Nonlinear Finite Element Analysis of Concrete Structures
Nonlinear Finite Element Analysis of Concrete Structures
Nonlinear Finite Element Analysis of Concrete Structures
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
- 61 -<br />
tensor c.. follows which in turn determines an approximative<br />
stress tensor o^ through the constitutive equation- It is obvious,<br />
however, that this stress tensor will not in all points<br />
<strong>of</strong> the structure satisfy the equilibrium condition, eq. (1), as<br />
no means have been taken for this purpose. However, the method<br />
<strong>of</strong> Galerkin ensures an approximative satisfaction <strong>of</strong> these equilibrium<br />
equations. Let us now consider this procedure in some<br />
detail.<br />
The true displacements u. are approximated by<br />
u. = u a = N. a a = 1, 2 n (4.1-10)<br />
i i la a<br />
where the tensor N.<br />
depends on position and is assumed to be<br />
known while the coefficients a are to be determined. It is cona<br />
venient to consider these coefficients as displacements <strong>of</strong> some<br />
points distributed all over the structure. These points are<br />
termed nodal displacements. Obviously, to obtain an accurate<br />
approximation by the available n degrees <strong>of</strong> freedom, the nodal<br />
points should be distributed closely where large changes in the<br />
displacement field are expected- It is a crucial feature <strong>of</strong> the<br />
finite element method that the approximative displacement functions<br />
given by the tensor N.<br />
and, in a finite element context,<br />
termed shape functions are not the same all through the structure,<br />
but render different expressions for each subdomain or<br />
element, the total <strong>of</strong> which covers the whole structure. Moreover,<br />
the finite element method assumes that within each element, the<br />
approximative displacements can be expressed solely by the nodal<br />
displacements located within or on the boundary <strong>of</strong> the element<br />
in question. However, with these remarks in mind we will retain<br />
the formulation given by eq. (10).<br />
The approximative strain tensor follows from eqs. (2) and (10)<br />
i.e.<br />
4j = B ija a a<br />
a - !' 2 n (4.1-11)<br />
where the tensor B .. depends on position and follows from knowledge<br />
<strong>of</strong> N i(j . Through the constitutive condition eq. (3) , the