⫽ 1 Lu).
dure diverge; Divac et al. (1996), Divac et al. (1998).
With regards to previous, the following procedure
of calculations is applied.
Initial stress field within a rock mass (that exists in
a nature before dam building) and a field of initial plas-
tic hardening are modeled by stress-strain analysis.
Dam building and accumulation formation, from
stress-strain aspect, is equivalent to application of pri-
mary load, that is, of water pressure to dam and applica-
tion of filtration forces due to underground water flow
(with its dead weight). Stress and strain fields that cor-
respond to exploitation conditions are then obtained.
After reaching exploitation condition, primary
load incrementally increases (dead weight, water
pressure and filtration forces) until the appearance of
numerical divergence. Integral safety factor of the
complete structure is defined here in two ways.
In the first case, safety factor is determined with
regards to water pressure on dam and to filtration
forces. After determining the initial stress field (from Figure 3. Isolines of potential (kcurtain ⫽ 1 Lu) on 320 m
the dead weight load only), pressure and filtration level.

724
velocity and gradients. Analyses are performed in 8 THE RESULTS OF STABILITY ANALYSIS
order to identify the influence of injection curtain
water permability. The total flow dependence of The calculation of the stress field is firstly done, where
the filtration coefficient for injection curtain is the load of the dead weigh of soil and dam is used.
determined. Safety factor of the whole structure for the first cri-
The results of analyses of filtration for curtain terion (Fig. 6) is 4. With further load increase non-
(kcurtain ⫽ 1 Lu) are shown on Figures 2–5 (field of stability occurred. Field of stress, in this case, is shown
potential, velocities and gradients). on Figure 6, and field of total displacement on Figure 7.
The flow dependence of the filtration coefficient Safety factor of the whole structure for second cri-
around dam and through the surrounding rock mass terion (Fig. 8) is 2. With further load increase non-
ranges from 33 l/s (for cases without curtain) to 7.3 l/s stability occurred. Field of stress, in this case, is shown
(for case with curtain of 1 Lu). on Figure 8 and field of displacement in x direction
(direction of river flow) on Figure 9.

Figure 4. Field of velocity (vertical section, decreased 1.3


times in palette). Figure 6. Field of effective stress at the moment of the
crash of structure (case 1).

Figure 5. Field of gradient (vertical section, decreased 8 Figure 7. Field of total displacement at the moment of the
times in palette). crash of structure (case 1).

725
9 CONCLUSION

The complex modeling methodology is applied for


interaction between underground water flow and
stress-strain process that occur due to formulation of
dam and accumulation in naturally stressed structure.
More realistic image of structure behavior is
gained by creating this model. Rationalization and
improved safety can be achieved by application of
these modeling results.
Applied methodology gives a very good assess-
ment of global safety of structure, but also make it
possible to locate critical zones regarding static and
filtration stability.
Based on achieved results it can be concluded that
design solutions of dam “St. Petka” have required
safety.
Considering significant potential of applied
Figure 8. Field of effective stress at the moment of the methodology, its generalization and application for
crash of structure (case 2). different structures is possible. Additionally, more
information should be collected through the inspec-
tion work and observation.

REFERENCES

Desai C.S. 1984. Free surface flow through porous media


using a residual procedure, in Finite Elements in Fluids.
Gallagher R.H., Oden J.T., Zienkiewicz O.C., Kawai T.,
Kawahara M., (eds.), John Wiley & Sons. Vol. 5, Ch. 18,
pp. 377–392.
Desai C.S. & Siriwardane H.J. 1984. Constitutive laws for
engineering materials – with emphases on geologic
materials, Prentice-Hall. Inc., Englewood Cliffs. New
Jersey.
Divac D., Vuckovic D., Masala S. & Denic D. 1996. A new
method of modeling subsidence above underground min-
ing excavitacion. News Journal, JERM.
Divac D., Vuckovic D., Zivkovic M. & Grujovic N. 1998.
Numerical modeling of rock mass stress strain changes
caused by underground excavation in the Bor copper
Figure 9. Field of x displacement at the moment of the mine. 4th Eur. conference on numerical methods in geot-
crash of structure (case 2). echnical engineering, Udine, Italy.
Kojic M. 1997. Computational procedures in inelastic
analyses of solids and structures, Center of Scientific
The following fact should be noted. If all loads Research of Serbian Academy of Sciences and Arts and
(dead weight, water pressure and filtration forces) are University of Kragujevac and Faculty of Mechanical
multiplied, safety factor has less value in comparison Engineering in Kragujevac, Kragujevac.
to the case with permanent dead weight. Regarding Kojic M., Slavkovic R., Zivkovic M. & Grujovic N.
PAK–Finite element program for static and dynamic
the biggest deformations upstream on the left side of
nonlinear analysis and heat transfer, Faculty of
dam, safety of a new structure (dam with rock mass) Mechanical Engineering in Kragujevac, Kragujevac.
probably is not less than safety of existing structure Majewski S. 1997. Elasto-Plastic Model with Isotropic
(rock mass in natural canyon). Hardening/Softening Rule for Cohesive-Frictional
The results show that the global structure stability Materials, Computational Plasticity, Fundamentals and
is satisfied, i.e. building of the dam and forming of Applications, edited by Owen D.R.J., Onate E., Hinton
the accumulation HEC “St. Petka” do not violate the E., CIMNE, Barcelona.
existing natural balance. A precise inspection is done Kojic M. & Bathe K.J. 2005. Inelastic of Solid and
by using detailed calculations with analysis of poten- Structures, Springer Berlin, Heidelberg New York.
tial unstable zones.

726

You might also like