1% Nonlinear mechanical behaviours of materials 2% Thomas Helfer 3% 2017 4 5\newcommand{\tenseur}[1]{\underline{#1}} 6\newcommand{\tenseurq}[1]{\underset{=}{\mathbf{#1}}} 7<!-- the previous works better than the standard solution: --> 8<!-- \newcommand{\tenseurq}[1]{\underline{\underline{{\mathbf{#1}}}}} --> 9\newcommand{\tns}[1]{{\underset{\tilde{}}{\mathbf{#1}}}} 10\newcommand{\transpose}[1]{{#1^{\mathop{T}}}} 11 12\newcommand{\paren}[1]{\left(#1\right)} 13\newcommand{\trace}[1]{\mathrm{tr}\left(#1\right)} 14\newcommand{\tsigma}{\underline{\sigma}} 15 16\newcommand{\Frac}[2]{{{\displaystyle \frac{\displaystyle #1}{\displaystyle #2}}}} 17\newcommand{\deriv}[2]{{\displaystyle \frac{\displaystyle \partial #1}{\displaystyle \partial #2}}} 18\newcommand{\sderiv}[2]{{\displaystyle \frac{\displaystyle \partial^{2} #1}{\displaystyle \partial #2^{2}}}} 19\newcommand{\dtot}{{{\mathrm{d}}}} 20\newcommand{\derivtot}[2]{{\displaystyle \frac{\displaystyle \dtot #1}{\displaystyle \dtot #2}}} 21\newcommand{\grad}[1]{{\displaystyle \overset{\longrightarrow}{\nabla} #1}} 22\newcommand{\Grad}[1]{\mathop{\mathrm{Grad}\,#1}} 23\newcommand{\diver}[1]{{\displaystyle \vec{\nabla} . #1}} 24\newcommand{\divergence}{\mathop{\mathrm{div}}} 25\newcommand{\Divergence}{\mathop{\mathrm{Div}}} 26 27\newcommand{\bts}[1]{\left.#1\right|_{t}} 28\newcommand{\mts}[1]{\left.#1\right|_{t+\theta\,\Delta\,t}} 29\newcommand{\ets}[1]{\left.#1\right|_{t+\Delta\,t}} 30 31\newcommand{\epsilonto}{\epsilon^{\mathrm{to}}} 32\newcommand{\tepsilonto}{\underline{\epsilon}^{\mathrm{to}}} 33\newcommand{\tdepsilonto}{\underline{\dot{\epsilon}}^{\mathrm{tox}}} 34\newcommand{\tepsilonel}{\underline{\epsilon}^{\mathrm{el}}} 35\newcommand{\tdepsilonel}{\underline{\dot{\epsilon}}^{\mathrm{el}}} 36\newcommand{\tepsilonth}{\underline{\epsilon}^{\mathrm{th}}} 37\newcommand{\epsilonvis}{\epsilon^{\mathrm{vis}}} 38\newcommand{\tepsilonvis}{\underline{\epsilon}^{\mathrm{vis}}} 39\newcommand{\depsilonvis}{\dot{\epsilon}^{\mathrm{vis}}} 40\newcommand{\tdepsilonvis}{\underline{\dot{\epsilon}}^{\mathrm{vis}}} 41\newcommand{\tepsilonp}{\underline{\epsilon}^{\mathrm{p}}} 42\newcommand{\tdepsilonp}{\underline{\dot{\epsilon}}^{\mathrm{p}}} 43 44\newcommand{\energieinterne}{e} 45\newcommand{\energielibre}{\Psi} 46\newcommand{\energielibreel}{{\Psi}^{el}} 47\newcommand{\energielibreine}{{\Psi}^{inél}} 48\newcommand{\energielibreduale}{\Psi^{\star}} 49\newcommand{\potentieldissip}{\Phi} 50\newcommand{\potentieldissipdual}{\Phi^{\star}} 51 52\newcommand{\discret}[1]{\mathbb{#1}} 53\newcommand{\residuEF}{\discret{\vec{R}}} 54\newcommand{\forceintEF}{\discret{\vec{F}}_{i}} 55\newcommand{\forceextEF}{\discret{\vec{F}}_{e}} 56\newcommand{\forceintElem}{\discret{\vec{F}}_{i}^{e}} 57\newcommand{\champEF}{\discret{\vec{v}}^{h}} 58 59 60# Preliminary mathematical results 61 62## Tensors 63 64Second order tensors can be represented by matrices: 65\[ 66 \tns{F}= 67 \begin{pmatrix} 68 F_{11} & F_{12} & F_{13} \\ 69 F_{21} & F_{22} & F_{23} \\ 70 F_{31} & F_{32} & F_{33} \\ 71 \end{pmatrix} 72 \] 73 74In `TFEL/MFront`, a tensor is stored as an array of values, as follows 75in \(3D\): 76\[ 77 \tenseur{s}= 78 \begin{pmatrix} 79 s_{\,11}\quad 80 s_{\,22}\quad 81 s_{\,33}\quad 82 s_{\,12}\quad 83 s_{\,21}\quad 84 s_{\,13}\quad 85 s_{\,31}\quad 86 s_{\,23}\quad 87 s_{\,32} 88 \end{pmatrix}^{T} 89\] 90 91The trace of a tensor is defined by: 92 93\[ 94 \trace{\tns{F}}=\sum_{i=0}^{3}F_{ii} 95\] 96 97The products of two tensors is defined by: 98 99\[ 100 \paren{\tns{F}\,.\,\tns{G}}_{ij}=\sum_{k=0}^{3}F_{ik}\,G_{kj}\neq\paren{\tns{G}\,.\,\tns{F}}_{ij} 101 \] 102 103The contracted product of two tensors defines a scalar product: 104 105\[ 106\tns{F}\,\colon\,\tns{\pi}=\trace{\tns{F}\,.\,\tns{\pi}}=\sum_{i=0}^{3}\sum_{j=0}^{3}F_{ij}\,\pi_{ji}=F_{ij}\,\pi_{ji} 107\] 108 109## Symmetric tensors 110 111A symmetric tensor is stored as an array of values (vector notation), 112as follows in \(3D\): 113 114\[ 115\tenseur{s}= 116\begin{pmatrix} 117 s_{\,11}\quad 118 s_{\,22}\quad 119 s_{\,33}\quad 120 \sqrt{2}\,s_{\,12}\quad 121 \sqrt{2}\,s_{\,13}\quad 122 \sqrt{2}\,s_{\,23} 123\end{pmatrix}^{T} 124\] 125 126The contracted product of two symmetric tensors is the scalar product 127 of their vector forms (hence the \(\sqrt{2}\)). 128 129### Diagonalisation of a symmetric tensor 130 131Symmetric tensors are diagonalisable: 132\[ 133\tenseur{s}=\sum_{i=0}^{3}\lambda_{i}\,\tenseur{n}_{i} 134\] 135 136- \(\lambda_{i}\) are the eigenvalues of the tensor. 137- The eigentensors \(\tenseur{n}_{i}\) are orthogonals. 138 139### Isotropic functions of symmetric tensors 140 141This decomposition allows the definition of isotropic functions of 142tensors (logarithm, exponential, square root, ..): 143 144\[ 145\tepsilonto_{\mathrm{log}}=\sum_{i=0}^{3}f\paren{\lambda_{i}}\,\tenseur{n}_{i} 146\] 147 148# Kinematics 149 150## Deformation of a body 151 152![Initial and deformed configurations](img/Configuration.png "Initial and deformed configurations"){width=75%} 153 154- at a given time \(t\), a structure is characterised by its 155 {\em configuration} \(\mathcal{C}_{t}\); 156- the initial configuration \(\mathcal{C}_{0}\) is often taken 157 as the reference configuration 158- the motion is described by a family of mappings \(\phi_{t}\), 159 the {\bf deformation}, which associates a point \(\vec{X}\) in 160 \(\mathcal{C}_{0}\) to its position \(\vec{x}\) in 161 \(\mathcal{C}_{t}\): 162 \[ 163 \vec{x}=\phi_{t}\paren{\vec{X}}=\vec{X}+\vec{u}_{t}\paren{\vec{X}} 164 \] 165- \(\vec{u}_{t}\) is the {\bf displacement field} 166 167### Deformation gradient 168 169- Locally, the material is mechanically loaded if the 170 deformation of the current point differs from the deformation of 171 its neighbours 172- The \nom{Taylor} expansion of the deformation leads to: 173 \[ 174 \phi\paren{\vec{X}+\dtot\,\vec{X}}-\phi\paren{\vec{X}}=\deriv{\phi}{\vec{X}}\,.\dtot\,\vec{X}+\underbrace{\vec{X}\,.\,\Frac{\partial^{2}\,\phi}{\partial\,\vec{X}^{2}}\,.\dtot\,\vec{X}+\ldots}_{\text{Higher order terms}} 175 \] 176- The deformation gradient \(\tns{F}\) is a tensor defined by: 177 \[ 178 \tns{F}=\deriv{\phi}{\vec{X}}=\tns{I}+\deriv{\vec{u}}{\vec{X}} 179 \] 180- Volume change: 181 \[ 182 J=\det\paren{\deriv{\vec{x}}{\vec{X}}}=\det{\tns{F}} 183 \] 184 185Higher order theories will not be considered in this course in this 186paper. 187 188 189### Polar decomposition of $\tns{F}$ 190 191![Polar decomposition (Wikipedia)](img/Polar_decomposition_of_F.png "Polar decomposition (Wikipedia)"){width=75%} 192 193- Unicity and existence of a polar decomposition of \(\tns{F}\): 194 \( 195 \quad\quad\tns{F}=\tns{R}\,.\tenseur{U}=\tenseur{V}\,.\tns{R} 196 \) 197- \(\tns{R}\) is a rotation 198- \(\tenseur{V}\) and \(\tenseur{U}\) are {\bf symmetric tensors} 199- \(\tenseur{V}\) is the eulerian stretch tensor: 200 - \(\bts{\tenseur{V}}\) and \(\ets{\tenseur{V}}\) are 201 expressed in two different configuration: 202 - \(\Delta\,\tenseur{V}=\ets{\tenseur{V}}-\bts{\tenseur{V}}\) 203 is {\bf not} well defined 204 - \(\tenseur{U}\) is the lagrangian stretch tensor: 205 - expressed in the reference configuration 206 - \(\Delta\,\tenseur{U}=\ets{\tenseur{U}}-\bts{\tenseur{U}}\) 207 is {\bf well defined} 208 - Volume change: 209 \[ 210 J=\det\,\tns{F}=\det\,\tenseur{U} 211 \] 212 213### Pure dilatation 214 215- Pure dilatations correspond to diagonal deformation gradients: 216 \[ 217 \tns{F}=\tenseur{U}= 218 \begin{pmatrix} 219 F_{11} & 0 & 0 \\ 220 0 & F_{22} & 0 \\ 221 0 & 0 & F_{33} \\ 222 \end{pmatrix} 223 \] 224- Example of thermal expansion. 225 226### Rate of deformation 227 228- velocity of a point: 229 \[ 230 \vec{v}=\derivtot{\vec{x}}{t} 231 \] 232- gradient velocity of a point: 233 \[ 234 \tns{L}=\deriv{\vec{\dot{x}}}{\vec{x}}=\deriv{\vec{\dot{x}}}{\vec{X}}\,.\,\deriv{\vec{X}}{\vec{x}}=\paren{\derivtot{}{t}\deriv{\vec{x}}{\vec{X}}}\,.\,\deriv{\vec{X}}{\vec{x}}=\tns{\derivtot{F}{t}}\,.\,\tns{F}^{-1} 235 \] 236- rate of deformation: 237 \[ 238 \tenseur{D}=\Frac{1}{2}\left(\tns{L}+\transpose{\tns{L}}\right) 239 \] 240- rate of : 241 \[ 242 \dot{J} = \trace{\tenseur{D}} 243 \] 244- rotation rate: 245 \[ 246 \tns{\omega}=\Frac{1}{2}\left(\tns{L}-\transpose{\tns{L}}\right) 247 \] 248 249### Rate of deformation in pure dilatations 250 251- for pure dilatations, we have: 252 \[ 253 \tenseur{D}= 254 \begin{pmatrix} 255 \frac{\dot{F}_{11}}{F_{11}} & 0 & 0 \\ 256 0 & \frac{\dot{F}_{22}}{F_{22}} & 0 \\ 257 0 & 0 & \frac{\dot{F}_{33}}{F_{33}} \\ 258 \end{pmatrix}= 259 \begin{pmatrix} 260 \frac{\dot{l}_{1}}{l_{1}\paren{t}} & 0 & 0 \\ 261 0 & \frac{\dot{l}_{2}}{l_{2}\paren{t}} & 0 \\ 262 0 & 0 & \frac{\dot{l}_{3}}{l_{3}\paren{t}} \\ 263 \end{pmatrix} 264 \] 265- \(\displaystyle\int_{0}^{t}D_{11}\,\dtot\,t=\int_{l_{0}}^{l\paren{t}}\Frac{\dtot\,l}{l}=\log\paren{\Frac{l\paren{t}}{l_{0}}}\) 266- \(\displaystyle\log\paren{\Frac{l\paren{t}}{l_{0}}}\) is 267 sometimes calls the true strain, altough the meaning of this is 268 dubious, as discussed later 269- The expression 270 \(\displaystyle\int_{l_{0}}^{l\paren{t}}\Frac{\dtot\,l}{l}\) is 271 sometimes used to justify an incremental framework of the 272 mechanics of deformable body at finite strain (updated lagrangian 273 formulations, hypoelasticity): it's a common pitfall of the 274 \(80's\) which is still present in major finite element solvers. 275 276# Mechanical equilibrium 277 278## Cauchy stress 279 280![Components of the Cauchy stress (Wikipedia)](img/CauchyStressComponents.png "Components of the Cauchy stress (Wikipedia)"){width=75%} 281 282- The stress state of a body at a given point is characterised 283 by a second order tensor called the Cauchy stress: 284 \(\quad\quad\quad\tsigma= 285 \begin{pmatrix} 286 \sigma_{11} & \sigma_{12} & \sigma_{13} \\ 287 \sigma_{21} & \sigma_{22} & \sigma_{23} \\ 288 \sigma_{31} & \sigma_{32} & \sigma_{33} 289 \end{pmatrix} 290 \) 291- The Cauchy stress satisfies:\\ 292 \( 293 \quad\quad\quad\dtot\vec{T} = \tsigma\,.\,\dtot\vec{s} 294 \) 295 - \(\dtot\vec{s}\): oriented unit surface 296 - \(\dtot\vec{T}\): traction acting on \(\dtot\vec{s}\) 297 - \(\dtot\vec{s}\) is defined of the current configuration 298- Example of a pressure \(p\) applied to the boundary of a body, 299 \(\vec{n}\) being the outer normal to the boundary: 300 \[ 301 \tsigma\,.\,\dtot\vec{n}=-p 302 \] 303 304\paragraph{Mechanical equilibrium} 305 306- Linear momentum conservation: 307 \[ 308 \divergence\tsigma+\vec{f}=\rho\,\vec{a}\equiv 309 \left\{ 310 \begin{aligned} 311 \deriv{\sigma_{11}}{x_{1}}+ 312 \deriv{\sigma_{21}}{x_{2}}+ 313 \deriv{\sigma_{31}}{x_{3}}+f_{1}=\rho\,a_{1}\\ 314 \deriv{\sigma_{12}}{x_{1}}+ 315 \deriv{\sigma_{22}}{x_{2}}+ 316 \deriv{\sigma_{32}}{x_{3}}+f_{2}=\rho\,a_{2} \\ 317 \deriv{\sigma_{13}}{x_{1}}+ 318 \deriv{\sigma_{23}}{x_{2}}+ 319 \deriv{\sigma_{33}}{x_{3}}+f_{3}=\rho\,a_{3} \\ 320 \end{aligned} 321 \right. 322 \] 323 with \(\vec{a}\) acceleration, \(\rho\) density, \(\vec{f}\) body forces (gravity) 324- Equilibrium is expressed in the unknown {\bf deformed} configuration: 325 - geometrical {\bf non linearity} of continuum mechanics 326 - Angular momentum conservation: without body momentum, the 327 Cauchy stress is {\bf symmetric} 328 \[ 329 \tsigma=\transpose{\tsigma} 330 \] 331 332## Infinitesimal perturbation theory 333 334- {\bf no rotation}: 335 \[ 336 \tns{R}\approx\tns{I} 337 \] 338- {\bf small strain (first order perturbation)}: 339 \[ 340 \tenseur{U}\approx\tenseur{I}+\tepsilonto 341 \] 342- \(\tepsilonto\) is the {\bf linearised strain tensor}: 343 \[ 344 \epsilonto_{ji}=\Frac{1}{2}\paren{\deriv{u_{i}}{X_{j}}+\deriv{u_{j}}{X_{i}}} 345 \quad\text{or}\quad 346 \tepsilonto=\Frac{1}{2}\paren{\Grad{\vec{u}}+\transpose{\Grad{\vec{u}}}} 347 \] 348- change of volume: 349 \[ 350 J=\det\tns{F}\approx{}=\det\tenseur{U}=1+\trace{\tepsilonto} 351 \] 352- \(\Omega_{t}\) can be replaced by \(\Omega_{0}\): 353 - no more geometrical non linearity 354 - Equilibrium: 355 \[ 356 \Divergence\tsigma+\vec{f}=\rho_{0}\,\vec{a} \quad\text{and}\quad 357 \tsigma=\transpose{\tsigma} 358 \] 359 360# Energy, strain measures, stress measures 361 362## Mechanical power 363 364- Power of body forces: 365 \[ 366 \begin{aligned} 367 \int_{\Omega}\vec{f}\,.\,\vec{v}\,\dtot\,v 368 &=-\int_{\partial\,\Omega}\left(\tsigma\,.\,\vec{n}\right)\,.\,\vec{v}\,\dtot\,s+\int_{\Omega}\tsigma\,\colon\,\tns{L}\,\dtot\,v\\ 369 &=-\int_{\partial\,\Omega}\left(\tsigma\,.\,\vec{n}\right)\,.\,\vec{v}\,\dtot\,s+\int_{\Omega}\tsigma\,\colon\,\tns{D}\,\dtot\,v\quad\text{(symmetry of \(\tsigma\))}\\ 370 \end{aligned} 371 \] 372- Conservation of energy: 373 \[ 374 \underbrace{\int_{\Omega}\tsigma\,\colon\,\tns{D}\,\dtot\,v}_{\text{inner forces power}} = \underbrace{\int_{\Omega}\vec{f}\,.\,\vec{v}\,\dtot\,v + \int_{\partial\,\Omega}\vec{t}\,.\,\vec{v}\,\dtot\,s}_{\text{external forces power}} 375 \] 376 377## Strain measures 378 379- A strain measure must satisfy the following hypotheses: 380 - tends to \(\tepsilonto\) when the infinitesimal strain 381 theory's assumptions are satisfied; 382 - is objective (filters finite body rotation) 383 - is symmetric 384 - many isotropic functions of \(\tenseur{U}\) satisfy those 385 requirements: 386 - Green-Lagrange strain: 387 \[ 388 \tepsilonto_{GL}=\Frac{1}{2}\paren{\tenseur{U}^{2}-\tns{I}}= \Frac{1}{2}\paren{\tns{F}^{T}\,.\,\tns{F}-\tns{I}} 389 \] 390 - Hencky strain: 391 \[\tepsilonto_{\mathrm{log}}=\log\tenseur{U}=\displaystyle\sum_{i=0}^{3}\log\lambda_{i}\,\tenseur{n}_{i}\] 392 where \(\lambda_{i}\) are the eigenvalues of \(\tenseur{U}\) and 393 \(\tenseur{n}_{i}\) are its eigentensors 394 - functions of \(\tns{V}\) will not be considered {\bf here} 395- \(\tepsilonto\) is {\bf not} a strain measure, \(\tns{F}\) is {\bf not} a strain measure 396- \(\tns{D}\) is {\bf not} the time derivative of a strain measure 397 398## Energetic conjugates (stress measures) 399 400- Mechanical work: 401 \[ 402 \int_{\Omega}\tsigma\,\colon\,\tenseur{D}\,\dtot\,v= 403 \int_{\Omega_{0}}\tsigma\,\colon\,\tenseur{D}\,J\,\dtot\,V_{0} 404 \] 405- For each strain measure \(\tepsilonto_{\star}\), one may 406 define its dual stress \(\tns{T}_{\star}\): 407 \[ 408 J\,\tsigma\,\colon\,\tenseur{D}=\tenseur{\tau}\,\colon\,\tenseur{D}=\tepsilonto_{\star}\,\colon\,\tns{T}_{\star} 409 \] 410- The tensor \(\tenseur{\tau}=J\,\tsigma\) is called the 411 Kirchhoff stress. 412- The dual of the Green-Lagrange strain is the second 413 Piola-Kirchhoff stress \(\tenseur{S}\): 414 \[ 415 \tenseur{S}=\tns{F}^{-1}\,.\,\tenseur{\tau}\,.\,\tns{F}^{-T} \quad\Leftrightarrow\quad\tau = \tns{F}\,.\,\tenseur{S}\,.\,\tns{F}^{T} 416 \] 417- There is no strain measure which is the dual of \(\tenseur{\tau}\), nor \(\tsigma\) 418 419## Choice of a stress/strain couple 420 421- All strain measures are {\bf equivalent}: there is no 422 theoretical reason to prefer one strain measure over 423 another. 424- However, there are {\bf pratical} reasons to do so. 425 426 427### The Green-Lagrange strain 428 429- The Green-Lagrange strain and its dual are: 430 - easy to compute (no computational penalty); 431 - there is no straight-forward relation between the 432 Green-Lagrange strain and the change of volume 433 - for small strain however: 434 \[ 435 J \approx 1+\trace{\tepsilonto_{\mathrm{GL}}} 436 \] 437 \begin{center} 438 439* The Green-Lagrange strain framework is indeed well suited for 440extending behaviours identified in the infinitesimal perturbation 441theory to finite rotation* 442 443### The logarithmic strain framework 444 445- There is a straight-forward relation between the 446 Hencky strain strain and the change of volume: 447 \[ 448 J = \exp\paren{\trace{\tepsilonto_{\mathrm{log}}}} 449 \] 450- For pure dilatations: 451 - \(\tepsilonto_{\mathrm{log}}=\displaystyle\int_{0}^{t}\tenseur{D}\,\dtot\,t\) 452 - \(\tenseur{T}_{\mathrm{log}}\) is equal to the Kirchhoff 453 stress \(\tenseur{\tau}\), which is egal to the Cauchy stress 454 \(\tsigma\) for isochoric deformation 455 - However, the Hencky strain and its dual are costly to compute 456 (see Miehe et al., 2005); 457 458** The logarithmic strain framework is indeed well suited for 459 plasticity and/or viscoplasticity ** 460 461# Elasticity 462 463- The mechanical work during the deformation process is 464 (time integral of the mechanical power): 465 \[ 466 w\paren{0,t}=\int_{0}^{t}J\,\tsigma\,\colon\,\tenseur{D}\,\dtot\,t 467 =\int_{0}^{t}\tenseur{T}_{\star}\,\colon\,\tepsilonto_{\star}\,\dtot\,t 468 =\int_{\left.\tepsilonto_{\star}\right|0}^{\left.\tepsilonto_{\star}\right|t}\tenseur{T}_{\star}\,\colon\,\dtot\,\tepsilonto_{\star} 469 \] 470- Elasticity assumes that the previous integral is 471 path-independent, i.e. the mechanical work only depends on the 472 current state of deformation and not the history of the 473 deformation process 474- The dual stress \(\tenseur{T}_{\star}\) then satisfies : 475 \[ 476 \tenseur{T}_{\star} = \deriv{w}{\tepsilonto_{\star}} 477 \] 478 479## Linear Elasticity 480 481- Linear elasticity assumes a linear relationship between 482 \(\tepsilonto_{\star}\) and \(\tenseur{T}_{\star}\): 483 \[ 484 \tenseur{T}_{\star} = \tenseurq{D}\,\colon\,\tepsilonto_{\star}\quad\Leftrightarrow\quad w=\Frac{1}{2}\tepsilonto_{\star}\,\colon\,\tenseurq{D}\,\colon\,\tepsilonto_{\star} 485 \] 486- The fourth order tensor \(\tenseurq{D}\) satisfies: 487 \[ 488 \tenseurq{D}=\Frac{\partial^{2}w}{\partial\tepsilonto_{\star}\partial\tepsilonto_{\star}} 489 \] 490- \(\tenseurq{D}\) have the following properties: 491 - \(\tenseurq{D}\) has minor symmetries (\(\tenseur{T}_{\star}\) and \(\tepsilonto_{\star}\) 492 are symmetric): 493 \[ 494 D_{ijkl}= D_{jikl}\quad\text{and}\quad D_{ijkl}= D_{ijlk} 495 \] 496 - \(\tenseurq{D}\) has major symmetries (Schwarz theorem): 497 \[ 498 D_{ijkl}=D_{klij} 499 \] 500 501 502<!-- - matrix notations for \(\tenseurq{D}\): --> 503<!-- \[ --> 504<!-- \tenseurq{D}= --> 505<!-- \begin{pmatrix} --> 506<!-- D_{1111} & D_{1122} & D_{1133} & \sqrt{2}\,D_{1112} & \sqrt{2}\,D_{1113} & \sqrt{2}\,D_{1123} \\ --> 507<!-- &D_{2222} & D_{2233} & \sqrt{2}\,D_{2212} & \sqrt{2}\,D_{2213} & \sqrt{2}\,D_{2223} \\ --> 508<!-- & & D_{3333} & \sqrt{2}\,D_{3312} & \sqrt{2}\,D_{3313} & \sqrt{2}\,D_{3323} \\ --> 509<!-- & & & 2\,D_{1212} & 2\,D_{1213} & 2\,D_{1223} \\ --> 510<!-- & & & & 2\,D_{1313} & 2\,D_{1323} \\ --> 511<!-- & & & & & 2\,D_{2323} \\ --> 512<!-- \end{pmatrix} --> 513<!-- \] --> 514<!-- - \(21\) independent coefficients in general --> 515 516 517## Isotropic elasticity 518 519- For an isotropic material, only two independant material properties remains: 520 \[ 521 \tenseur{T}_{\star} 522 = \lambda\,\paren{\trace{\tepsilonto_{\star}}}\,\tenseur{I}+2\,\mu\,\tepsilonto_{\star} 523 \] 524- \(\lambda\) and \(\mu\) are called the Lamé coefficients 525- \(\mu\) is also called the shear modulus 526- In the infinitesimal perturbation hypothesis, this is the Hooke law. 527- In the Green-Lagrange strain framework, this is the Saint-Venant Kirchhoff law. 528- In the logarithmic strain framework, this is the Hencky-Biot law. 529 530### Bulk modulus 531 532- Separate change of volume from deviatoric part: 533 \[ 534 \begin{aligned} 535 \tenseur{T}_{\star}&= \lambda\,\paren{\trace{\tepsilonto_{\star}}}\,\tenseur{I}+2\,\mu\,\tepsilonto_{\star} \\ 536 &= \paren{\lambda+\Frac{2\,\mu}{3}}\,\underbrace{\paren{\trace{\tepsilonto_{\star}}}\,\tenseur{I}}_{\text{volume change}}+2\,\mu\,\underbrace{\paren{\tepsilonto_{\star}-\Frac{\trace{\tepsilonto_{\star}}}{3}\,\tenseur{I}}}_{\text{deviator}} \\ 537 &= K\,\paren{\trace{\tepsilonto_{\star}}}\,\tenseur{I}+2\,\mu\,\paren{\tepsilonto_{\star}-\Frac{\trace{\tepsilonto_{\star}}}{3}\,\tenseur{I}} \\ 538 \end{aligned} 539 \] 540- \(K=\lambda+\Frac{2\,\mu}{3}\) is called the bulk modulus 541 542### Young modulus and Poisson coefficient 543 544- consider an uniaxial tensile state along \(11\): 545 \[ 546 \begin{aligned} 547 \tepsilonto_{\star}& = 548 \begin{pmatrix} 549 \epsilonto_{\star\,11} & 550 \epsilonto_{\star\,22} & 551 \epsilonto_{\star\,33} & 552 0 & 0 & 0 553 \end{pmatrix}^{T}\\ 554 \tenseur{T}_{\star}&= 555 \begin{pmatrix} 556 T_{\star\,11} & 557 0 & 0 & 0 & 0 & 0 558 \end{pmatrix}^{T}\\ 559 \end{aligned} 560 \] 561- The Young modulus \(E\) is defined by: 562 \[ 563 T_{\star\,11} = E\,\epsilonto_{\star\,11} 564 \] 565- The Poisson ratio \(\nu\) is defined by: 566 \[ 567 \epsilonto_{\star\,22} = -\nu\,\epsilonto_{\star\,11} 568 \] 569- \(\lambda=\Frac{E\,\nu}{\paren{1+\nu}\,\paren{1-2\,\nu}}\quad\mu=\Frac{E}{2\,\paren{1+\nu}}\quad K=\Frac{E}{3\,\paren{1-2\,\nu}}\) 570 571## Orthotropy 572 573- An orthotropic material introduces a preferential material 574 frame: 575 \[ 576 \tenseurq{D}= 577 \begin{pmatrix} 578 D_{1111} & D_{1122} & D_{1133} & 0 & 0 & 0 \\ 579 D_{1122} & D_{2222} & D_{2233} & 0 & 0 & 0 \\ 580 D_{1133} & D_{2233} & D_{3333} & 0 & 0 & 0 \\ 581 0 & 0 & 0 & D_{1212} & 0 & 0 \\ 582 0 & 0 & 0 & 0 & D_{1313} & 0 \\ 583 0 & 0 & 0 & 0 & 0 & D_{2323} \\ 584 \end{pmatrix} 585 \] 586- \(9\) independent coefficients 587 588## Thermo-elasticty 589 590### Thermal strain 591 592- The total strain is splitted into an elastic part and a 593 thermal part: 594 \[ 595 \tepsilonto_{\star}= 596 \tepsilonel_{\star}+ 597 \tepsilonth_{\star} 598 \] 599- The elastic part \(\tepsilonel_{\star}\) defines the stresses 600 \(\tenseur{T}_{\star}\) through the Hooke law: 601 \[ 602 \tenseur{T}_{\star}=\tenseurq{D}\,\colon\,\tepsilonel_{\star} 603 \] 604 605#### Isotropic thermal expansion 606 607- The thermal expansion is given by: 608 \[ 609 \Frac{\Delta\,l}{l_{T^{\alpha}}}=\Frac{l_{T}-l_{T^{\alpha}}}{l_{T^{\alpha}}}=\alpha\paren{T}\,\paren{T-T^{\alpha}} 610 \] 611- If the reference temperature \(T^{\alpha}\) for the thermal 612 expansion is different than the reference temperature \(T^{i}\) of the 613 geometry: 614 \[ 615 \Frac{\Delta\,l}{l_{T^{i}}} = \Frac{1}{1+\alpha\paren{T^{i}}\,\paren{T^{i}-T^{\alpha}}}\,\left[\alpha\paren{T}\,\paren{T-T^{\alpha}}-\alpha\paren{T^{i}}\,\paren{T^{i}-T^{\alpha}}\right] 616 \] 617 618# Isotropic damage 619 620# Visco-plasticity 621 622## A first approach to viscoplastic behaviour 623 624- The total strain is splitted into an elastic part and a 625 viscoplastic part: 626 \[ 627 \tepsilonto_{\star}= 628 \tepsilonel_{\star}+ 629 \tepsilonvis_{\star} 630 \] 631- The plastic flow is generally isochoric\footnote{This is an 632 approximation for all strain measures execpt the logarithmic 633 strain}: 634 \[ 635 \trace{\tepsilonvis_{\star}}=0 636 \] 637- Without internal state, the mechanical dissipation associated 638 with plasticity is: 639 \[ 640 \tenseur{T}_{\star}\,\colon\,\tdepsilonto_{\star}= 641 \underbrace{\tenseur{T}_{\star}\,\colon\,\tdepsilonel_{\star}}_{\text{stored reversibly}}+ 642 \underbrace{\tenseur{T}_{\star}\,\colon\,\tdepsilonvis_{\star}}_{\text{dissipated}} 643 \] 644- The expression dissipated power can be rewritten using the deviator of the stress \(\tenseur{s}_{\star}\): 645 \[ 646 \tenseur{T}_{\star}\,\colon\,\tdepsilonvis_{\star}= 647 \tenseur{s}_{\star}\,\colon\,\tdepsilonvis_{\star} 648 \quad\text{with}\quad\tenseur{s}_{\star}=\tenseur{T}_{\star}-\Frac{1}{3}\,\trace{\tenseur{T}_{\star}}\,\tenseur{I} 649 \] 650- The dissipation is maximal if the \(\tdepsilonvis_{\star}\) is 651 colinear with \(\tenseur{s}_{\star}\). 652 653## The Von Mises stress 654 655- The material is now assumed {\bf isotropic} 656- A convenient isotropic norm for deviatoric stress tensor is 657 the Von Mises norm: 658 \[ 659 T^{eq}_{\star}=\sqrt{\Frac{3}{2}\tenseur{s}_{\star}\,\colon\,\tenseur{s}_{\star}} 660 \] 661- The \(\Frac{3}{2}\) factor is here so that in uniaxial tensile tests: 662 \[ 663 T^{eq}_{\star}=\left|T_{xx}\right| 664 \] 665- The Von Mises norm is one the three invariants of the stress 666 (the other ones are the pressure and the determinant) 667- In term of eigen values: 668 \[ 669 T^{eq}_{\star}=\sqrt{\Frac{1}{2}\left[\paren{T_{1}-T_{2}}^{2}+\paren{T_{1}-T_{3}}^{2}+\paren{T_{2}-T_{3}}^{2}\right]} 670 \] 671 672## The normal tensor 673 674- The equation: 675 \[ 676 T^{eq}_{\star}=\text{Cste} 677 \] 678 defines a sphere in the deviatoric space 679- The normal to this surface is: 680 \[ 681 \tenseur{n}_{\star}=\deriv{T^{eq}_{\star}}{\tenseur{T}_{\star}}=\Frac{3\,\tenseur{s}_{\star}}{2\,T^{eq}_{\star}} 682 \] 683- \(\tenseur{n}_{\star}\colon\tenseur{n}_{\star}=\Frac{3}{2}\) 684- The normal is colinear to \(\tenseur{s}_{\star}\), thus an 685 isochoric viscoplastic flow of the form: 686 \[ 687 \tdepsilonvis_{\star}=f\paren{\tenseur{T}_{\star}}\tenseur{n}_{\star} 688 \] 689 would maximise the mechanical dissipation. 690- In uniaxial tensile tests: 691 \[ 692 \tenseur{n}_{\star}= 693 \begin{pmatrix} 694 1 & 695 -\frac{1}{2} & 696 -\frac{1}{2} & 697 0 & 698 0 & 699 0 700 \end{pmatrix}^{T} 701 \] 702 703## The Norton behaviour 704 705- The viscoplastic flow is: 706 \( 707 \tdepsilonvis_{\star}=f\paren{\tenseur{T}_{\star}}\tenseur{n}_{\star} 708 \) 709- The material being isotropic, \(f\) must be a function of the 710 invariants of the stresses: the pressure, the Von Mises stress, 711 the determinant. 712- Experimentally, viscoplastic behaviour is found to be pressure 713 insensitive. 714- The effect of the third invariant is neglected in general. 715- Thus, a simple viscoplastic model for an isotropic 716 incompressible material is: 717 \[ 718 \tdepsilonvis_{\star}=f\paren{T^{eq}_{\star}}\tenseur{n}_{\star} 719 \] 720- Restrictions: 721 - \(f\) must be positive for the dissipation to be positive 722 - \(f\paren{\tenseur{0}}\) must be null 723 - The Norton behaviour correspond to a power function: 724 \[ 725 \tdepsilonvis_{\star}=A\,\paren{T^{eq}_{\star}}^{n}\tenseur{n}_{\star}=\dot{\varepsilon}^{0}\paren{\Frac{T^{eq}_{\star}}{T^{0}}}^{n}\tenseur{n}_{\star} 726 \] 727 728## Equivalent viscoplastic strain 729 730- A convenient choice for the viscoplastic strain rate norm is: 731 \[ 732 \dot{p}=f\paren{T^{eq}_{\star}}=\sqrt{\Frac{2}{3}\,\tdepsilonvis_{\star}\,\colon\,\tdepsilonvis_{\star}} 733 \] 734- In uniaxial tensile tests: 735 \[ 736 \dot{p} = \left|\paren{\depsilonvis_{\star}}_{xx}\right| 737 \] 738- The equivalent viscoplastic strain is defined by: 739 \[ 740 p=\int_{0}^{t}\dot{p}\,\dtot\,t 741 \] 742- This quantity is a convenient measure of the viscoplastic 743 history of the material and is widely used as a damage criterium. 744 745 746## Dissipation potential 747 748- The Norton behaviour can be expressed as: 749 \[ 750 \tdepsilonvis_{\star}=\deriv{\potentieldissipdual}{\tenseur{T}_{\star}} 751 \quad\text{with}\quad 752 \potentieldissipdual\paren{T^{eq}_{\star}}=\Frac{T^{0}\dot{\varepsilon}^{0}}{n+1}\paren{\Frac{T^{eq}_{\star}}{T^{0}}}^{n+1} 753 \] 754- \(\potentieldissipdual\) is called the dissipation potential 755- Other expressions of the dissipation potential defines used the 756 define other viscoplastic behaviours. 757- The resulting viscoplastic behaviours will lead to a positive 758 dissipation if \(\potentieldissipdual\) is {\bf convex} and {\bf 759 minimal} at zero. 760- The introduction of dissipation potentials is the departure of 761 a theorical developments which can ease the formulation of 762 mechanical behaviours: 763 - thermodynamical consistent behaviours 764 - numerically efficient behaviours 765 766# Thermodynamics 767 768## First principle 769 770# The Finite Element Method 771 772## Principle of virtual power 773 774- Let \(\partial_{u}\Omega\) the boundary part where 775 displacements are prescribed 776- \(\partial_{t}\Omega=\partial\Omega\setminus\partial_{u}\Omega\) 777 is the boundary part where tractions are prescribed 778- Let \(\vec{v}^{\star}\) be a vector field compatible with 779 prescribed displacement and 780 \(\delta\,\vec{v}^{\star}=\vec{v}^{\star}-\vec{v}\), then: 781 \[ 782 \underbrace{\int_{\Omega}\tsigma\,\colon\,\delta\,\tns{D}^{\star}\,\dtot\,v}_{\text{virtual inner forces power}} = \underbrace{\int_{\Omega}\vec{f}\,.\,\delta\,\vec{v}^{\star}\,\dtot\,v + \int_{\partial\,\Omega_{t}}\vec{t}\,.\,\delta\,\vec{v}^{\star}\,\dtot\,s}_{\text{virtual external forces power}} 783 \] 784- this is the principle of virtual power which is the basis of 785 the Finite Element Method (FEM) 786 787## Finite element method 788 789- The principle of virtual power is used to find the best 790 approximation of the solution on a finite space. 791- Finite elements are a widely used way of defining such a 792 finite space by discretizing the real geometry by subdomains 793 called finite elements: 794 - Given values at specified points of the finite element (the 795 nodes), the function value is approximated by interpolation 796 functions. 797 798### Resolution 799 800- Mechanical equilibrium: find\(\Delta\discret{\vec{u}}\) such as: 801 \[ 802 \small 803 \residuEF\paren{\Delta\discret{\vec{u}}}=\discret{\vec{O}}\quad\text{ 804 avec 805 }\quad\residuEF\paren{\Delta\discret{\vec{u}}}=\forceintEF\paren{\Delta\discret{\vec{u}}}-\forceextEF 806 \] 807- element contribution to inner forces: 808 \[ 809 \small 810 \begin{aligned} 811 \forceintElem&=\int_{V^{e}}\tsigma_{t+\Delta 812 t}\paren{\Delta\,\tepsilonto,\Delta\, t}\colon\tenseur{B}\;\dtot V \\ 813 &= \sum_{i=1}^{N^{G}} \paren{\tsigma_{t+\Delta\,t}\paren{\Delta\tepsilonto\paren{\vec{\eta}_{i}},\Delta\, t}\colon\tenseurq{B}\paren{\vec{\eta}_{i}}}w_{i} 814 \end{aligned} 815 \] 816 where \(\tenseur{B}\) gives the relationship between \(\Delta\,\tepsilonto\) and \(\Delta\discret{\vec{u}}\) 817 818 \[ 819 \forceintElem = \sum_{i=1}^{N^{G}} \paren{\tsigma_{t+\Delta\,t}\paren{\Delta\tepsilonto\paren{\vec{\eta}_{i}},\Delta\, t}\colon\tenseurq{B}\paren{\vec{\eta}_{i}}}w_{i} 820 \] 821 822### Resolution using the \nom{Newton-Raphson} algorithm 823 824<!-- % \[ --> 825<!-- % \Delta\discret{\vec{u}}^{n+1}=\Delta\discret{\vec{u}}^{n}-\paren{\left.\deriv{\residuEF}{\Delta\discret{\vec{u}}}\right|_{\Delta\discret{\vec{u}}^{n}}}^{-1}.\residuEF\paren{\Delta\discret{\vec{u}}^{n}}= \Delta\discret{\vec{u}}^{n}-\tenseurq{\mathbb{K}}^{-1}.\residuEF\paren{\Delta\discret{\vec{u}}^{n}}} --> 826<!-- % \] --> 827 828<!-- % \[ --> 829<!-- % \Delta\discret{\vec{u}}^{n+1}=\Delta\discret{\vec{u}}^{n}-\tenseurq{\mathbb{K}}^{-1}.\residuEF\paren{\Delta\discret{\vec{u}}^{n}} --> 830<!-- % \] --> 831<!-- - element contribution to the stiffness: --> 832<!-- \[ --> 833<!-- \small --> 834<!-- \tenseurq{\mathbb{K}}^{e}=\displaystyle\sum_{i=1}^{N^{G}} --> 835<!-- \mbox{}^{t}\tenseurq{B}\paren{\vec{\eta}_{i}}\colon\deriv{\Delta\tsigma}{\Delta\tepsilonto}\paren{\vec{\eta}_{i}}\colon\tenseurq{B}\paren{\vec{\eta}_{i}}w_{i} --> 836<!-- \] --> 837<!-- \(\scriptsize\deriv{\Delta\tsigma}{\Delta\tepsilonto}\) is the --> 838<!-- {\bf consistent tangent operator} --> 839 840 841## Mechanical behaviours 842 843<!-- \paragraph{Main functions of the mechanical behaviour} --> 844 845<!-- \[ --> 846<!-- \paren{ --> 847<!-- \bts{\tepsilonto_{\star}} \,, --> 848<!-- \bts{\vec{Y}} \,, --> 849<!-- \Delta\,\tepsilonto_{\star} \,, --> 850<!-- \Delta\,t --> 851<!-- } --> 852<!-- \underbrace{\Longrightarrow}_{behaviour} --> 853<!-- \paren{ --> 854<!-- \ets{\tenseur{T}_{\star}} \,, --> 855<!-- \ets{\vec{Y}} \,, --> 856<!-- \deriv{\Delta\,\tepsilonto_{\star}}{\Delta\,\tepsilonto_{\star}} --> 857<!-- } --> 858<!-- \] --> 859<!-- - Given a strain increment \(\Delta\,\tepsilonto_{\star}\) over --> 860<!-- a time step \(\Delta\,t\), the mechanical behaviour must compute: --> 861<!-- - The value of the stress \(\ets{\tenseur{T}_{\star}}\) at the --> 862<!-- end of the time step. --> 863<!-- - The value of internal state variables, noted --> 864<!-- \(\ets{\vec{Y}}\) at the end of the time step. --> 865<!-- - The consistent tangent operator: --> 866<!-- \( --> 867<!-- \deriv{\Delta\,\tenseur{T}_{\star}}{\Delta\,\tepsilonto_{\star}} --> 868<!-- \) --> 869<!-- - For specific cases, the mechanical behaviour shall also provide: --> 870<!-- - a prediction operator --> 871<!-- - the elastic operator (Abaqus-Explicit, Europlexus) --> 872<!-- - estimation of the stored and dissipated energies (Abaqus-Explicit) --> 873 874 875<!-- \paragraph{Other functions of the mechanical behaviour} --> 876 877<!-- - Provide a estimation of the next time step for time step automatic adaptation --> 878<!-- - Check bounds: --> 879<!-- - Physical bounds --> 880<!-- - Standard bounds --> 881<!-- - Clear error messages --> 882<!-- - Parameters --> 883<!-- - It is all about AQ! --> 884<!-- - Parametric studies, identification, etc… --> 885<!-- - Generate mtest files on integration failures --> 886<!-- - Generated example of usage: --> 887<!-- - Generation of MODELISER/MATERIAU instructions (Cast3M) --> 888<!-- - Input file for Abaqus --> 889<!-- - Provide information for dynamic resolution of inputs (MTest/Aster/Europlexus): --> 890<!-- - Numbers Types (scalar, tensors, symmetric tensors) --> 891<!-- - Entry names /Glossary names… --> 892 893 894<!-- \paragraph{Mechanical behaviour integration} --> 895 896<!-- - The evolution of the state variables are usually expressed by --> 897<!-- a ordinary differential equation: --> 898<!-- \[ --> 899<!-- \vec{\dot{Y}}=G\paren{\vec{Y},\tdepsilonto_{\star}} --> 900<!-- \] --> 901<!-- - Example of the Norton behaviour: --> 902<!-- \[ --> 903<!-- \left\{ --> 904<!-- \begin{aligned} --> 905<!-- \tenseur{T}_{\star}&=\tenseurq{D}\,\colon\,\tepsilonel \\ --> 906<!-- \tdepsilonel+\tdepsilonvis &= \tdepsilonto_{\star} \\ --> 907<!-- \tdepsilonvis &= \dot{\varepsilon}^{0}\paren{\Frac{T^{eq}_{\star}}{T^{0}}}^{n}\tenseur{n}_{\star} \\ --> 908<!-- \end{aligned} --> 909<!-- \right. --> 910<!-- \] --> 911 912 913<!-- \paragraph{Explicit schemes} --> 914 915<!-- - This ode may be solved using one of the many Runge-Kutta algorithms: --> 916<!-- - This is not recommended as one can't derive the consistent --> 917<!-- tangent operator. --> 918<!-- - There are intrisic pitfalls with those algorithms when the --> 919<!-- behaviour depends on external state variables, such as the --> 920<!-- temperature. --> 921<!-- - Plasticity and damage are not treated exactly. --> 922<!-- - Poor numerical performances. --> 923 924 925<!-- \paragraph{Implicit schemes} --> 926 927<!-- - The previous ordinary differential equations can be rewritten --> 928<!-- as a system of non linear equations: --> 929<!-- \[ --> 930<!-- F\paren{\Delta\,\vec{Y}}= --> 931<!-- \Delta\,\vec{Y}-\Delta\,t\,G\paren{\mts{\vec{Y}},\Delta\,\tepsilonto_{\star}}=\vec{0} --> 932<!-- \] --> 933<!-- with --> 934<!-- \(\mts{\vec{Y}}=\bts{\vec{Y}}+\theta\,\Delta\,\vec{Y}\) --> 935<!-- - \(\theta\) is a numerical parameter: --> 936<!-- - \(\theta\in\left[0:1\right]\) --> 937<!-- - \(\theta=\frac{1}{2}\) leads to a second order method: --> 938 939 940<!-- \paragraph{Example} --> 941 942<!-- - The Norton behaviour integrated by an implicit schemes leads --> 943<!-- to: --> 944<!-- \[ --> 945<!-- \begin{aligned} --> 946<!-- \mts{\tenseur{T}_{\star}}&=\tenseurq{D}\,\colon\,\mts{\tepsilonel}\\ --> 947<!-- \Delta\,\tepsilonel+\Delta\,\tepsilonvis - \Delta\,\tepsilonto_{\star} &= \tenseur{0}\\ --> 948<!-- \Delta\,\tepsilonvis -\Delta\,t\,\dot{\varepsilon}^{0}\,\paren{\Frac{\mts{T^{eq}_{\star}}}{T^{0}}}^{n}\,\mts{\tenseur{n}_{\star}}&=\tenseur{0} \\ --> 949<!-- \end{aligned} --> 950<!-- \] --> 951 952 953<!-- \paragraph{Resolution of the implicit scheme} --> 954 955<!-- - The previous equation is generally solved using a --> 956<!-- variant of the Newton-Raphson algorithm: --> 957<!-- \[ --> 958<!-- \Delta\,\vec{Y}^{(n+1)}=\Delta\,\vec{Y}^{(n)}-J^{-1}\paren{\Delta\,\vec{Y}^{(n)}}\,F\paren{\Delta\,\vec{Y}^{(n)}} --> 959<!-- \] --> 960<!-- - \(J=\derivtot{F}{\Delta\,\vec{Y}}\) is the jacobian of the system: --> 961<!-- - Computing \(J\) is the difficult part ! --> 962<!-- - One can use a finite-difference approximation or Broyden algorithm: --> 963<!-- - Poorer performances (although better than explicit schemes) --> 964 965 966<!-- \paragraph{Block decomposition} --> 967 968<!-- - The implicit system can be decomposed by blocks: --> 969<!-- \[ --> 970<!-- F = --> 971<!-- \begin{pmatrix} --> 972<!-- f_{y_{1}} \\ --> 973<!-- \vdots \\ --> 974<!-- f_{y_{N}} \\ --> 975<!-- \end{pmatrix} --> 976<!-- \] --> 977<!-- - The jacobian system can be also be decomposed by blocks: --> 978<!-- \[ --> 979<!-- J = \deriv{F}{Y} = --> 980<!-- \begin{pmatrix} --> 981<!-- \deriv{f_{y_{1}}}{y_{1}} & \ldots & \ldots & \ldots & \ldots \\ --> 982<!-- \vdots & \vdots & \vdots & \vdots & \vdots \\ --> 983<!-- \vdots & \vdots & \deriv{f_{y_{i}}}{y_{j}} & \vdots & \vdots \\ --> 984<!-- \vdots & \vdots & \vdots & \vdots & \vdots \\ --> 985<!-- \ldots & \ldots & \ldots & \ldots & \deriv{f_{y_{N}}}{y_{N}} \\n --> 986<!-- \end{pmatrix} --> 987<!-- \] --> 988 989 990<!-- Local IspellDict: english --> 991