next up previous
Next: Implementation of the Natural Up: Natural Neighbor Interpolation Previous: Natural Neighbor Interpolation

Governing Equations and Weak Form

In order to study a model PDE, let us consider small displacement elastostatics, which is governed by the equation of equilibrium:

$\displaystyle \boldsymbol{\nabla} \cdot \boldsymbol{\sigma} + \mathbf{b} = 0 \ $   in $\displaystyle \Omega$ (7)

where

$\displaystyle \boldsymbol{\sigma} = \mathbf{C}: \boldsymbol{\varepsilon} \, , \ \ \boldsymbol{\varepsilon} = \boldsymbol{\nabla}_s \mathbf{u} \,.$ (8)

In the above equations, $ \Omega \subset {\mathbf{R}\,}^2$ is the domain of the body, $ \boldsymbol{\sigma}$ is the Cauchy stress tensor, $ \boldsymbol{\varepsilon}$ is the small strain tensor, $ \mathbf{b}$ is the body force per unit volume, $ \mathbf{C}$ is the material moduli tensor, $ \mathbf{u}$ is the displacement, $ \boldsymbol{\nabla}$ is the gradient operator, and $ \boldsymbol{\nabla}_s$ is the symmetric gradient operator.

The essential and natural boundary conditions are

$\displaystyle \mathbf{u} = \bar{\mathbf{u}} \ $   on $\displaystyle \Gamma_u \,, \ \ \mathbf{n} \cdot \boldsymbol{\sigma} = \bar{\mathbf{t}} \ $   on $\displaystyle \Gamma_t \,,$ (9)

where $ \Gamma$ is the boundary of $ \Omega$, and $ \bar{\mathbf{u}}$ and $ \bar{\mathbf{t}}$ are prescribed displacements and tractions, respectively.

The weak form (principle of virtual work) is

$\displaystyle \int_{\Omega} \nabla_s \delta \mathbf{v} : \boldsymbol{\sigma}\, ...
...bf{t}}\, d\Gamma \ \ \forall \ \delta {\mathbf{v}} \in {\cal{H}\,}_E^1 (\Omega)$ (10)

On substituting the trial and test functions in the above equation and using the arbitrariness of nodal variations, the following discrete system of equations is obtained:

$\displaystyle {\boldsymbol{K}} {\mathbf{d}} = {\mathbf{f}}^{ext} \,,$ (11)

where

$\displaystyle {\mathbf{K}}_{IJ} = \int_\Omega {\mathbf{B}}_I^T {\mathbf{D}} {\m...
... \, d \Gamma + \int_\Omega {\boldsymbol{\Phi}}_I^T {\mathbf{b}} \, d \Omega \,.$ (12)

In the above equation, $ \boldsymbol{\Phi}_I$ is the shape function vector and $ {\mathbf{B}}_I$ is the matrix of shape function derivatives.


next up previous
Next: Implementation of the Natural Up: Natural Neighbor Interpolation Previous: Natural Neighbor Interpolation
N. Sukumar