# Continuum mechanics/Strains and deformations: Wikis

Note: Many of our articles have direct quotes from sources you can cite, within the Wikipedia article! This article doesn't yet, but we're working on it! See more info or our list of citable articles.

# Study guide

Up to date as of January 14, 2010

## Strain Measures in three dimensions

 The motion of a body

Initial orthonormal basis:

$(\boldsymbol{E}_1, \boldsymbol{E}_2, \boldsymbol{E}_3)$

Deformed orthonormal basis:

$(\mathbf{e}_1, \mathbf{e}_2, \mathbf{e}_3)$

We assume that these coincide.

### Motion

$\mathbf{x} = \boldsymbol{\varphi}(\mathbf{X}, t) = \mathbf{x}(\mathbf{X}, t)$

\begin{align} \boldsymbol{F} & = \frac{\partial \boldsymbol{\varphi}}{\partial \mathbf{X}} = \boldsymbol{\nabla}_o \boldsymbol{\varphi} \ & = \frac{\partial \mathbf{x}}{\partial \mathbf{X}} = \boldsymbol{\nabla}_X \boldsymbol{\varphi} \end{align}

Effect of $\boldsymbol{F}$:

\begin{align} d\mathbf{x}_1 & = \boldsymbol{F}\bullet d\mathbf{X}_1 ~;& d\mathbf{x}_2 & = \boldsymbol{F}\bullet d\mathbf{X}_2 \end{align}

$\boldsymbol{F} = F_{iJ}~\mathbf{e}_i\otimes\boldsymbol{E}_J$

Index notation:

$F_{iJ} = \frac{\partial x_i}{\partial X_J}$

The determinant of the deformation gradient is usually denoted by J and is a measure of the change in volume, i.e.,

$J = \det{\boldsymbol{F}}$

#### Push Forward and Pull Back

Forward Map:

$\mathbf{x} = \boldsymbol{\varphi}(\mathbf{X},t)$

$\boldsymbol{F} = \frac{\partial \mathbf{x}}{\partial \mathbf{X}} = \boldsymbol{\nabla}_o \boldsymbol{\varphi}$

$\boldsymbol{F} = \sum_{i,J=1}^3 \frac{\partial x_i}{\partial X_J}~\mathbf{e}_i\otimes\boldsymbol{E}_J$

$d\mathbf{x} = \boldsymbol{F}\bullet d\mathbf{X} = \boldsymbol{\varphi}_{*}[d\mathbf{X}]$

Push Forward operation:

$\boldsymbol{\varphi}_{*}[\bullet]$
• $d\mathbf{X}$ = material vector.
• $d\mathbf{x}$ = spatial vector.

Inverse map:

$\mathbf{X} = \boldsymbol{\varphi}^{-1}(\mathbf{x},t)$

$\boldsymbol{F}^{-1} = \frac{\partial \mathbf{X}}{\partial \mathbf{x}} = \boldsymbol{\nabla} \boldsymbol{\varphi}^{-1}$

$\boldsymbol{F}^{-1} = \sum_{i,J=1}^3 \frac{\partial X_I}{\partial x_j}~\boldsymbol{E}_I\otimes\mathbf{e}_j$

$d\mathbf{X} = \boldsymbol{F}^{-1}\bullet d\mathbf{x} = \boldsymbol{\varphi}^{*}[d\mathbf{x}]$

Pull Back operation:

$\boldsymbol{\varphi}^{*}[\bullet]$
• $d\mathbf{X}$ = material vector.
• $d\mathbf{x}$ = spatial vector.
##### Example
 Push forward and pull back

Motion:

\begin{align} x_1 & = \cfrac{1}{4}(18 + 4X_1 + 6X_2) \ x_2 & = \cfrac{1}{4}(14 + 6X_2) \end{align}

$F_{ij} = \frac{\partial x_i}{\partial X_j} \implies \mathbf{F} = \frac{1}{2}\begin{bmatrix} 2 & 3 \\ 0 & 3 \end{bmatrix}$

$\mathbf{F}^{-1} = \cfrac{1}{3}\begin{bmatrix} 3 & -3 \\ 0 & 2 \end{bmatrix}$

Push Forward:

\begin{align} \boldsymbol{\varphi}_{*}[\boldsymbol{E}_1] & = \mathbf{F}\begin{bmatrix} 1 \\ 0 \end{bmatrix} = \begin{bmatrix} 1 \\ 0 \end{bmatrix} \ \boldsymbol{\varphi}_{*}[\boldsymbol{E}_2] & = \mathbf{F}\begin{bmatrix} 0 \\ 1 \end{bmatrix} = \begin{bmatrix} 1.5 \\ 1.5 \end{bmatrix} \end{align}

Pull Back:

\begin{align} \boldsymbol{\varphi}^{*}[\mathbf{e}_1] & = \mathbf{F}^{-1}\begin{bmatrix} 1 \\ 0 \end{bmatrix} = \begin{bmatrix} 1 \\ 0 \end{bmatrix} \ \boldsymbol{\varphi}^{*}[\mathbf{e}_2] & = \mathbf{F}^{-1}\begin{bmatrix} 0 \\ 1 \end{bmatrix} = \begin{bmatrix} -1 \\ 2/3 \end{bmatrix} \end{align}

### Cauchy-Green Deformation Tensors

#### Right Cauchy-Green Deformation Tensor

Recall:

$d\mathbf{x}_1 = \boldsymbol{F}\bullet d\mathbf{X}_1 ~;~~ d\mathbf{x}_2 = \boldsymbol{F}\bullet d\mathbf{X}_2$

Therefore,

$d\mathbf{x}_1\bullet d\mathbf{x}_2 = (\boldsymbol{F}\bullet d\mathbf{X}_1)\bullet(\boldsymbol{F}\bullet d\mathbf{X}_2)$

Using index notation:

\begin{align} d\mathbf{x}_1\bullet d\mathbf{x}_2 & = (F_{ij}~dX^1_j)(F_{ik}~dX^2_k)\ & = dX^1_j~(F_{ij}~F_{ik})~dX^2_k \ & = d\mathbf{X}_1\bullet(\boldsymbol{F}^T\bullet\boldsymbol{F})\bullet d\mathbf{X}_2 \ & = d\mathbf{X}_1\bullet\boldsymbol{C} \bullet d\mathbf{X}_2 \end{align}

Right Cauchy-Green tensor:

$\boldsymbol{C} = \boldsymbol{F}^T\bullet\boldsymbol{F}$

#### Left Cauchy-Green Deformation Tensor

Recall:

$d\mathbf{X}_1 = \boldsymbol{F}^{-1}\bullet d\mathbf{x}_1 ~;~~ d\mathbf{X}_2 = \boldsymbol{F}^{-1}\bullet d\mathbf{x}_2$

Therefore,

$d\mathbf{X}_1\bullet d\mathbf{X}_2 = (\boldsymbol{F}^{-1}\bullet d\mathbf{x}_1)\bullet(\boldsymbol{F}^{-1}\bullet d\mathbf{x}_2)$

Using index notation:

\begin{align} d\mathbf{X}_1\bullet d\mathbf{X}_2 & = (F^{-1}_{ij}~dx^1_j)(F^{-1}_{ik}~dx^2_k)\ & = dx^1_j~(F^{-1}_{ij}~F^{-1}_{ik})~dx^2_k \ & = d\mathbf{x}_1\bullet(\boldsymbol{F}^{-T}\bullet\boldsymbol{F}^{-1})\bullet d\mathbf{x}_2 \ & = d\mathbf{x}_1\bullet(\boldsymbol{F}\bullet\boldsymbol{F}^T)^{-1}\bullet d\mathbf{x}_2 \ & = d\mathbf{x}_1\bullet\mathbf{b}^{-1}\bullet d\mathbf{x}_2 \end{align}

Left Cauchy-Green (Finger) tensor:

$\mathbf{b} = \boldsymbol{F}\bullet\boldsymbol{F}^T$

### Strain Measures

#### Green (Lagrangian) Strain

\begin{align} \frac{1}{2}(d\mathbf{x}_1\bullet d\mathbf{x}_2 & - d\mathbf{X}_1\bullet d\mathbf{X}_2) \ & = \frac{1}{2}d\mathbf{X}_1\bullet(\boldsymbol{C} - \boldsymbol{I}) \bullet d\mathbf{X}_2 \ & = d\mathbf{X}_1\bullet\boldsymbol{E} \bullet d\mathbf{X}_2 \end{align}

Green strain tensor:

\begin{align} \boldsymbol{E} & = \frac{1}{2}(\boldsymbol{C} - \boldsymbol{I}) \ & = \frac{1}{2}(\boldsymbol{F}^T\bullet\boldsymbol{F} - \boldsymbol{I}) \ & = \frac{1}{2}\left[\boldsymbol{\nabla}_o \mathbf{u} + (\boldsymbol{\nabla}_o \mathbf{u})^T + \boldsymbol{\nabla}_o \mathbf{u}\bullet(\boldsymbol{\nabla_o \mathbf{u})^T}\right] \end{align}

Index notation:

\begin{align} E_{ij} & = \frac{1}{2}(F_{ki}~F_{kj} - \delta_{ij})\ & = \frac{1}{2}\left(\frac{\partial u_i}{\partial X_j} + \frac{\partial u_j}{\partial X_i} + \frac{\partial u_k}{\partial X_i}\frac{\partial u_k}{\partial X_j}\right) \end{align}

#### Almansi (Eulerian) Strain

\begin{align} \frac{1}{2}(d\mathbf{x}_1\bullet d\mathbf{x}_2 & - d\mathbf{X}_1\bullet d\mathbf{X}_2) \ & = \frac{1}{2}d\mathbf{x}_1\bullet(\boldsymbol{I} - \mathbf{b}^{-1})\bullet d\mathbf{x}_2 \ & = d\mathbf{x}_1\bullet\mathbf{e} \bullet d\mathbf{x}_2 \end{align}

Almansi strain tensor:

\begin{align} \mathbf{e} & = \frac{1}{2}(\boldsymbol{I} - \mathbf{b}^{-1}) \ & = \frac{1}{2}(\boldsymbol{I} - \boldsymbol{F}^{-T}\bullet\boldsymbol{F}^{-1}) \end{align}

Index notation:

\begin{align} e_{ij} & = \frac{1}{2}(\delta_{ij} - F^{-1}_{ki}~F^{-1}_{kj}) \end{align}

#### Push Forward and Pull Back

Recall:

$d\mathbf{x}_1\bullet\mathbf{e} \bullet d\mathbf{x}_2 = d\mathbf{X}_1\bullet\boldsymbol{E} \bullet d\mathbf{X}_2$

Now,

\begin{align} d\mathbf{x}_1\bullet\mathbf{e} \bullet d\mathbf{x}_2 & =(\boldsymbol{F}\bullet d\mathbf{X}_1)\bullet\mathbf{e}\bullet(\boldsymbol{F}\bullet d\mathbf{X}_2) \ & =d\mathbf{X}_1\bullet(\boldsymbol{F}^T\bullet\mathbf{e}\bullet\boldsymbol{F})\bullet d\mathbf{X}_2 \ & =d\mathbf{X}_1\bullet\boldsymbol{E}\bullet d\mathbf{X}_2 \end{align}

Therefore,

\begin{align} \boldsymbol{E} & = \boldsymbol{F}^T\bullet\mathbf{e}\bullet\boldsymbol{F} \ \implies \mathbf{e} & = \boldsymbol{F}^{-T}\bullet\boldsymbol{E}\bullet\boldsymbol{F}^{-1} \end{align}

Push Forward:

$\mathbf{e} = \boldsymbol{\varphi}_{*}[\boldsymbol{E}] =\boldsymbol{F}^{-T}\bullet\boldsymbol{E}\bullet\boldsymbol{F}^{-1}$

Pull Back:

$\boldsymbol{E} = \boldsymbol{\varphi}^{*}[\mathbf{e}] =\boldsymbol{F}^T\bullet\mathbf{e}\bullet\boldsymbol{F}$

### Some useful results

#### Derivative of J with respect to the deformation gradient

We often need to compute the derivative of $J = \det{\boldsymbol{F}}$ with respect the the deformation gradient $\boldsymbol{F}$. From tensor calculus we have, for any second order tensor $\boldsymbol{A}$

$\cfrac{\partial}{\partial \boldsymbol{A}}( \det{\boldsymbol{A}}) = \det{\boldsymbol{A}}~\boldsymbol{A}^{-T}$

Therefore,

 $\cfrac{\partial J}{\partial \boldsymbol{F}} = J~\boldsymbol{F}^{-T}$

#### Derivative of J with respect to the right Cauchy-Green deformation tensor

The derivative of J with respect to the right Cauchy-Green deformation tensor ($\boldsymbol{C}$) is also often encountered in continuum mechanics.

To calculate the derivative of $J = \det{\boldsymbol{F}}$ with respect to $\boldsymbol{C}$, we recall that (for any second order tensor $\boldsymbol{T}$)

$\frac{\partial \boldsymbol{C}}{\partial \boldsymbol{F}}:\boldsymbol{T} = \frac{\partial }{\partial \boldsymbol{F}}(\boldsymbol{F}^T\cdot\boldsymbol{F}):\boldsymbol{T} = (\boldsymbol{\mathsf{I}}^T:\boldsymbol{T})\cdot\boldsymbol{F} + \boldsymbol{F}^T\cdot(\boldsymbol{\mathsf{I}}:\boldsymbol{T}) = \boldsymbol{T}^T\cdot\boldsymbol{F} + \boldsymbol{F}^T\cdot\boldsymbol{T}$

Also,

$\frac{\partial J}{\partial \boldsymbol{F}}:\boldsymbol{T} = \frac{\partial J}{\partial \boldsymbol{C}}:(\frac{\partial \boldsymbol{C}}{\partial \boldsymbol{F}}:\boldsymbol{T}) = \frac{\partial J}{\partial \boldsymbol{C}}:(\boldsymbol{T}^T\cdot\boldsymbol{F} + \boldsymbol{F}^T\cdot\boldsymbol{T}) = \left[\boldsymbol{F}\cdot\frac{\partial J}{\partial \boldsymbol{C}}\right]:\boldsymbol{T} + \left[\boldsymbol{F}\cdot\left(\frac{\partial J}{\partial \boldsymbol{C}}\right)^T\right]:\boldsymbol{T}$

From the symmetry of $\boldsymbol{C}$ we have

$\frac{\partial J}{\partial \boldsymbol{C}} = \left(\frac{\partial J}{\partial \boldsymbol{C}}\right)^T$

Therefore, involving the arbitrariness of $\boldsymbol{T}$, we have

$\frac{\partial J}{\partial \boldsymbol{F}} = 2~\boldsymbol{F}\cdot\frac{\partial J}{\partial \boldsymbol{C}}$

Hence,

$\frac{\partial J}{\partial \boldsymbol{C}} = \frac{1}{2}~\boldsymbol{F}^{-1}\cdot\frac{\partial J}{\partial \boldsymbol{F}} ~.$

Also recall that

$\frac{\partial J}{\partial \boldsymbol{F}} = J~\boldsymbol{F}^{-T}$

Therefore,

 $\frac{\partial J}{\partial \boldsymbol{C}} = \frac{1}{2}~J~\boldsymbol{F}^{-1}\cdot\boldsymbol{F}^{-T} = \cfrac{J}{2}~\boldsymbol{C}^{-1}$

In index notation,

 $\frac{\partial J}{\partial C_{IJ}} = \cfrac{J}{2}~C^{-1}_{IJ}$

#### Derivative of the inverse of the right Cauchy-Green tensor

Another result that is often useful is that for the derivative of the inverse of the right Cauchy-Green tensor ($\boldsymbol{C}$).

Recall that, for a second order tensor $\boldsymbol{A}$,

$\frac{\partial \boldsymbol{A}^{-1}}{\partial \boldsymbol{A}}:\boldsymbol{T} = -\boldsymbol{A}^{-1}\cdot\boldsymbol{T}\cdot\boldsymbol{A}^{-1}$

In index notation

$\frac{\partial A^{-1}_{ij}}{\partial A_{kl}}~T_{kl} = B_{ijkl}~T_{kl} = -A^{-1}_{ik}~T_{kl}~A^{-1}_{lj}$

or,

$\frac{\partial A^{-1}_{ij}}{\partial A_{kl}} = B_{ijkl} = -A^{-1}_{ik}~A^{-1}_{lj}$

Using this formula and noting that since $\boldsymbol{C}$ is a symmetric second order tensor, the derivative of its inverse is a symmetric fourth order tensor we have

 $\frac{\partial C^{-1}_{IJ}}{\partial C_{KL}} = -\frac{1}{2}~(C^{-1}_{IK}~C^{-1}_{JL} + C^{-1}_{JK}~C^{-1}_{IL})$