tdoc.tex - slidergrid - grid of elastic sliders on a frictional surface
 (HTM) git clone git://src.adamsgaard.dk/slidergrid
 (DIR) Log
 (DIR) Files
 (DIR) Refs
 (DIR) README
 (DIR) LICENSE
       ---
       tdoc.tex (14294B)
       ---
            1 \documentclass[11pt,a4paper]{article}
            2 
            3 \usepackage{a4wide}
            4 
            5 %\usepackage[german, english]{babel}
            6 %\usepackage{tabularx}
            7 %\usepackage{cancel}
            8 %\usepackage{multirow}
            9 %\usepackage{supertabular}
           10 %\usepackage{algorithmic}
           11 %\usepackage{algorithm}
           12 %\usepackage{amsthm}
           13 %\usepackage{float}
           14 %\usepackage{subfig}
           15 %\usepackage{rotating}
           16 \usepackage{amsmath}
           17 \setcounter{MaxMatrixCols}{20}  % allow more than 10 matrix columns
           18 
           19 \usepackage[T1]{fontenc} % Font encoding
           20 \usepackage{charter} % Serif body font
           21 \usepackage[charter]{mathdesign} % Math font
           22 \usepackage[scale=0.8]{sourcecodepro} % Monospaced fontenc
           23 \usepackage[lf]{FiraSans} % Sans-serif font
           24 
           25 \usepackage{listings}
           26 
           27 \usepackage{hyperref}
           28 
           29 \usepackage{soul} % for st strikethrough command
           30 
           31 %\usepackage[round]{natbib}
           32 \usepackage[natbib=true, style=authoryear, bibstyle=authoryear-comp, 
           33 maxbibnames=10,
           34 maxcitenames=2, backend=bibtex8]{biblatex}
           35 \bibliography{/home/ad/articles/own/BIBnew.bib}
           36 
           37 
           38 \begin{document}
           39 
           40 \title{Lagrangian model of the elastic, viscous and plastic deformation of a 
           41     series of bonded nodes moving on a frictional surface}
           42 
           43 \author{Anders Damsgaard}
           44 \date{{\small Institute of Geophysics and Planetary Physics\\Scripps Institution 
           45 of Oceanography\\University of California, San Diego}\\[3mm] Last revision: 
           46 \today}
           47 
           48 \maketitle
           49 
           50 \section{Methods}
           51 Our approach treats the short-temporal scale behavior of stick-slip as a 
           52 rigid-body dynamics problem.  The material is represented as a discrete number 
           53 of Lagrangian points (\emph{nodes}) which are mechanically interacting with each 
           54 other and the boundary conditions.
           55 
           56 The Lagrangian nodes are connected with visco-elastic beam elements.
           57 The bonds are resistive to tension and compression, shearing, twisting, and 
           58 bending, which ensures elastic uniformity regardless of geometric node 
           59 arrangement \citep{Bolander1998, Radjai2011}.  Alternatively, the node 
           60 interaction could be parameterized as simple springs which exclusively provide 
           61 resistance to tension and compression, and resistance to shearing, bending and 
           62 twisting would be introduced by discretizing the elastic material into an 
           63 irregular network of many more nodes \citep[e.g.][]{Topin2007, Topin2009}.  Here 
           64 we chose the former approach which allows us to keep the number of nodes low.
           65 
           66 The kinematic degrees of freedom are determined by explicit integration of 
           67 Newton's second law of motion for translation and rotation.  For a point $i$ 
           68 with bonded interactions to nodes $j\in N_c$, the translational accelerations 
           69 ($\boldsymbol{a}$) are found from the sums of forces:
           70 \begin{equation}
           71     \boldsymbol{a}_i =
           72     \frac{
           73     \boldsymbol{f}_i^\text{d}
           74     + \boldsymbol{f}_i^\text{f}
           75     + \sum^{N_c}_j \left[
           76         \boldsymbol{f}_{i,j}^\text{p} +
           77         \boldsymbol{f}_{i,j}^\text{s}
           78     \right]
           79 }{m_i}
           80     + \boldsymbol{g}
           81 \label{eq:n2-tran}
           82 \end{equation}
           83 where $\boldsymbol{f}_i^\text{d}$ is the gravitational driving stress due to 
           84 surface slope, $\boldsymbol{g}$ is the gravitational acceleration, and
           85 $\boldsymbol{f}_i^\text{f}$ is the frictional force provided if the point is 
           86 resting on the lower surface. Bonded interaction with another point $j$ 
           87 contributes to translational acceleration through bond-parallel and bond-normal 
           88 shear forces, $\boldsymbol{f}_{i,j}^\text{p}$ and 
           89 $\boldsymbol{f}_{i,j}^\text{s}$, respectively.
           90 
           91 The angular accelerations ($\boldsymbol{\alpha}$) are found from the sums of 
           92 torques:
           93 \begin{equation}
           94     \boldsymbol{\alpha}_i =
           95     \sum^{N_c}_j
           96     \left[
           97         \frac{\boldsymbol{t}^{i,j}_{\bar{x}}}{I^\text{p}_{i,j}} +
           98         \frac{\boldsymbol{t}^{i,j}_{\bar{y}}}{I^\text{n}_{i,j}} +
           99         \frac{\boldsymbol{t}^{i,j}_{\bar{z}}}{I^\text{n}_{i,j}}
          100     \right]
          101 \label{eq:n2-ang}
          102 \end{equation}
          103 here, $\boldsymbol{t}^\text{s}$ is the torque resulting from shearing motion of 
          104 the bond, while the torque $\boldsymbol{t}^{t}$ results from relative twisting.
          105 $I^\text{n}_{i,j}$ is the bond-normal mass moment of inertia at the point, and 
          106 $I^\text{p}$ is polar mass moment of inertia of the bond.  The above equation 
          107 implies the simplifying assumption that the nodes are bonded in a configuration 
          108 with geometric symmetry, which is a good approximation inside the grid but 
          109 slightly worse at the grid edges.
          110 
          111 
          112 \subsection{Visco-elastic interaction between nodes}
          113 The total force ($\boldsymbol{f}$) and torque ($\boldsymbol{t}$) on two nodes 
          114 ($i$ and $j$) is found by determining the stress response of a three-dimensional 
          115 elastic Timoshenko beam to strain \citet{Schlangen1996, Austrell2004, 
          116     Aastroem2013}.
          117 In the following the forces and torques are described for node $i$ but are
          118 of equal magnitude and with opposite sign for node $j$.  The interaction 
          119 accounts for resistance to tension and compression, shear, torsion, and bending.  
          120 The equations below are derived from the stiffness matrix in 
          121 \citet{Austrell2004}.  The components for the three-dimensional force vector on 
          122 node $i$ are:
          123 \begin{equation}
          124     \begin{split}
          125         f_{\bar{x}}^i & = \frac{EA}{L}
          126         \left( p_{\bar{x}}^{*,i} - p_{\bar{x}}^{*,j} \right)\\
          127 %
          128         f_{\bar{y}}^i & = \frac{12EI_A}{L^3}
          129         \left( p_{\bar{y}}^{*,i} - p_{\bar{y}}^{*,j} \right)
          130         + \frac{6EI_A}{L^2}
          131         \left( \Omega_{\bar{z}}^{*,i} + \Omega_{\bar{z}}^{*,j} \right)\\
          132 %
          133         f_{\bar{z}}^i & = \frac{12EI_A}{L^3}
          134         \left( p_{\bar{z}}^{*,i} - p_{\bar{z}}^{*,j} \right)
          135         - \frac{6EI_A}{L^2}
          136         \left( \Omega_{\bar{y}}^{*,i} + \Omega_{\bar{y}}^{*,j} \right)
          137     \end{split}
          138 \end{equation}
          139 where $\bar{x}, \bar{y}, \bar{z}$ are the bond-relative coordinates.
          140 The linear and angular relative displacement of the nodes is described by $p^*$ 
          141 and $\Omega^*$.  $E$ is Young's modulus, $G$ is the shear modulus, $A$ is the 
          142 beam cross-sectional area, and $L$ is the original beam length. $I_A$ is the 
          143 area moment of inertia of the beam ($I_A = a^4 12^{-1}$ where $a$ is the beam 
          144 side length).  The torque on node $i$ is:
          145 \begin{equation}
          146     \begin{split}
          147         t_{\bar{x}}^i & = \frac{GJ}{L}
          148         \left( \Omega_{\bar{x}}^{*,i} + \Omega_{\bar{x}}^{*,j} \right)\\
          149 %
          150         t_{\bar{y}}^i & = \frac{6EI_A}{L^2}
          151         \left(p_{\bar{z}}^{*,j} - p_{\bar{z}}^{*,i} \right)
          152         + \frac{4EI_A}{L}
          153         \left( \Omega_{\bar{y}}^{*,i} + \frac{\Omega_{\bar{y}}^{*,j}}{2} 
          154             \right)\\
          155 %
          156         t_{\bar{z}}^i & = \frac{6EI_A}{L^2}
          157         \left(p_{\bar{y}}^{*,i} - p_{\bar{y}}^{*,j} \right)
          158         + \frac{4EI_A}{L}
          159         \left( \Omega_{\bar{z}}^{*,i} + \frac{\Omega_{\bar{z}}^{*,j}}{2} 
          160     \right)\\
          161     \end{split}
          162 \end{equation}
          163 $GJ$ is the Saint-Venant torsional stiffness, where $J$ called the torsional 
          164 rigidity multiplier or torsion constant. For a beam with a solid square 
          165 cross-sectional shape it can be approximated as $J \approx 0.140577 a^4$ where 
          166 $a$ is the side length  \citep{Timoshenko1951, Roark1954, Weisstein2016}.  
          167 
          168 % Torsional constant:
          169 % https://en.wikipedia.org/wiki/Torsion_constant
          170 % http://mathworld.wolfram.com/TorsionalRigidity.html (K: K_v)
          171 % http://physics.stackexchange.com/questions/83148/where-i-can-find-a-torsional-stiffness-table-for-different-types-of-stainless-st
          172 % St Venant torsion: K_v = 1/G  (Austrell et al. 2004, table 3) Does it make sense?
          173 % chapter4.pdf: K_v = J
          174 % https://skyciv.com/free-moment-of-inertia-calculator/
          175 
          176 
          177 The deformation and reactive forces are determined relative to the orientation 
          178 of the bond.  Common geometrical vectors include the inter-distance vector 
          179 $\boldsymbol{d}$ between nodes $\boldsymbol{p}_i$ and $\boldsymbol{p}_j$:
          180 \begin{equation}
          181     \boldsymbol{d}_{i,j} = \boldsymbol{p}_i - \boldsymbol{p}_j
          182 \end{equation}
          183 which in normalized form constitutes the bond-parallel unit vector:
          184 \begin{equation}
          185     \boldsymbol{n}_{i,j} = \frac{\boldsymbol{d}_{i,j}}{||\boldsymbol{d}_{i,j}||}
          186 \end{equation}
          187 
          188 The nodes move by translational and rotational velocities.  The combined 
          189 relative velocity between the nodes is found as \citep[e.g.][]{Hinrichsen2004, 
          190     Luding2008}:
          191 \begin{equation}
          192     \boldsymbol{v}_{i,j} = \boldsymbol{v}_i - \boldsymbol{v}_j +
          193     \frac{\boldsymbol{d}_{i,j}}{2} \times \boldsymbol{\omega}_i +
          194     \frac{\boldsymbol{d}_{i,j}}{2} \times \boldsymbol{\omega}_j
          195 \end{equation}
          196 The velocity can be decomposed into spatial components relative to the bond 
          197 orientation, e.g.\ the bond-parallel and bond-shear velocity, respectively:
          198 \begin{equation}
          199     v^\text{p}_{i,j} = \boldsymbol{v}_{i,j} \cdot \boldsymbol{n}_{i,j}
          200 \end{equation}
          201 \begin{equation}
          202     \boldsymbol{v}^\text{s}_{i,j} = \boldsymbol{v}_{i,j} - \boldsymbol{n}_{i,j}
          203     \left(
          204         \boldsymbol{v}_{i,j}
          205         \cdot
          206         \boldsymbol{n}_{i,j}
          207     \right)
          208 \end{equation}
          209 
          210 
          211 
          212 \subsection{Temporal integration}
          213 Once the force and torque sum components at time $t$ have been determined, the 
          214 kinematic degrees of freedom at time $t+\Delta t$ can be found by explicit 
          215 temporal integration of moment balance equations~\ref{eq:n2-tran} 
          216 and~\ref{eq:n2-ang}.
          217 We use an integration scheme based on the third-order Taylor expansion, which 
          218 results in a truncation error on the order of $O(\Delta t^4)$ for positions and 
          219 $O(\Delta t^3)$ for velocities.  This scheme includes changes in acceleration as 
          220 the highest order term, which are approximated by backwards differences.  For 
          221 the translational degrees of freedom:
          222 \begin{equation}
          223     \boldsymbol{p}^i_{t+\Delta t} =
          224     \boldsymbol{p}^i_{t} +
          225     \boldsymbol{v}^i_{t} \Delta t +
          226     \frac{1}{2} \boldsymbol{a}^i_{t} \Delta t^2 +
          227     \frac{1}{6} \frac{\boldsymbol{a}^i_{t}
          228         - \boldsymbol{a}^i_{t - \Delta t}}{\Delta t} \Delta t^3
          229 \end{equation}
          230 \begin{equation}
          231     \boldsymbol{v}^i_{t+\Delta t} =
          232     \boldsymbol{v}^i_{t} +
          233     \boldsymbol{a}^i_{t} \Delta t +
          234     \frac{1}{2} \frac{\boldsymbol{a}^i_{t}
          235         - \boldsymbol{a}^i_{t - \Delta t}}{\Delta t} \Delta t^2
          236 \end{equation}
          237 At $t=0$ the acceleration change term is defined as zero.  The angular degrees 
          238 of freedom are found correspondingly:
          239 \begin{equation}
          240     \boldsymbol{\Omega}^i_{t+\Delta t} =
          241     \boldsymbol{\Omega}^i_{t} +
          242     \boldsymbol{\omega}^i_{t} \Delta t +
          243     \frac{1}{2} \boldsymbol{\alpha}^i_{t} \Delta t^2 +
          244     \frac{1}{6} \frac{\boldsymbol{\alpha}^i_{t}
          245         - \boldsymbol{\alpha}^i_{t - \Delta t}}{\Delta t} \Delta t^3
          246 \end{equation}
          247 \begin{equation}
          248     \boldsymbol{\omega}^i_{t+\Delta t} =
          249     \boldsymbol{\omega}^i_{t} +
          250     \boldsymbol{\alpha}^i_{t} \Delta t +
          251     \frac{1}{2} \frac{\boldsymbol{\alpha}^i_{t}
          252         - \boldsymbol{\alpha}^i_{t - \Delta t}}{\Delta t} \Delta t^2
          253 \end{equation}
          254 
          255 The numerical time step $\Delta t$ is found by considering the largest elastic 
          256 stiffness in the system relative to the smallest mass:
          257 \begin{equation}
          258     \Delta t =
          259     \epsilon
          260     \left[
          261         \min (m_i)^{-1}
          262         \max \left(
          263             \max \left(
          264                 \frac{E A_{i,j}}{||\boldsymbol{d}_{0}^{i,j}||}
          265             \right)
          266             ,
          267             \max \left(
          268                 \frac{G A_{i,j}}{||\boldsymbol{d}^{i,j}||}
          269             \right)
          270         \right)
          271     \right]^{-1/2}
          272 \end{equation}
          273 where $\epsilon$ is a safety factor related to the geometric structure of the 
          274 bonded network.  We use $\epsilon = 0.07$.
          275 
          276 
          277 \appendix
          278 \section{Stiffness matrix}
          279 \begin{equation}
          280     \begin{bmatrix}
          281         f_{\bar{x}}^i\\[0.6em]
          282         f_{\bar{y}}^i\\[0.6em]
          283         f_{\bar{z}}^i\\[0.6em]
          284         t_{\bar{x}}^i\\[0.6em]
          285         t_{\bar{y}}^i\\[0.6em]
          286         t_{\bar{z}}^i\\[0.6em]
          287         f_{\bar{x}}^j\\[0.6em]
          288         f_{\bar{y}}^j\\[0.6em]
          289         f_{\bar{z}}^j\\[0.6em]
          290         t_{\bar{x}}^j\\[0.6em]
          291         t_{\bar{y}}^j\\[0.6em]
          292         t_{\bar{z}}^j\\
          293     \end{bmatrix}
          294     =
          295     \begin{bmatrix}
          296         \frac{EA}{L} & 0 & 0 & 0 & 0 & 0 & \frac{-EA}{L} & 0 & 0 & 0 & 0 & 0\\[0.5em]
          297         0 & \frac{12EI_{\bar{z}}}{L^3} & 0 & 0 & 0 & \frac{6EI_{\bar{z}}}{L^2} & 
          298         0 & \frac{-12EI_{\bar{z}}}{L^3} & 0 & 0 & 0 & 
          299         \frac{6EI_{\bar{z}}}{L^2}\\[0.5em]
          300         0 & 0 & \frac{12EI_{\bar{y}}}{L^3} & 0 & \frac{-6EI_{\bar{y}}}{L^2} & 0 
          301         & 0 & 0 & \frac{-12EI_{\bar{y}}}{L^3} & 0 & \frac{-6EI_{\bar{y}}}{L^2} & 
          302         0\\[0.5em]
          303         0 & 0 & 0 & \frac{GJ}{L} & 0 & 0 & 0 & 0 & 0 & \frac{-GJ}{L} & 0 & 
          304         0\\[0.5em]
          305         0 & 0 & \frac{-6EI_{\bar{y}}}{L^2} & 0 & \frac{4EI_{\bar{y}}}{L} & 0 & 0 
          306         & 0 & \frac{6EI_{\bar{y}}}{L^2} & 0 & \frac{2EI_{\bar{y}}}{L} & 
          307         0\\[0.5em]
          308         0 & \frac{6EI_{\bar{z}}}{L^2} & 0 & 0 & 0 & \frac{4EI_{\bar{z}}}{L} & 0 
          309         & \frac{-6EI_{\bar{z}}}{L^2} & 0 & 0 & 0 & 
          310         \frac{2EI_{\bar{z}}}{L}\\[0.5em]
          311         \frac{-EA}{L} & 0 & 0 & 0 & 0 & 0 & \frac{EA}{L} & 0 & 0 & 0 & 0 & 0\\[0.5em]
          312         0 & \frac{-12EI_{\bar{z}}}{L^3} & 0 & 0 & 0 & \frac{-6EI_{\bar{z}}}{L^2} 
          313         & 0 & \frac{12EI_{\bar{z}}}{L^3} & 0 & 0 & 0 & 
          314         \frac{-6EI_{\bar{z}}}{L^2}\\[0.5em]
          315         0 & 0 & \frac{-12EI_{\bar{y}}}{L^3} & 0 & \frac{-6EI_{\bar{y}}}{L^2} & 0 
          316         & 0 & 0 & \frac{12EI_{\bar{y}}}{L^3} & 0 & \frac{6EI_{\bar{y}}}{L^2} & 
          317         0\\[0.5em]
          318         0 & 0 & 0 & \frac{-GJ}{L} & 0 & 0 & 0 & 0 & 0 & \frac{GJ}{L} & 0 & 
          319         0\\[0.5em]
          320         0 & 0 & \frac{-6EI_{\bar{y}}}{L^2} & 0 & \frac{2EI_{\bar{y}}}{L} & 0 & 0 
          321         & 0 & \frac{6EI_{\bar{y}}}{L^2} & 0 & \frac{4EI_{\bar{y}}}{L} & 
          322         0\\[0.5em]
          323         0 & \frac{6EI_{\bar{z}}}{L^2} & 0 & 0 & 0 & \frac{2EI_{\bar{z}}}{L} & 0 
          324     & \frac{-6EI_{\bar{z}}}{L^2} & 0 & 0 & 0 & \frac{4EI_{\bar{z}}}{L}\\
          325     \end{bmatrix}
          326     \begin{bmatrix}
          327         p_{\bar{x}}^i\\[0.6em]
          328         p_{\bar{y}}^i\\[0.6em]
          329         p_{\bar{z}}^i\\[0.6em]
          330         \Omega_{\bar{x}}^i\\[0.6em]
          331         \Omega_{\bar{y}}^i\\[0.6em]
          332         \Omega_{\bar{z}}^i\\[0.6em]
          333         p_{\bar{x}}^j\\[0.6em]
          334         p_{\bar{y}}^j\\[0.6em]
          335         p_{\bar{z}}^j\\[0.6em]
          336         \Omega_{\bar{x}}^j\\[0.6em]
          337         \Omega_{\bar{y}}^j\\[0.6em]
          338         \Omega_{\bar{z}}^j\\
          339     \end{bmatrix}
          340 \end{equation}
          341 
          342 
          343 
          344 \printbibliography{}
          345 
          346 \end{document}