09 Chapter 1
09 Chapter 1
09 Chapter 1
1.0 INTRODUCTION
Dams are generally constructed for flood control and conservation. Various
purposes for conservation are irrigation, power generation, navigation, domestic,
recreational and industrial purposes. The selection of dam type depends on the purpose
for which it is built, site topography, storage capacity, hydrology, availability of local
construction materials etc.[1,2] Consideration of the tri-dimensional arch action to hold
water load in a narrow gorge brought the idea of arch dam. Of all the dam types
creating water storage facilities, arch dams curving across the breadth of the valley with
one mighty vault are doubtlessly the most economic and elegant, exacting the needs of
concrete and rock.[3,4] The conventional methods adopted for the analysis of all types
of arch dams are cylinder theory, method of independent arches, trial load and model
analysis which are found to be of limitations for multiple radius arch dams of height
greater than 100 m.[5,6] Later, accurate methods are necessitated by eliminating many
assumptions made in the traditional methods for ensuring safety and economy which
led to numerical methods such as finite difference, finite element and boundary element
for arch dams. Of these, finite element is the most effective method for handling a
continuum like arch dam since, it gives a more realistic stress distribution and more
flexibility with regard to geometry and boundary conditions than other methods.[6-9]
Hence, a critical study on how the finite element method resolves the complexity in the
case of an arch dam of varying geometry is presented in this thesis.
1.1
THE NEED
Earlier, finite element method with two dimensional analysis using plane stress
and plane strain as well as shell theory that actually approximates three-dimensional
problem by two-dimensional one were used. Though it gives good results for a thin
arch dam, thick arch dam requires a rigorous three dimensional analysis.[9-12] Studies
as well as software are seen developed so far using isoparametric brick elements; 8-21
nodes, as well as shell elements; 16 nodes, for modeling the geometry which will not
accurately define the complex geometry of a thick arch dam of variable curvature
especially the extrados curve.[12-20] Hence, there is an urgent need for considering the
effect of variable curvature by approximating the geometry with higher order
polynomials incorporating more nodal points at element level itself while modeling.
The available literature and software show that the hydrostatic pressure on the
curved surface is seen approximated as normal to the surface by means of certain
global coefficients to the horizontal pressure on vertical surface.[14-19] In fact, the
magnitude as well as direction will be varying at each point, i.e. water pressure will be
normal to the curved surface, horizontal and vertical extrados, with components in the
three directions. In the finite element method, water pressure needs to be considered
more accurately as actual distributed surface forces on each element by direction
cosines and numerical integration.[20,21] Similarly the silt pressure, uplift and dynamic
effect of the reservoir water also will have to be considered at element level itself.
which is also approximate.[15-20] The load vector due to gravity, if taken as a body
force by numerical integration throughout the volume of each element, can give more
dependable results. The seismic inertial effects by way of equivalent static load
components as horizontal and vertical peak ground acceleration then can also be
incorporated in the body force like this.
Trial load analysis gives comparable results with 3D Finite element method only
for the simple cylindrical shape and for variable curvature arch dams of greater height
the trial load assumptions are dubious. Recent studies considering the various aspects
of design, variable curvature arch dams are found to be more practical for valleys of
heights exceeding 100m.[4,5,22,23] Hence, it is felt that there is a great need for
rigorous three dimensional finite element analysis to study the actual performance of a
practical arch dam structure by approximating the actual geometry with higher order
polynomials considering the surface pressure components in all the three directions and
body force components in the respective direction by developing a software using
advanced programming technique equipped with effective pre-post processing
facilities.[20-27]
1.2
The definition for an arch dam by ICOLD includes all curved dams, where the
base-thickness is less than 0.6 times the height.[2] Mainly arch dams are grouped into:
(i)
Constant radius
(ii)
Variable radius
(iii)
Constant angle
1.3
(iv)
Multiple arch
(v)
Cupola (shell)
(vi)
Arch gravity
(vii)
Mixed type
Preliminary methods
a. Thin cylinder theory
b. Thick cylinder theory
c. Elastic theory
d. Active arch method
e. Cains method
f. U.S.B.R. criteria
g. Institution of Engineers, London
h. R. S. Varshneys equations
(ii)
Elaborate methods
a. Inclined arch method
b. Tolke method
(iii)
(iii)
Experimental method
Model studies
(i)
Cylinder Theory
The simplest and the earliest of the methods available for the design of an arch
dam is the cylinder theory. In this theory, the stress in an arch dam is assumed to be the
same as in a cylindrical ring of equal external radius. The arch thickness is calculated
by the thin cylinder formula. The cylinder theory does not allow for the discontinuity of
the arch at the abutment and is, therefore, highly approximate. The use of cylinder
theory has been restricted to dams less than 30 m in height located in narrow valleys. A
low value of permissible stress in concrete, usually about 60 per cent of the permissible
stress, issued to allow for the highly approximate nature of the formula. The cylinder
theory is only of historical importance now.
(ii)
interaction between them. It is assumed that all horizontal water loads are carried
horizontally to the arch abutments by arch action and that only the dead load weights
plus the vertical water loads in the case of sloping upstream face are carried vertically
to the foundation by cantilever action. If the canyon is relatively regular and narrow
and the dam is of low height so that a symmetrical thin structure with large central
angle can be adopted this method may give reasonably satisfactory results.
Practically the water load is transferred to the foundation and abutments, both
by horizontal arch action and vertical cantilever action. The vertical cantilevers are
restrained at the foundation and must bend under their share of water load until their
deflected positions coincide with the deflected positions of horizontal arch elements.
The theory that the entire water load is carried horizontally to the abutments is
therefore, incorrect and the design that ignores vertical cantilever action can seldom be
considered as wholly satisfactory.
(iii)
horizontal water load is divided between the arches and cantilevers so that the
calculated arch and cantilever deflections are equal at all conjugate points in all parts of
the structure. Because the required agreement of all deformations is obtained by
estimating various load distributions and computing the resulting movements until the
specified conditions are fulfilled, the procedure is logically called trial-load method.
Trial load analyses may be classified into the following types depending on their
relative accuracy and corresponding complexity.
a. Crown Cantilever Analysis
Crown-cantilever analysis consists of an adjustment of radial deflections at the
crown cantilever with the corresponding deflections at the crowns of arches. This type
of analysis assumes a uniform distribution of radial load from the crowns of arches to
their abutments and neglects the effects of tangential shear and twist. While the results
obtained from this analysis are rather crude, it has the advantage of very short time to
complete the analysis. If used with judgment, it is an effective tool for appraisal studies.
b. Radial Deflection Analysis
A radial deflection analysis is one in which radial deflection agreement is
obtained at arch quarter points with several representative cantilevers by an adjustment
of radial loads between these structural elements. With the use of this type of analysis,
loads may be varied between the crowns and abutments of arches, thus producing a
more realistic distribution of load in the dam. A radial deflection analysis may be used
for a feasibility study.
c. Complete Trial Load Analysis
A complete trial-load analysis is carried out by properly dividing the radial,
tangential and twists loads between the arch and cantilever elements until-agreement is
reached for arch of the three axial and three rotational movements for each arch
cantilever node point. The accuracy of this analysis is limited only by the exactness of
the basic assumptions, the number of vertical and horizontal elements chosen, and the
magnitude of the error permitted in the slope and deflection adjustments. In view of the
comprehensive and involved nature of the complete trial-load analysis, it is desirable
that preliminary studies of tentative dams are first carried out by simplified methods;
crown cantilever analysis and radial deflection analysis, to obtain a dam; proposed for
complete trial load analysis, which is most suitable for the given site and whose
dimensions are as close to the final as practicable.
(iv)
1.4
Although the detailed analysis of an arch dam can be done by various methods
as explained in Section 1.3, the most reliable and accurate method used is the finite
element method.
contains a set of differential equations will be more reliable than a physical model;
experimental investigation, because the small scale model may not always simulate all
the features of the actual continua. For this purpose, it is appropriate to go for a
mathematical description of the physical model; each ordinary as well as partial
differential equations expresses a certain conservative principle. Here, for an arch dam,
a three dimensional continua is to be analysed. Each variable inside the continua as
well as its dependent variable must be in balance with various factors that influence the
variable. Thus, for the quantitative description of physical phenomena an analyst go
for a system of partial and ordinary differential equations valid for a certain domain and
impose on the system suitable boundary and initial conditions. Various forms of
discretisations exist in which the infinite set of unknown functions is replaced by a
finite number of unknown parameters.
The merits of a computer aided finite element analysis of an arch dams are:
Computer solution gives detailed and complete information for all the
relevant variables throughout the domain of interest.
1.5
LITERATURE REVIEW
The researches done for the arch dam design, analysis, construction and
functioning are summarized as under:
The search for the best shape for arch dams and the optimization of the shapes with
relevant analysis methods taking in to account the effect of curvature and constraints of
construction evolved various solutions like constant/variable angle, constant/variable
curvature, single/double/multiple curvature arches, membrane shapes etc. These aspects
have been very critically studied, papers consolidated and published.[4] A numerical
model for membrane shape is seen developed to find the equilibrium position of a
membrane shell under external loadings and boundary conditions. The automatic
checking of the stresses were then done by the general finite element programs like
ADAP, SAP etc or dependable methods of analysis like shell theory and complete
adjustment methods. US Commission on Large Dams has concluded the next iteration
of reporting on the numerical analysis of dams in June 1999, indicates that computer
simulation of dams continues to be a topic of pressing national interest.[8]
In the shape optimization studies, the main methods adopted for finite element
mesh formulation and refinement were movements of selected boundary nodes,
boundary shapes and spline functions which eliminates higher order polynomials.[4]
Since conventional methods were used for analysis, the most common trial load method
in which the dam as an assemblage of arches and cantilevers, the finite element mesh
was also developed along vertical and horizontal directions in resemblance with the
arches and cantilevers.[26,27] Criteria for analysis and design of arch dams have been
dealt with by many engineers and methods evolved as explained in 1.3.
10
Extensive researches are seen in the area of rock mechanics, since the
characteristics of the foundation strata has of vital importance in the stability, using
boundary and block element methods. Recently block theory has been widely used for
the elastic visco plastic and stochastic analysis of discontinuous rock masses.[28-32]
Arch dams were also analyzed by block elements and coupled trial load and block
element method where the arch dam is considered as an arch cantilever system and
foundation as block element system.[33] Stability analysis of dam abutments by 3D
elasto plastic finite element method is seen in a case study with Hough gravity arch
dam in China.[34] Sliding and stability analysis of dam foundation as well as slope
stability analysis with finite elements was attempted for the stability of rock slopes for
foundation and abutments.[35-39] Simulation and stability of cracks and fractures have
been experimentally as well as with crack model studied, stability checked.[40-44]
Fracture analysis of concrete with discrete crack model was done by the boundary
element method seepage analysis of crack and cracking analysis with a three
dimensional finite element model.[45] In many of the conferences it has came a matter
of importance the numerical analysis of arch dams and the ultimate strength evaluation
numerically.[4-9] Shell elements were widely used for the finite element analysis of
thin curved dams and for which software have been developed.[14,46,47]
11
structure interaction studies in time domain.[54,55] Shaking table test for fragments of
gravity and arch dams were also carried out.[56,57]. Dynamic response of arch dams in
earthquake is experimentally and theoretically studied.[58-62] Researches are carried
out for the seismic analysis which include dam water interaction and dam foundation
interaction with instrumented structures case studies California and Kobe dam
foundation rock interaction effects in frequency domain response functions.[63-66]
Three dimensional analysis of spatially varying ground motions around a uniform
canyon and impedance functions for foundation in a homogeneous half space.[67,68]
The seismic response of a dam in frequency domain by boundary element, modal
approach, (FE-HE)-BE model which allows for the rigorous representation of the
dynamic interaction between the dam, foundation rock and water are studied.[69-74]
Three dimensional fluid hyper element for dynamic analysis of concrete arch dams and
impedance functions for three dimensional foundation supported on an infinitely long
canyon of uniform cross-section in a homogeneous half-space are also studied.[75-77]
Analysis of arch dam including seismic effect is attempted by continuum damage
concrete model for gravity dam reservoir systems.[78-81] Distributed memory parallel
element by element scheme based on Jacobi conditioned conjugate gradient for 3D
finite element analysis is seen developed.[82]
12
scheme for plane and curved surfaces using isoparametric coordinates in which a
complex region is divided into eight noded quadrilaterals which are viewed in the form
of a rectangular pattern.[10] Saini in 1991, has attempted finite element static analysis
for the Kasari dam. As two-dimensional analysis is an approximation in the transition
zone of the dam where rapid changes occur in the thickness of the dam, in order to
obtain a realistic picture of the stress distribution and the region of stress concentration,
a three-dimensional finite element analysis was carried out using twenty noded
elements. But this analysis is applicable for straight gravity dams only. Dynamic
analysis was also done for the effect of seismic forces.
13
GDAP has been specifically designed for the analysis of arch dams. It uses the
basic program organization and numerical techniques of SAP-IV but has pre and post
processing capabilities. It uses thick shell elements, which is represented by its mid
surface nodes. The 16 noded shell element in the GDAP and general 3D solid elements
of 8 and 21 nodes are found used for modeling the geometry. An appropriate finite
element mesh for an arch dam can only be achieved by careful consideration of the dam
geometry and the type of analysis for which the dam is modeled. For example, the
finite element model of a double curvature thin shell structure differs from the model of
a thick gravity arch section.[15]
ABAQUS (1988), NISA (2001), STAD PRO 2005 etc., can also be used in the analysis
of three dimensional continuums by shell or tri linear quadratic elements, but not
specific to the point of thick curved elements. One of the most important requirements
in arch dam analysis is to develop accurate models, representative of the actual 3D
behavior of the system.
The necessary Geometric Data for constructing a finite element mesh of an arch
dam is obtained from drawings containing information defining the geometry of the
dam shape.[7,9,15] These include the plan view and section along the reference plane.
In practice, arch dams are geometrically described as multi centered arches with their
centers varied by elevation in addition to the arch opening angles and radii varying for
each side with elevation. Preparation of finite element mesh data from these geometric
data is very time consuming especially for multi layer meshes because most general
purpose finite element programs cannot directly handle these data which is another
draw back.
14
1.6
15
The arch dam descretisation are seen as an assemblage of horizontal arches and
vertical cantilevers. The 16 node shell and general 3D solid elements of 8 and 21 nodes
provided for modeling the geometry are not sufficient to accurately define the shape of
a thick variable curvature arch dam. This limitation in shape approximation as
explained above is to be resolved by the help of higher order polynomial shape
functions including more number of nodes at element level relevant to the curve in each
direction to take care of the curvature effect.[24,26,83,84] The descretisation using tri
quadratic brick element and vertical arch-cantilever system is shown in Fig 1.4.
16
In most of the studies and programs intended for concrete gravity dams as
explained above, the hydrostatic pressure is considered to act normal to the surface and
thus acts in the same direction for all the elements of that face. Even in the finite
element method analysis studies for an arch dam, the load vectors assumed with certain
coefficients for horizontal hydrostatic pressure. For an arch dam of complex geometry,
the water pressure on each point acts in a direction normal to the surface; the surface
being curved in both directions the magnitude as well as direction of water pressure
varies from point to point; element to element depending on the height of water
column. This normal hydrostatic pressure to the surface will have components in all the
three directions which have to be dealt accurately in the finite element method rather
using coefficients.[20] Similarly the pressure due to the accumulated silt is also to be
considered. This aspect is to be effectively solved taking in to consideration, the
direction cosines of the pressure at each point and by numerical integration in the
finite elements itself for which a program which resolves the pressure to all the three
directions at a particular point is required. Fig1.5 gives a schematic representation of
the distribution of loads and radial deflection of cantilevers in the Trial load method.
17
Fig 1.5: Distribution of Loads and radial deflection cantilevers; Trial Load Method
The other possible load vectors due to change in chemical properties with aging,
temperature effects, silt pressure, uplift, fluctuation of reservoir level, gravity and body
forces, seismic effect etc which affects deflection and stress values will have to be
18
considered in the finite element program.[5,9,85,86] This will be required for obtaining
a reliable computer simulation.
It is felt from the above limitations that, apart from the general purpose
software, a problem specific software for the 3D analysis of an arch gravity dam
structure using quadratic hexahedral elements, defining the geometry accurately with
the help of advanced computer programming technique with verification by plotting
capable of handling various load vectors and applicable to a variety of structural
mechanics problems is essential to be developed.
Hence the scope of the present work is to find out suitable methods for
resolving the above limitations using finite element method and the development of a
finite element analysis software for the elastostatic analysis of an arch dam having
efficient pre and post processing capabilities equipped with the above load vectors
which is capable of handling a variety of structural mechanic problems using the
advanced object oriented programming technique: Visual C++, a simultaneous finiteelement plotting program in Matlab corresponding to the Visual C++ program arrived
with various options of plotting the original shape, deflected shape, sectional plan.[87]
Established works of similar nature/case studies are intended for verification and
comparison of results along with parametric studies.
1.7 OBJECTIVES
19
To develop an accurate basic Finite Element Software having pre and post
processing capabilities for three dimensional elasto static analysis of a solid
continuum using linear and
quadratic
hexahedral elements
capable
of
To develop a method to handle the varying hydrostatic pressure for each reservoir
levels acting on the curved boundary of an arch dam in magnitude and direction
from element to element so as to improve the result of analysis and further
verification of the actual instrumentation values. Similarly, silt pressure as well as
hydrodynamic pressure due to reservoir water during earthquake, self weight as
body force due to gravity at element level by numerical integration and inertial
forces of dam due to horizontal and vertical earthquake accelerations.
20
1.8
METHODOLOGY
A flow diagram to arrive the above objectives of the research is shown below.
Finite element program is developed for a three dimensional solid continuum in Visual
C++ and plots using Matlab.
DEVELOPMENT OF FEA PROGRAM FOR 3D SOLID CONTINUA BY
08 NODED HEXAHEDRAL ELEMENTS
20 NODED HEXAHEDRAL ELEMENTS
PROGRAM VALIDATION
LxTxH PARALLELEPIPED
2x2x2 CUBE WITH CENTRE (1,1,1)
2x2x2 CUBE WITH CENTRE (0,0,0)
21