Algorithmic stiffness

Differentiation of the elastic stress-strain relations (4.8) and the discrete flow rule (4.9) yields
$\displaystyle d$$\displaystyle \mbox{\boldmath$\sigma$}$$\displaystyle _{n+1}$ $\displaystyle =$ $\displaystyle \mbox{\boldmath$D$}$$\displaystyle \left(d\mbox{\boldmath$\varepsilon$}_{n+1}-d\mbox{\boldmath$\varepsilon$}^p_{n+1}\right)$ (4.24)
$\displaystyle d$$\displaystyle \mbox{\boldmath$\varepsilon $}$$\displaystyle ^p_{n+1}$ $\displaystyle =$ $\displaystyle \sum\left(\lambda^i\partial_{\sigma\sigma}gd\mbox{\boldmath$\sigm...
...a}\mbox{\boldmath$\kappa$}d\lambda^i\right)+d\lambda^i\partial_{\sigma}g\right)$ (4.25)

Combining this two equations, one obtains following relation

$\displaystyle d$$\displaystyle \mbox{\boldmath$\sigma$}$$\displaystyle =$   $\displaystyle \mbox{\boldmath$\Xi$}$$\displaystyle _{n+1} \left\{d\mbox{\boldmath$\varepsilon$}_{n+1}-\sum\lambda^i\...
...}\mbox{\boldmath$\kappa$}d\lambda^i - \sum d\lambda^i\partial_{\sigma}g\right\}$ (4.26)

where $ \Xi$$ _{n+1}$ is the algorithmic moduli defined as

$\displaystyle \mbox{\boldmath$\Xi$}$$\displaystyle _{n+1}=\left[\mbox{\boldmath$D$}^{-1}+\sum\lambda^i\partial_{\sig...
...\lambda\partial_{\sigma\kappa}g\partial_{\sigma}\mbox{\boldmath$\kappa$}\right]$ (4.27)

Differentiation of discrete consistency condition yields

$\displaystyle \partial_\sigma f^i d$$\displaystyle \mbox{\boldmath$\sigma$}$$\displaystyle + \partial_\kappa f^i (\partial_\sigma$   $\displaystyle \mbox{\boldmath$\kappa$}$$\displaystyle d$$\displaystyle \mbox{\boldmath$\sigma$}$$\displaystyle + \partial_\lambda$   $\displaystyle \mbox{\boldmath$\kappa$}$$\displaystyle d$$\displaystyle \mbox{\boldmath$\lambda$}$$\displaystyle ) = 0$ (4.28)

By substitution of (4.26) into (4.28) the following relation is obtained

$\displaystyle d$$\displaystyle \mbox{\boldmath$\lambda$}$$\displaystyle =$   $\displaystyle \mbox{\boldmath$G$}$$\displaystyle \left\{\mbox{\boldmath$V$}\mbox{\boldmath$\Xi$}d\mbox{\boldmath$\varepsilon$}\right\}$ (4.29)

where matrix $ G$ is defined as

$\displaystyle \mbox{\boldmath$G$}$$\displaystyle =\left[\mbox{\boldmath$V$}^T\mbox{\boldmath$\Xi$}\mbox{\boldmath$...
...kappa}\mbox{\boldmath$f$}\partial_{\lambda}\mbox{\boldmath$\kappa$}\right]^{-1}$ (4.30)

Finally, by substitution of (4.30) into (4.26) one obtains the algorithmic elastoplastic tangent moduli

$\displaystyle \mbox{$\displaystyle\frac{\rm {d}\mbox{\boldmath$\sigma$}}{\rm {d}\mbox{\boldmath$\varepsilon$}}$}$$\displaystyle \vert_{n+1}=$$\displaystyle \mbox{\boldmath$\Xi$}$$\displaystyle -$$\displaystyle \mbox{\boldmath$\Xi$}$$\displaystyle \mbox{\boldmath$U$}$$\displaystyle \left(\mbox{\boldmath$V$}\mbox{\boldmath$\Xi$}\mbox{\boldmath$U$}...
...ambda}\mbox{\boldmath$\kappa$}]\right) \mbox{\boldmath$V$}\mbox{\boldmath$\Xi$}$ (4.31)

Borek Patzak 2017-12-30