F-bar method is a element technique used in linear elements to alleviate volumetric locking. This method replaces F of each integration point with F-bar.
Index notation
Displacement interpolation
u_i=N_{iK}d_k
Deformation gradient is calculated as
F_{iA}=\delta_{iA}+N_{iK,A}d_K=\delta_{iA}+B_{iKA}d_K
where d_K
is nodal displacement
Total lagrangian shceme without traction and body force is
\delta U=\int P_{iA}\delta F_{iA}dV
and nodal force
R_K=\int P_{iA}B_{iKA}dV
F-bar method: decomposition
\bar{F}_{iA}=(J_0/J)^{1/3}F_{iA}
where J_0
is determinant of the centriod integration point
Replace all F with F-bar in constitutive law
\bar{P}_{iA}=P_{iA}(\bar{F})
Linearization
K_{UU}=\frac{\partial R_K}{\partial d_M}=\int\frac{\partial\bar{P}_{iA}}{\partial\bar{F}_{jB}}\frac{\partial\bar{F}_{jB}}{\partial d_M}dV
where
\begin{align*}
\frac{\partial\bar{F}_{jB}}{\partial d_M}&=\frac{\partial}{\partial d_M}\left(\frac{J_0}{J}\right)^{1/3}F_{jB}+\left(\frac{J_0}{J}\right)^{1/3}\frac{\partial F_{jB}}{\partial d_M}\\
&=\frac13\left(\frac{J_0}{J}\right)^{-2/3}\left[J^{-1}\frac{\partial J_0}{\partial d_M}-\frac{J_0}{J^2}\frac{\partial J}{\partial d_M}\right]F_{jB}+\left(\frac{J_0}{J}\right)^{1/3}B_{jMB}
\end{align*}
and
\frac{\partial J_0}{\partial d_M}=\frac{\partial J_0}{\partial(F_0)_{kC}}\frac{\partial(F_0)_{kC}}{\partial d_M}=J_0(F^{-1}_0)_{Ck}(B_0)_{kMC}
and similarly
\frac{\partial J}{\partial d_M}=JF^{-1}_{Ck}B_{kMC}
now we will have
\frac{\partial\bar{F}_{jB}}{\partial d_M}=\frac13\left(\frac{J_0}{J}\right)^{1/3}[(F^{-1}_0)_{Ck}(B_0)_{kMC}-F^{-1}_{Ck}B_{kMC}]F_{jB}+\left(\frac{J_0}{J}\right)^{1/3}N_{jMB}
Now we will have consistent tangent
K^{UU}_{KM}=\int\bar{C}_{iAjB}B_{iKA}\left\{\frac13[(F^{-1}_0)_{Ck}(B_0)_{kMC}-F^{-1}_{Ck}B_{kMC}]F_{jB}+B_{jMB}\right\}dV
where
\bar{C}_{iAjB}=(J_0/J)^{1/3}\frac{\partial\bar{P}_{iA}}{\partial\bar{F}_{jB}}
ugly huh?
Matrix form
Let’s define an operator to condense 2 indices in one specific order
O[a_{ij}]=\{a\}
Suppose we’re using row-order:
O[a_{ij}]=O\begin{bmatrix}
a_{11}&a_{12}&a_{13}\\
a_{21}&a_{22}&a_{23}\\
a_{31}&a_{32}&a_{33}\\
\end{bmatrix}\rightarrow\begin{Bmatrix}
a_{11}\\a_{12}\\a_{13}\\a_{21}\\a_{22}\\a_{23}\\a_{31}\\a_{32}\\a_{33}\\
\end{Bmatrix}
You can use whatever order you like, just make it consistent
We condense iA,jB
at first
\{u\}=[N]\{d\}
and
\{F\}=[B]\{d\}
now
\frac{\partial\{\bar{F}\}}{\partial\{d\}}=\{F\}\{F^{-1}_0\}^T[B_0]-\{F^{-1}\}^T[B]=[\bar{B}]
Note: \{F^{-1}\}=Q[F^{-T}]
So
K_{UU}=\sum_i[\bar{B}]^T[\bar{C}][\bar{B}]^T(\xi_i)w_i
Here Gaussian integration is applied
Box for F-bar in Q4
- read coord. and disp.
- calculate
F_0
, the deforamtion gradient at central intpt.\xi=0
- Loop over 4 intpt
- read
\xi_i
, calculateF(\xi), [B(\xi)], [N(\xi)]
- calculate
\bar{F}=(J_0/J)^{1/3}F
- calculate stress and tangent
P(\bar{F}),[\bar{C}],[\bar{B}]
- calculate residual contribution
r_i=[N]^T\bar{P}(\xi_i)w_i
- calculate stiffness contribution
[K]_i=[B]^T[\bar{C}][\bar{B}]w_i
- do
[K]=[K]+[K]_i
and\{r\}=\{r\}+\{r\}_i
- read
Leave a Reply