A Stabilized Finite Element Method for Morpho-Visco-Poroelastic Model
Extracted source text · arXiv 2512.10718
%%
%% Copyright 2007-2025 Elsevier Ltd
%%
%% This file is part of the 'Elsarticle Bundle'.
%% ---------------------------------------------
%%
%% It may be distributed under the conditions of the LaTeX Project Public
%% License, either version 1.3 of this license or (at your option) any
%% later version. The latest version of this license is in
%% http://www.latex-project.org/lppl.txt
%% and version 1.3 or later is part of all distributions of LaTeX
%% version 1999/12/01 or later.
%%
%% The list of all files belonging to the 'Elsarticle Bundle' is
%% given in the file `manifest.txt'.
%%
%% Template article for Elsevier's document class `elsarticle'
%% with harvard style bibliographic references
\documentclass[preprint,12pt]{elsarticle}
\usepackage{a4wide}
%% Use the option review to obtain double line spacing
%% \documentclass[preprint,review,12pt]{elsarticle}
%% Use the options 1p,twocolumn; 3p; 3p,twocolumn; 5p; or 5p,twocolumn
%% for a journal layout:
%% \documentclass[final,1p,times]{elsarticle}
%% \documentclass[final,1p,times,twocolumn]{elsarticle}
%% \documentclass[final,3p,times]{elsarticle}
%% \documentclass[final,3p,times,twocolumn]{elsarticle}
%% \documentclass[final,5p,times]{elsarticle}
%% \documentclass[final,5p,times,twocolumn]{elsarticle}
%% For including figures, graphicx.sty has been loaded in
%% elsarticle.cls. If you prefer to use the old commands
%% please give \usepackage{epsfig}
%% The amssymb package provides various useful mathematical symbols
%\usepackage{amssymb}
%% The amsmath package provides various useful equation environments.
%\usepackage{amsmath}
\usepackage{type1cm} % activate if the above 3 fonts are
% not available on your system
%
\usepackage{makeidx} % allows index generation
\usepackage{graphicx} % standard LaTeX graphics tool
% when including figure files
\usepackage{multicol} % used for the two-column index
\usepackage[bottom]{footmisc}% places footnotes at page bottom
\usepackage{bbm} % voor \N, \R, \C, \Q in blackboard-bold
\usepackage{csquotes}
\usepackage{newtxtext} %
\usepackage{bm}
\usepackage{bbm}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{amsthm}
\theoremstyle{definition}
\newtheorem{thm}{Theorem}[section]
%\newtheorem{thma}{Theorem A.}
%\newtheorem{defn}{thm}[section]
\newtheorem{lem}{Lemma}[section]
\newtheorem{cor}{Corollary}[section]
\newtheorem{prop}{Proposition}[section]
\newtheorem{rem}{Remark}[section]
%\newtheorem{rema}{Remark A.}
\usepackage{graphicx}
%\usepackage{newtxmath} % selects Times Roman as basic font
\usepackage[title]{appendix}
\usepackage{empheq}
%\usepackage{natbib}
%\usepackage[square,numbers]{natbib}
\usepackage{subcaption}
\newcommand{\R}{\mathbbm{R}} % reele getallen
\newcommand{\N}{\mathbbm{N}} % natuurlijke getallen
\newcommand{\Z}{\mathbbm{Z}} % gehele getallen
\newcommand{\C}{\mathbbm{C}} % complexe getallen
\newcommand{\Q}{\mathbbm{Q}} % rationale getallen
%\usepackage[left]{lineno}
%\linenumbers
% see the list of further useful packages
% in the Reference Guide
\makeindex % used for the subject index
% please use the style svind.ist with
% your makeindex program
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\usepackage{hyperref}
\begin{document}
\begin{frontmatter}
\title{A Stabilized Finite Element Method for Morpho-Visco-Poroelastic Model}
\author[label1,label2]{S. Asghar\corref{cor1}}
\ead{sabia.asghar@uhasselt.be}
\author[label3]{D.R. den Bakker}
\author[label4]{E. Javierre}
\author[label5,label6]{Q. Peng}
\author[label1,label2]{F.J. Vermolen}
\cortext[cor1]{Corresponding author}
\affiliation[label1]{organization={Computational Mathematics Group (CMAT), Department of Mathematics and Statistics, University of Hasselt},addressline={Diepenbeek}, country={Belgium}}
\affiliation[label2]{organization={Data Science Institute (DSI), University of Hasselt},addressline={Diepenbeek}, country={Belgium}}
\affiliation[label3]{organization={Delft Institute of Applied Mathematics, Delft University of Technology}, addressline={Delft}, country={The Netherlands}}
\affiliation[label4]{organization={IUMA and Applied Mathematics Department, University of Zaragoza}, addressline={Zaragoza}, country={Spain}}
\affiliation[label5]{organization={Mathematics for AI in Real-world Systems, School of Mathematical Sciences, Lancaster University}, addressline={Lancaster}, country={UK}}
\affiliation[label6]{organization={Mathematical Institute, Leiden University}, addressline={Leiden}, country={The Netherlands}}
%% Abstract
\begin{abstract}
We propose a mathematical model that combines elastic, viscous and porous effects with growth or shrinkage due to microstructural changes. This phenomenon is important in tissue or tumor growth, as well as in dermal contraction. Although existence results of the solution to the problem are not given, the current study assesses stability of the equilibria for both the continuous and semi-discrete versions of the model. Furthermore, a numerical condition for monotonicity of the numerical solution is described, as well as a way to stabilize the numerical solution so that spurious oscillations are avoided. The derived stabilization result is confirmed by computer simulations. In order to have a more quantitative picture, the total variation has been evaluated as a function of the stabilization parameter.
\end{abstract}
%% Keywords
\begin{keyword}
Morpho-viscoporoelasticity \sep Stability of Equilibria \sep Monotonicity \sep Stabilized FEM
\end{keyword}
\end{frontmatter}
\section{Introduction}
Biological tissues are complex materials that consist of a solid matrix, mainly composed of cells and the extracellular matrix, permeated by interstitial fluids that play a crucial role in their biomechanical response \cite{Simon1990}. They are frequently subjected to relatively high mechanical stresses and respond with a non-linear relationship between stress and strain, in particular at large deformations \cite{Taylor2020}. In addition, soft tissues are capable of growth, remodeling, and adaptation \cite{Rodriguez1994,Humphrey2002}, processes that generate residual stresses and modify the mechanical configuration of the tissue independently of external loading.
Capturing these intricate phenomena requires the development of mathematical frameworks capable of describing the coupled interaction between the solid and fluid phases.
Poroelastic models have emerged as a powerful tool for characterizing the mechanical behavior of fluid-saturated biological tissues. Originally developed by Biot in the context of soil mechanics \cite{Biot1941}, the theory of poroelasticity has been extensively adapted to describe biological systems such as articular cartilage \cite{Mow1980}, intervertebral discs, brain tissue, and tumor microenvironments \cite{BaxterJain1989,NettiETAL1995}. The fundamental premise of poroelasticity lies in the recognition that mechanical deformation and fluid transport are inherently coupled phenomena: mechanical loading induces fluid flow through the porous solid matrix, while fluid pressure gradients, in turn, generate mechanical stresses that deform the tissue. This bidirectional coupling between fluid dynamics and solid mechanics is essential for understanding numerous physiological and pathological processes. For instance, the time-dependent mechanical response of cartilage under compressive loading—characterized by creep and stress relaxation—arises from the gradual redistribution of interstitial fluid within the tissue \cite{Mow1980}. Similarly, the accumulation of interstitial fluid pressure in solid tumors, resulting from abnormal vascular leakage and impaired lymphatic drainage, significantly affects tumor mechanics and can impede drug delivery \cite{JainETAL2014,HeldinETAL2004}. While poroelastic models successfully capture the coupled fluid-solid mechanics of biological tissues, they assume that the solid matrix exhibits purely elastic behavior, which may be insufficient to describe the time-dependent mechanical response of certain soft tissues. Poro-viscoelastic models extend the classical poroelastic framework by incorporating the intrinsic viscoelastic properties of the solid matrix, thereby accounting for both the viscous drag associated with fluid flow through the porous medium and the inherent viscous dissipation within the solid skeleton itself \cite{Mak1986}. Mathematical models based on (visco-)poroelastic theory provide a quantitative framework to predict these coupled phenomena, offering insights that are difficult to obtain through experimental observation alone.
The numerical solution of Biot's consolidation model presents significant challenges related to the stability of finite element discretizations. Standard Galerkin formulations often exhibit spurious oscillations in the pressure field, particularly when strong pressure gradients occur or when the permeability is small relative to the mesh size \cite{Aguilar2008,BergerETAL2015}. These spurious oscillations arise from the failure of certain finite element pairs to satisfy the inf-sup (or LBB) condition uniformly with respect to the physical parameters of the problem. Several stabilization strategies have been developed to address these difficulties. One approach involves the use of finite element spaces that satisfy appropriate inf-sup conditions, such as the classical Taylor-Hood elements or the MINI element \cite{MuradLoula1994}. While these methods provide theoretical stability, they may still exhibit oscillatory behavior in the presence of very sharp boundary layers or when dealing with materials of low permeability \cite{RodrigoETAL2016}. The key issue here is that the algebraic equation for the pressure should contain an {\em M-matrix}. An alternative stabilization technique is based on perturbation of the variational formulation. Aguilar et al. \cite{Aguilar2008} proposed adding a time-dependent artificial term to the flow equation, with a stabilization parameter depending on the elastic properties of the solid and the mesh size. This parameter has been shown to be optimal in the one-dimensional case. The method provides oscillation-free solutions independently of the chosen discretization parameters while maintaining the use of linear finite elements for both displacements and pressure. Furthermore, Rodrigo et al. \cite{RodrigoETAL2018} introduced a stabilization approach for the three-field formulation of Biot's model based on enriching the piecewise linear continuous finite element space for displacement with edge or face bubble functions. By applying a consistent perturbation of the bilinear form, the bubble functions can be eliminated locally, resulting in a stable scheme with the same number of degrees of freedom as the classical P1-RT0-P0 discretization. This approach ensures uniform stability with respect to both physical and discretization parameters. Similarly, Berger et al. \cite{BergerETAL2015} developed a stabilized conforming mixed finite element method for the three-field formulation using the lowest possible approximation order. Their method incorporates a local pressure jump stabilization term in the mass conservation equation, leading to a symmetric linear system while ensuring stability and avoiding pressure oscillations.
Next to poroelastic effects, many tissues are subject to growth or shrinkage as a result of microstructural changes. These microstructural changes are often caused by cells that adjust their direct environment. Examples are cancer cells that are able to secrete large amounts of fibrous tissue that possibly has different mechanical properties from the fibrous tissue of the original (embryonic) tissue found in the organs, or myofibroblasts that change the structure and orientation of collagen. These changes may lead to changes in mechanical properties, as well as to the occurrence of permament (residual) stresses and strains in the tissue. An important example is an occurrence of nasty dermal contractures that may occur as a result of a burn injury or other deep tissue injury. For this reason, there is a need of mathematical models that not only incorporate the porous, viscous and elastic nature of tissues, but also deal with microstructural changes that lead to permanent deformations, such as growth or shrinkage, and permanent strains. An important model that combines microstructural changes with mechanical deformations is based on morphoelasticity. This model was first formulated by Rodriguez {\em et al.} \cite{Rodriguez1994}, and later applied and clearly described by Goriely \cite{Goriely2015}. Hall \cite{Hall2017} extended the morphoelastic formalism to multiple spatial dimensions. Despite the model combining mechanical loading with microstructural changes is already very useful for tumor growth, tissue growth, organ development of skin contraction, the porous nature of tissue has not been taken into account. For this reason, we combine the poroelastic nature with morphoelasticity to build a more complete model. We remark that the current model is based on linear elasticity with infinitesimal strain. In future studies, we will relax these formulations so that large deformations can be treated in a physically more sound manner.
\section{The mathematical model}
Morphoelasticity allows to model elastic growth and thus capture permanent deformations resulting from tissue growth or shrinkage. While morphoelasticity shares some similarities with classical plasticity, it is fundamentally distinct. Unlike plasticity, where deformation occurs only in regions exceeding a yield stress, morphoelastic remodeling typically takes place throughout the entire tissue. Additionally, morphoelasticity can involve changes in the total tissue mass, such as during growth, and may also lead to increase in internal energy. In contrast, plastic flow is always mass-conserving and inherently dissipative, whereas morphoelasticity models microstructural changes. In this study we consider the evolution equation of the strain tensor obtained in \cite{Hall2017, Hall2008}.
\par The morpho-viscoporoelastic model reads as:
\begin{subequations} \label{morphoporoelas}
\begin{empheq}[left=\empheqlbrace]{align}
& \frac{D}{Dt} (\rho {\bf w}) + \rho (\nabla \cdot {\bf w}) {\bf w} - \nabla \cdot \boldsymbol{\sigma}({\bf w},\boldsymbol{\varepsilon}) + \nabla p = {\bf f}_{\bf u}, \label{mvpe_momentum}\\[1ex]
& \boldsymbol{\sigma}({\bf w},\boldsymbol{\varepsilon}) = \boldsymbol{\sigma}_{el}(\boldsymbol{\varepsilon}) + \boldsymbol{\sigma}_{vis}({\bf w}),
\label{mvpe_stress} \\[1ex]
& \nabla \cdot {\bf w} - \nabla \cdot (\kappa \nabla p) = f_p, \label{mvpe_continuity} \\[1ex]
& \frac{D \boldsymbol{\varepsilon}}{Dt} + \boldsymbol{\varepsilon} ~ \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \boldsymbol{\varepsilon} + (\text{tr}(\boldsymbol{\varepsilon}) - 1) \text{sym}(\nabla {\bf w}) = -\mathbf{G}, \label{mvpe_strain} \\[1ex]
& \frac{D {\bf u} }{Dt} = {\bf w}, \label{mvpe_dudt}
\end{empheq}
\end{subequations}
in $\Omega(t)\times (0,T]$, where
\begin{align}
\boldsymbol{\sigma}_{el}(\boldsymbol{\varepsilon}) = 2\mu \boldsymbol{\varepsilon} + \lambda ~\text{tr}(\boldsymbol{\varepsilon}) {\bf I},
\quad
\boldsymbol{\sigma}_{vis}({\bf w}) = \mu_1 ~ \text{sym}(\nabla {\bf w}) + \mu_2 ~ \text{tr}(\text{sym}(\nabla {\bf w})) \mathbf{I}. \tag*{(1f)}
\label{mvpe_se_sv}
\end{align}
Here ${\bf u}$ and $p$, respectively, represent the local displacement of the skeleton and the pore pressure, ${\bf w}$ the displacement velocity, $\boldsymbol{\varepsilon}$ the strain tensor, ${\bf G}$ the net growth tensor, and $\kappa$ represents the ratio of the permeability of the porous medium and the fluid viscosity. Further, $D(\cdot)/Dt$ denotes the material derivative, and $\text{sym}(.)$ and $\text{skw}(.)$, respectively, denote the symmetric and skew-symmetric part of a matrix or tensor. Eq. (\ref{mvpe_momentum}) represents the balance of momentum where inertial terms are taken into account, $\rho$ denotes the tissue density, $\boldsymbol{\sigma}$ denotes the effective stress tensor and the right-hand term ${\bf f}_{\bf u}$ represents applied body force(s). Eq. (\ref{mvpe_stress}) accounts for the elastic and viscoelastic behavior of the material, with their respective constitutive relations given in \ref{mvpe_se_sv}. There $\lambda$ and $\mu$ denote the Lam\'e parameters, $\mu_1$ and $\mu_2$ the viscosity coefficients and $\mathbf{I}$ the identity tensor. Eq. (\ref{mvpe_continuity}) represents the conservation equation for the fluid, where Darcy's law,
\begin{align*}
{\bf v} + \kappa ~ \nabla p = 0,
\end{align*}
has already been incorporated. Here ${\bf v}$ represents the fluid flow velocity. In Eq. (\ref{mvpe_continuity}) the source term $f_p$ represents a forced fluid extraction or injection process. Eq. (\ref{mvpe_strain}) represents the evolution equation of the strain tensor. Note that the primary variables in Eqs. (\ref{morphoporoelas}) are ${\bf w}$, $\boldsymbol{\varepsilon}$ and $p$. The displacement ${\bf u}$ can be obtained by integrating ${\bf w}$ over one time step as a post-processing step.
Note that Eq. (\ref{mvpe_strain}) models the evolution of the strain of the tissue as a result of microstructural changes. This will lead to permanent (local) displacements since the ${\bf G}$-tensor quantifies either growth or shrinkage of the medium as a result of microstructural changes. If ${\bf G} = {\bf 0}$, then the model does not predict any permanent displacements.
The model needs to be completed with appropriate boundary and initial conditions. Unless stated otherwise, we consider:
\begin{subequations} \label{mpve_ibc}
\begin{empheq}[left=\empheqlbrace]{align}
& {\bf w} = \mathbf{0}, \quad \kappa \nabla p \cdot \mathbf {n} = g_N , & \text{on $\Gamma_1(t) \times (0,T]$,} \label{mpve_bc1} \\[1ex]
& \boldsymbol{\sigma}\cdot\boldsymbol{n} = \mathbf{f}_b, \quad p = p_0, & \text{on $\Gamma_2(t) \times (0,T]$,} \label{mpve_bc2} \\[1ex]
& {\bf w}({\bf x},0) = {\bf 0}, & \text{in $\Omega(0)$,} \label{mpve_ic1} \\[1ex]
& {\boldsymbol \varepsilon}({\bf x},0) = {\boldsymbol 0}, & \text{in $\Omega(0)$,} \label{mpve_icw}
\end{empheq}
\end{subequations}
where $g_N$ represents the fluid flow on $\Gamma_1$, ${\bf f}_b$ represents an external force on the boundary and $p_0$ denotes the environmental atmospheric pressure, which is taken constant.
We note that if $\kappa$ is a constant (not depending on the position or time) and if the fluid flow velocity stays finite, then from Darcy's Law heuristically it follows that $\nabla p \longrightarrow {\bf 0}$ as $\kappa \longrightarrow \infty$. This implies that $p = c$, for some $c \in \R$. Since $p = p_0$ on $\Gamma_2(t)$, it follows that $p = p_0$ in $\Omega$.
It is to note that the standard morphoelastic formulation \cite{EgbertsStab2021} is recovered when $\kappa \to +\infty$ in Eqs. \eqref{mpve_ibc}.
\section{Stability of the model}
The term "stable" informally means resistant to change. For technical use, the term has to be defined
more precisely in terms of the mathematical model, but the same connotation applies. Stability analysis is crucial to ensuring the reliability and accuracy of the numerical methods used to solve these problems. In this section, we discuss the stability of the morpho-viscoporoelastic model. We first prove the symmetry of the strain tensor which further leads to the stability of the model. Then we assess the \textit{linear} stability of the steady state problems using the Fourier series.
\subsection{Symmetry of the strain tensor}
We demonstrate that the strain tensor will remain symmetric for all later times, if it is initially symmetric. Before we prove the result on the symmetry of $\boldsymbol{\varepsilon}$, recall the following operations with tensors. If $A,B\in \mathbb{R}^{n\times n}$, then
\begin{align}
A : B = \sum_{i,j = 1}^n A_{ij}B_{ij}.
\label{ten-scal-prod1}
\end{align}
Multiplication of the matrices $A^T$ and $B$ gives component-wisely
$$\displaystyle (A^T B)_{ij} = \sum_{k=1}^n A_{ki} B_{kj}$$ and consequently
\begin{align}
\text{tr}(A^T B) = \sum_{i=1}^n (A^T B)_{ii} = \sum_{i=1}^n \sum_{k=1}^n A_{ki} B_{ki} = A : B.
\label{ten-scal-prod2}
\end{align}
\begin{lem}\label{lemma1}
Let ${\bf v}, {\bf L} \in \mathbb{R}^{d \times d}$ be d-dimensional tensors. Suppose that ${\bf L}$ is skew-symmetric (${\bf L}^T = -{\bf L}$), then for all ${\bf v} \in \mathbb{R}^{d \times d}$ the tensorial scalar product satisfies
\begin{align*}
{\bf v} : ({\bf L} {\bf v}) = {\bf v} : ({\bf v} {\bf L} ) = 0.
\end{align*}
\end{lem}
\noindent {\bf Proof.}
Choose any $\boldsymbol{v} \in \mathbb{R}^{d \times d}$, use Eq. (\ref{ten-scal-prod2}) and ${\bf L}^T = -{\bf L}$: %, $(AB)^T = B^T A^T$, and equation (\ref{ten-scal-prod2}), then we arrive at
\begin{align*}
\boldsymbol{v} : (\boldsymbol{L} \boldsymbol{v})
=\text{tr}(\boldsymbol{v}^T \boldsymbol{L}\boldsymbol{v})
=-\text{tr}(\boldsymbol{v}^T\boldsymbol{L}^T\boldsymbol{v})
=-\text{tr}((\boldsymbol{L}\boldsymbol{v})^T\boldsymbol{v})
=-\boldsymbol{v} : (\boldsymbol{L}\boldsymbol{v}).
\end{align*}
Hence $\displaystyle{ {\boldsymbol{v}} : ({\bf L} {\boldsymbol{v}}) = 0 }.$ \\
Furthermore, since $A : B = A^T : B^T$, take $\boldsymbol{w} = \boldsymbol{v}^T$ and we use ${\bf L}^T=-{\bf L}$: %, we get using the above relation:
\begin{align*}
(\boldsymbol{w} \boldsymbol{L}) : \boldsymbol{w} = (\boldsymbol{w} \boldsymbol{L})^T : \boldsymbol{w}^T = (\boldsymbol{L}^T \boldsymbol{v}) : \boldsymbol{v} = -(\boldsymbol{L} \boldsymbol{v}) : \boldsymbol{v} = -\boldsymbol{v} : (\boldsymbol{L} \boldsymbol{v}) = 0.
\end{align*}
This proves the lemma. \hfill $\Box$ \\[2ex]
Note that the above products can be generalized to scalar products of anti-Hermitian operators on Hilbert spaces. Using this fact, the only thing that remained to be done in an alternative proof was demonstrating that ${\bf v} : ({\bf w} L)$ was a proper inner product. Based on the above lemma, we demonstrate the following claim:
\begin{thm}\label{Th_aux}
Let $\bf{G}$ be a symmetric tensor for $t\geq 0$ and $\boldsymbol{\varepsilon}$ satisfy
\begin{align}
\frac{D \boldsymbol{\varepsilon}}{Dt} + \boldsymbol{\varepsilon} ~ \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \boldsymbol{\varepsilon} + (\text{tr}(\boldsymbol{\varepsilon}) - 1) \text{sym}(\nabla {\bf w}) = -\bf{G}
\label{eps}
\end{align}
in an open Lipschitz domain $\Omega\in\mathbb{R}^d$ for $t>0$. Suppose that $\boldsymbol{\varepsilon}$ is symmetric on $t=0$, then $\boldsymbol{\varepsilon}$ remains symmetric for $t>0$.
\end{thm}
\noindent {\bf Proof}. Taking the transpose of Eq. (\ref{eps}) %
and using that $\text{skw}(\nabla{\bf w})^T=-\text{skw}(\nabla{\bf w})$ and $\text{sym}(\nabla{\bf w})^T=\text{sym}(\nabla{\bf w})$ we obtain
\begin{align}
\frac{D \boldsymbol{\varepsilon}^T}{Dt} + \boldsymbol{\varepsilon}^T \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \boldsymbol{\varepsilon}^T + (\text{tr}(\boldsymbol{\varepsilon}) - 1) \text{sym}(\nabla {\bf w}) = -\bf{G}^T.
\label{eps_T}
\end{align}
Subtraction of Eqs. (\ref{eps}) and (\ref{eps_T}) gives
\begin{align*}
\frac{D \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right)}{Dt} + \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right) ~ \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right) = - \left( \bf{G} - \bf{G}^T \right).
\end{align*}
If the growth tensor $\bf G$ remains symmetric on time, then we have
\begin{align}
\frac{D \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right)}{Dt} + \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right) ~ \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \left( \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T \right) = 0.
\label{aux1}
\end{align}
From this equation it is clear that $\boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T = 0$ represent an equilibrium solution. We furthermore can prove that $\boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T = 0$ is the only solution provided that $\boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T = 0$ is fulfilled at $t=0$. \\
\noindent Taking $\boldsymbol{v} = \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T $ and $\bf{L} = \text{skw}(\nabla {\bf w})$, Eq. (\ref{aux1}) can be rewritten as
\begin{align*}
\frac{D \boldsymbol{v} }{Dt} + \boldsymbol{v} ~ \bf{L} - \bf{L} ~ \boldsymbol{v} = 0.
\end{align*}
Then we have that
\begin{align*}
\boldsymbol{v} : \frac{D \boldsymbol{v} }{Dt} + \boldsymbol{v} : \left( \boldsymbol{v} \bf{L} \right) - \boldsymbol{v} : \left( \bf{L} \boldsymbol{v} \right) = 0.
\end{align*}
Using Lemma \ref{lemma1}, the previous equation reduces to
\begin{align*}
\boldsymbol{v} : \frac{D \boldsymbol{v} }{Dt} = 0.
\end{align*}
Define $||\boldsymbol{v}||^2 := \boldsymbol{v} : \boldsymbol{v}$, then it follows that
\begin{equation*}
\frac12 \frac{{D}}{{D}t} \, ||\boldsymbol{v}||^2 = 0
\end{equation*}
from which we obtain that $||\boldsymbol{v}({\bf x},t)||^2= ||\boldsymbol{v}({\bf x},0)||^2$ for any $t>0$. This proves that if
$\boldsymbol{\varepsilon}$ is symmetric on $t=0$ (i.e. $\boldsymbol{v}({\bf x},0) = 0$) then $\boldsymbol{\varepsilon}$ is symmetric for $t>0$.
\hfill $\Box$ \\
The previous result can be further generalized to a broader class of growth tensors.
Further, we demonstrate that small perturbations around symmetric strain tensor remain small, which is a characteristic of stability.
\begin{thm}\label{Theorem5}
Let $\boldsymbol{\varepsilon}$ satisfy
\begin{align*}
\frac{D \boldsymbol{\varepsilon}}{Dt} + \boldsymbol{\varepsilon} ~ \text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w}) \boldsymbol{\varepsilon} + (\text{tr}(\boldsymbol{\varepsilon}) - 1) \text{sym}(\nabla {\bf w}) = -\alpha \boldsymbol{\varepsilon},
\end{align*}
in an open Lipschitz domain $\Omega\in\mathbb{R}^d$ for $t>0$. Suppose that $\boldsymbol{\varepsilon}$ is symmetric on $t=0$, then $\varepsilon$ remains symmetric for $t>0$, and symmetry is asymptotically stable with respect to perturbations if and only if $\alpha>0$ (and stable if and only if $\alpha\geq 0$).
\end{thm}
\noindent {\bf Proof}. Reproducing the steps of the proof of Theorem \ref{Th_aux} we obtain the following equation for $\boldsymbol{v} = \boldsymbol{\varepsilon} - \boldsymbol{\varepsilon}^T $:
\begin{align*}
\frac{D \boldsymbol{v} }{Dt} + \boldsymbol{v} ~ \bf{L} - \bf{L} ~ \boldsymbol{v} = -\alpha \boldsymbol{v}
\end{align*}
where $\bf{L} = \text{skw}(\nabla {\bf w})$. Using the tensor product and Lemma \ref{lemma1} we obtain that
\begin{equation*}
\boldsymbol{v} : \frac{{D}}{{D}t} \, \boldsymbol{v} = -\alpha\,\boldsymbol{v} : \boldsymbol{v}
\end{equation*}
and
\begin{equation*}
\frac12 \frac{{D}}{{D}t} \, ||\boldsymbol{v}||^2 = -\alpha \, ||\boldsymbol{v}||^2.
\end{equation*}
Integrating over time from $t = 0$ and using $\boldsymbol{v} = 0$ at $t = 0$ and $\alpha \geq 0$, gives
\begin{equation}
0 \le \frac{1}{2} ||\boldsymbol{v}||^2 = - \int_0^t \alpha \, ||\boldsymbol{v}||^2 \, \mathrm{d}s \le 0.
\end{equation}
This implies that $||\boldsymbol{v}|| = 0$ on $t > 0$ if $||\boldsymbol{v}|| = 0$ on $t = 0$. Hence $\boldsymbol{v} = 0$ for $t > 0$, which represents symmetry, is the only possibility if $\boldsymbol{v} = 0$ on $t = 0$. Further, stability is also proved by this argument.
\hfill $\Box$
\subsection{Stability of the steady state}
Let us consider the one-dimensional version of Eqs. \eqref{mpve_ibc} on the fixed domain $\Omega = (0,1)$. If we further assume $\rho$ to remain constant, $G = \alpha \varepsilon$ ($\alpha \ge 0$), $f_u = 0$ and $f_p = 0$, we obtain:
\begin{subequations} \label{mpve_1D}
\begin{empheq}[left=\empheqlbrace]{align}
& \rho \left [\frac{\partial w}{\partial t} + 2w \frac{\partial w}{\partial x} \right ]- E \frac{\partial \varepsilon}{\partial x} - \mu_v \frac{\partial^2 w}{\partial x^2} + \frac{\partial p}{\partial x}= 0, \\%[1ex]
& \frac{\partial w}{\partial x} - \frac{\partial}{\partial x}\left(\kappa \frac{\partial p}{\partial x}\right) = 0, \\[0.5ex]
& \frac{\partial \varepsilon}{\partial t} + (\varepsilon - 1) \frac{\partial w}{\partial x} = -\alpha \varepsilon, \\%[1ex]
& \frac{\partial u }{\partial t} = w,
\end{empheq}
\end{subequations}
where $E = 2\mu+\lambda$ and $\mu_v = \mu_1+\mu_2$.
For simplicity we take the following boundary and initial conditions:
\begin{subequations} \label{mpve_ibc_1D}
\begin{empheq}[left=\empheqlbrace]{align}
& w = 0, \quad \kappa \frac{\partial p}{\partial x} = 0, & \text{on $\Gamma_1 \times (0,T]$,} \\%[1ex]
& \frac{\partial \sigma}{\partial x} = f_b, \quad p = p_0, & \text{on $\Gamma_2 \times (0,T]$,} \\%[1ex]
& w(x,0) = 0, & \text{in $(0,1)$,} \\%[1ex]
& \varepsilon(x,0) = 0, & \text{in $(0,1)$.} %\label{mpve_icw}
\end{empheq}
\end{subequations}
The steady state solutions of the one-dimensional model, Eqs. (\ref{mpve_1D})-(\ref{mpve_ibc_1D}) are $(\overline{w},\overline{\varepsilon},\overline{p}) = (0,0,p_0)$, where $p_0$ was introduced earlier as a constant.
In the following subsections we analyze the linear stability of the steady states of the one-dimensional model. The analysis that we perform is commonly known as Von Neumann stability analysis in cases of equidistant meshes and constant coefficients, which is based on decomposition of motion into normal modes, often using Fourier analysis, and superposition. The Von Neumann stability analysis provides sufficient and necessary conditions for numerical stability \cite{Fletcher1998}. The analysis looks at the growth or
decay of perturbations from one step to the next, and can be implemented using standard linear algebraic
procedures. A more severe restriction is that it strictly applies only to linear systems. Despite this limitation, it is frequently applied to nonlinear systems through linearization.
\subsubsection{Stability of steady states in the continuous problem}
In this subsection, we analyze the {\em linear} stability of the one-dimensional model (\ref{mpve_1D})-(\ref{mpve_ibc_1D}) %(\ref{morphoporoelas})
with respect to sinusoidal perturbations around the equilibrium using Fourier series. We do this analysis in order to understand the a-priori behavior of the solution.
\par {From now on, we make the following considerations:
\begin{itemize}
\item For simplicity, we take $\Omega = (0,1)$.
\item The stability conditions will be formulated in terms of the input parameters.
\item We indicate the equilibria by overlines and the perturbations by hats and proceed as in \cite{EgbertsStab2021}.
\end{itemize}}
\noindent Let us consider the following perturbations %(denoted by hats)
of the steady state solutions
\begin{equation}
{w=\overline{w}+\hat{w}=\hat{w}},\quad \varepsilon=\overline{\varepsilon}+\hat{\varepsilon}=\hat{\varepsilon},\quad p=\overline{p}+\hat{p} = p_0+\hat{p}.
\end{equation}
The linearised equations around the equilibria ${\overline{w}}=0,\ \overline{\varepsilon}=0$, and $\overline{p} = p_0$, are as follows
\begin{equation}
\left\{
\begin{array}{ll}
\displaystyle{ \rho \frac{\partial\hat{w}}{\partial t}-\mu_v\frac{\partial^2 \hat{w}}{\partial x^2}-E\frac{\partial\hat{\varepsilon}}{\partial x}+\frac{\partial\hat{p}}{\partial x}=0,} \\[1.5ex]
\displaystyle{\frac{\partial\hat{w}}{\partial x} =\kappa \frac{\partial^2 \hat{p}}{\partial x^2},} \\[1.5ex]
\displaystyle{\frac{\partial\hat{\varepsilon}}{\partial t}-\frac{\partial\hat{w}}{\partial x}=-\alpha \hat\varepsilon,} \\[1.5ex]
\end{array}
\right.
\label{S1Dmorphoporoelas}
\end{equation}
where %the variations
$\hat{w}, \hat{\varepsilon}\ \text{and}\ \hat{p}$ are written in terms of complex Fourier series:
\begin{equation}
f(x,t)= \sum_{k=-\infty}^{\infty}{\hat{f}_{k}(t) e^{2 i \pi k x}},
\label{Var}
\end{equation}
with $i$ being the imaginary unit. % number with $i^2=-1$.
For the computations in the proof of this analysis, see \cite{Asghar12025}. Linear stability is guaranteed if and only if
\begin{align}
4\pi^2 l^2 (\alpha \mu_v + E) + \dfrac{\alpha}{\kappa} \ge 0, \qquad 4\pi^2 l^2 \mu_v + \dfrac{1}{\kappa} + \alpha \rho \ge 0,
\end{align}
which implies linear stability for all modes if and only if $\alpha \ge 0$.
The findings of the analysis are summarised below:
\\[1ex]
\begin{thm}\label{Theorem6}
Let $\alpha \ge 0$, then the equilibria $(\overline w,\overline{\varepsilon},\overline p) = (0,0,p_0)$ in the one-dimensional (continuous) morpho-viscoporoelastic model are linearly stable.
\end{thm}
\subsubsection{Stability of steady states in the semi-discrete problem}
Next, we analyze the stability of the semi-discrete problem under the same consideration as above. Let $h$ be the gridsize, taken uniform, and $k$ denote the index of the meshpoint positioned at $k h$. We consider $n$ nodes in total, hence $n h = 1$. The finite difference method (FDM), based on central differences for the first- and second-order spatial derivatives, gives:
\begin{equation}
\left\{
\begin{array}{ll}
\displaystyle{{w_k}^{'}(t) = \frac{\mu_v}{\rho} \frac{w_{k-1}-2 w_k+w_{k+1}}{h^2}+\frac{E}{\rho}\frac{{\varepsilon_{k+1}-\varepsilon_{k-1}}}{2 h}-\frac{1}{\rho}\frac{p_{k+1}-p_{k-1}}{2 h},} \\[1.5ex]
\displaystyle{\frac{w_{k+1}-w_{k-1}}{2 h}-\kappa\frac{p_{k-1}-2 p_k+p_{k+1}}{h^2}=0,} \\[1.5ex]
\displaystyle{{\varepsilon_k}^{'}(t) = \frac{w_{k+1}-w_{k-1}}{2 h}-\alpha\varepsilon_k.} \\[1.5ex]
\end{array}
\right.
\label{Dis1D}
\end{equation}
Let
\begin{equation}
w_k= \sum_{j=1}^{n-1}{\hat{w}_{j}(t) e^{-2 i \pi k j h}},
\quad
\varepsilon_k= \sum_{j=1}^{n-1}{\hat{\varepsilon}_{j}(t) e^{-2 i \pi k j h}},
\quad
p_k= \sum_{j=1}^{n-1}{\hat{p}_{j}(t) e^{-2 i \pi k j h}}.
\label{VarDist1}
\end{equation}
Substitution of the variations (\ref{VarDist1}) into the simplified linearised equations (\ref{Dis1D}) gives
\begin{equation}
\left\{
\begin{array}{ll}
\displaystyle{\sum_{j=1}^{n-1} {\hat{w}_j^{'}(t)}= \frac{\mu_v}{h^2\rho}\sum_{j=1}^{n-1} \hat{w}_j(t) \{e^{-2 i \pi (k-1) j h }-2 e^{-2i\pi kjh}+ e^{-2i\pi (k+1)jh}\}}\\[1.5ex]
+\displaystyle{\frac{E}{2h\rho}\sum_{j=1}^{n-1} \hat{\varepsilon}_j(t) \{e^{-2i \pi (k+1)jh }-{e^{-2i\pi (k-1)jh}}\}
-\frac{1}{2h\rho}\sum_{j=1}^{n-1} \hat{p}_j(t) \{e^{-2i \pi (k+1)jh }-{e^{-2i\pi (k-1)jh}}\}},\\[1.5ex]
\displaystyle{\frac{1}{2h}\sum_{j=1}^{n-1} \hat{w}_j(t) \{e^{-2 i \pi (k+1) j h }- e^{-2i\pi (k-1)jh}\}-\frac{\kappa}{h^2}\sum_{j=1}^{n-1} \hat{p}_j(t) \{e^{-2 i \pi (k-1) j h }-2 e^{-2i\pi kjh}+ e^{-2i\pi (k+1)jh}\}=0}, \\[1.5ex]
\displaystyle{\sum_{j=1}^{n-1}{\hat{\varepsilon}_j}^{'}(t) = \frac{1}{2h}\sum_{j=1}^{n-1} \hat{w}_j(t) \{e^{-2 i \pi (k+1) j h }-e^{-2i\pi (k-1)jh}\}-\alpha \sum_{j=1}^{n-1} \hat{\varepsilon}_j(t) e^{-2 i \pi k j h }.} \\[1.5ex]
\end{array}
\right.
\label{Dis1DFourier}
\end{equation}
Multiplication of the above set of equations by $ e^{2 i \pi k l h }$, $l\in \{1,\ldots,n-1\}$, using Euler's Formula and $2\cos{(2 \pi l h)-2}=-4 \sin^2{(\pi l h)}$ results in
\begin{equation}
\left\{
\begin{array}{ll}
\displaystyle{ {\hat{w}_l}^{'}(t) = -\frac{4\mu_v}{h^2\rho} \sin^2{(\pi l h)}} ~ \hat{w}_l(t)
-\frac{i E}{h\rho} \sin{(2 \pi l h)} ~ \hat{\varepsilon}_l(t)
+\frac{i}{h\rho} \sin{(2 \pi l h)} ~ \hat{p}_l(t),\\[1.5ex]
\displaystyle{-\frac{i}{h} \sin{(2 \pi lh)}~\hat{w}_l(t)+\frac{4 \kappa}{h^2} \sin^2{(\pi l h)}~\hat{p}_l(t)=0}, \\[1.5ex]
\displaystyle{{\hat{\varepsilon}_l}^{'}(t) =-\frac{i}{h} \sin{(2 \pi l h)}~\hat{w}_l(t)-\alpha ~ \hat{\varepsilon}_l(t).} \\[1.5ex]
\end{array}
\right.
\label{Dis1Dsys}
\end{equation}
After incorporating the value of $\hat{p}_{l}(t)=\frac{ih}{2\kappa} \cot{(\pi l h)} ~ \hat{w}_{l}$ in the above system, we get
\begin{equation}
\setlength{\arraycolsep}{2em}
\frac{d}{dt}
\begin{pmatrix}
\hat{w}_{l}(t) \\ \\[1ex] \hat{\varepsilon}_{l}(t)\\
\end{pmatrix} +
\begin{pmatrix}
\frac{4\mu_v}{h^2 \rho}\sin^2{(\pi lh)}+\frac{1}{\kappa \rho}\cos^2{(\pi lh)} &\frac{iE}{h \rho}{\sin{(2 \pi lh)}} \\ \\[1ex]
\frac{i}{h}\sin{(2\pi lh)} & \alpha \\
\end{pmatrix}
\begin{pmatrix}
\hat{w}_{l}(t) \\ \\[1ex] \hat{\varepsilon}_{l}(t)\\
\end{pmatrix} = \bf{0}.\\[1.5ex]
\end{equation}
The fact that the eigenvalues of the above matrix need to be non-negative (or have a non-negative real part), leads to the following linear stability critera:
\begin{align}
\!\!\! \frac{4\sin^2 (\pi l h)}{h^2} \left[ \mu_v \alpha + E \cos^2(\pi l h)\right] + \frac{\alpha}{\kappa} \cos^2(\pi l h) \ge 0,
\quad
\frac{4\mu_v}{h^2} \sin^2(\pi l h) + \frac{1}{\kappa} \cos^2(\pi l h) + \alpha \rho \ge 0.
\label{eq:stab_discrete}
\end{align}
Hence, we conclude:
\begin{thm}\label{Theorem7}
{
Let $\alpha \ge 0$, then the equilibria $(w,\varepsilon,p) = (0,0,p_o)$ in the one-dimensional semi-discrete morpho-viscoporoelastic model are linearly stable.
}
\end{thm}
\begin{rem}
Recall that
\begin{equation*}
\lim_{h \rightarrow 0} \frac{4\sin^2{(\pi lh)}}{h^2}=4 \pi^2 l^2.
\end{equation*}
Hence, Theorem \ref{Theorem7} is consistent with the stability analysis of the continuous model, Theorem \ref{Theorem6}, for $h\to 0$.
\end{rem}
\begin{rem}
Applying, for instance, an implicit Euler time integration method, will always lead to a stable numerical solution.
\end{rem}
\section{The numerical scheme}
For the sake of presentation, we are dealing with a problem in two spatial dimensions, $d=2$. Then, we can write
\[
{\bf w} = \begin{pmatrix} w^1 \\ w^2 \end{pmatrix},
\qquad
\boldsymbol{\varepsilon} = \begin{pmatrix} \varepsilon^{11} & \varepsilon^{12} \\ \varepsilon^{12} & \varepsilon^{22} \end{pmatrix}.
\]
Furthermore, we limit the presentation to the case ${\bf G}=\alpha \boldsymbol{\varepsilon}$ in Eq. (\ref{mvpe_strain}).
\subsection{Galerkin approximation}
In this section we consider the finite element discretization of the morpho-viscoporoelastic model, Eqs. (\ref{morphoporoelas})-(\ref{mpve_ibc}). In order to obtain the weak formulation of the problem, for each $t\in (0,T]$ we consider the following functional spaces:
\begin{align*}
& \mathcal{W}_t = \left\{ \mathbf{w}(\cdot,t) \in \Big( H^1(\Omega(t)) \Big)^d \,\, \Big| \,\, \mathbf{w}(\cdot, t) = \mathbf{0} \text{ on } \Gamma_1(t) \right\}, \\
& \mathcal{Q}_t = \left\{ p(\cdot,t) \in H^1(\Omega(t)) \,\, \mid \,\, p = p_0 \text{ on } \Gamma_2(t) \right\}, \\
& \mathcal{Q}_{0,t} = \left\{ p(\cdot,t) \in H^1(\Omega(t)) \,\, \mid \,\, p = 0 \text{ on } \Gamma_2(t) \right\}, \\
& \mathcal{E}_t = \left\{ \boldsymbol{\varepsilon}(\cdot,t) \in \Big( L^2(\Omega(t)) \Big)^{d\times d} \,\, \Big| \,\, \boldsymbol{\varepsilon} \; \text{symmetric} \right\}.
\end{align*}
Applying Reynold's Transport Theorem \cite{Reynolds1903}, the weak formulation of (\ref{morphoporoelas})-(\ref{mpve_ibc}) reads: for each $t\in (0,T]$, find $(\mathbf{w}(t), p(t), \boldsymbol{\varepsilon}(t)) \in \mathcal{W}_t \times \mathcal{Q}_t \times \mathcal{E}_t$ such that
\begin{subequations} \label{mvpe_weakform}
\begin{align}
& \rho \frac{d}{dt} \int_{\Omega(t)} {\bf w} \cdot {\boldsymbol \phi} d \Omega + \int_{\Omega(t)} \left( \boldsymbol{\sigma}_{el}(\boldsymbol{\varepsilon}) + \boldsymbol{\sigma}_{vis}(\bf w) \right) : \text{sym}(\nabla {\boldsymbol \phi}) d \Omega - \int_{\Omega(t)} p \nabla \cdot {\boldsymbol \phi} d \Omega \nonumber \\
& \qquad\qquad = \int_{\Omega(t)} {\bf f}_{\bf u} \cdot {\boldsymbol \phi} d \Omega + \int_{\Gamma_2(t)} {\boldsymbol f}_b \cdot {\boldsymbol \phi} d \Gamma - \int_{\Gamma_2(t)} p_0 \boldsymbol{\phi}\cdot \mathbf{n} d\Gamma,
\label{wf_w}
\\[1ex]
& \frac{d}{dt}\int_{\Omega(t)} \boldsymbol{\varepsilon}:\boldsymbol{\zeta} d\Omega
+
\int_{\Omega(t)} \left[ \boldsymbol{\varepsilon}\,\text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w})\,\boldsymbol{\varepsilon}
+
\left( \text{tr}(\boldsymbol{\varepsilon})-1 \right)\,\text{sym}(\nabla {\bf w})
-
(\nabla {\bf w}) \boldsymbol{{\varepsilon}}
\right] : \boldsymbol{\zeta} ~ d\Omega \nonumber \\
& \qquad\qquad + \int_{\Omega(t)} \alpha \boldsymbol{\varepsilon} : \boldsymbol{\zeta} d\Omega = 0
\label{wf_varepsilon}
\\[1ex]
& \int_{\Omega(t)} \psi \nabla \cdot {\bf w} d \Omega + \int_{\Omega(t)} \kappa \nabla p \cdot \nabla \psi d \Omega = \int_{\Omega(t)} f_p \psi d\Omega - \int_{\Gamma_1(t)} g_N \psi d\Gamma
\label{wf_p}
\end{align}
\end{subequations}
for all $(\boldsymbol{\phi}, \psi, \boldsymbol{\zeta}) \in \mathcal{W}_t \times \mathcal{Q}_{0,t} \times \mathcal{E}_t$.
Prior to obtain the Galerkin finite element approximation of the weak formulation, Eqs. \eqref{mvpe_weakform}, let us work out the central term in Eq. \eqref{wf_varepsilon}.
\begin{rem} \label{rem:wf_eps}
Recall that $\boldsymbol{\zeta}\in \mathcal{E}_t$ is symmetric. Let us consider the following choices of $\boldsymbol{\zeta}$:
\begin{itemize}
\item If $\boldsymbol{\zeta} = \begin{pmatrix} \zeta & 0 \\ 0 & 0 \end{pmatrix}$, then
\begin{align*}
& \int_{\Omega(t)} \left[ \boldsymbol{\varepsilon}\,\text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w})\,\boldsymbol{\varepsilon}
+
\left( \text{tr}(\boldsymbol{\varepsilon})-1 \right)\,\text{sym}(\nabla {\bf w})
-
(\nabla {\bf w}) \boldsymbol{{\varepsilon}} \right] : \boldsymbol{\zeta} ~ d\Omega = \\
& \qquad
= \int_{\Omega(t)} \left[ -\varepsilon^{11} \dfrac{\partial w^2}{\partial y} + \varepsilon^{12} \left( \frac{\partial w^2}{\partial x} - \frac{\partial w^1}{\partial y} \right) + \varepsilon^{22} \frac{\partial w^1}{\partial x} \right] \zeta ~d\Omega
-
\int_{\Omega(t)} \frac{\partial w^1}{\partial x} \zeta ~d\Omega.
\end{align*}
\item If $\boldsymbol{\zeta} = \begin{pmatrix} 0 & \zeta \\ 0 & 0 \end{pmatrix}$, then
\begin{align*}
& \int_{\Omega(t)} \left[ \boldsymbol{\varepsilon}\,\text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w})\,\boldsymbol{\varepsilon}
+
\left( \text{tr}(\boldsymbol{\varepsilon})-1 \right)\,\text{sym}(\nabla {\bf w})
-
(\nabla {\bf w}) \boldsymbol{{\varepsilon}} \right] : \boldsymbol{\zeta} ~ d\Omega = \\
& \qquad
= \int_{\Omega(t)} \left[ \varepsilon^{11}\frac{\partial w^1}{\partial y} - \varepsilon^{12}\left( \frac{\partial w^1}{\partial x} + \frac{\partial w^2}{\partial y} \right) + \varepsilon^{22} \frac{\partial w^2}{\partial x} \right] \zeta ~d\Omega
-
\frac{1}{2}\int_{\Omega(t)} \left( \frac{\partial w^1}{\partial y} + \frac{\partial w^2}{\partial x} \right) \zeta ~d\Omega.
\end{align*}
\item If $\boldsymbol{\zeta} = \begin{pmatrix} 0 & 0 \\ 0 & \zeta \end{pmatrix}$, then
\begin{align*}
& \int_{\Omega(t)} \left[ \boldsymbol{\varepsilon}\,\text{skw}(\nabla {\bf w}) - \text{skw}(\nabla {\bf w})\,\boldsymbol{\varepsilon}
+
\left( \text{tr}(\boldsymbol{\varepsilon})-1 \right)\,\text{sym}(\nabla {\bf w})
-
(\nabla {\bf w}) \boldsymbol{{\varepsilon}} \right] : \boldsymbol{\zeta} ~ d\Omega = \\
& \qquad
= \int_{\Omega(t)} \left[ \varepsilon^{11} \dfrac{\partial w^2}{\partial y} - \varepsilon^{12} \left( \frac{\partial w^2}{\partial x} - \frac{\partial w^1}{\partial y} \right) - \varepsilon^{22} \frac{\partial w^1}{\partial x} \right] \zeta ~d\Omega
-
\int_{\Omega(t)} \frac{\partial w^2}{\partial y} \zeta ~d\Omega.
\end{align*}
\end{itemize}
\end{rem}
Let $\mathcal{T}_h(0)$ be a regular triangulation of $\Omega(0)$. For the Galerkin finite-element formulation with time-dependent linear basis functions we write
\begin{align*}
\mathbf{w}_h(\mathbf{x},t) = \sum_{k=1}^{n_w} \mathbf{w}_k(t) \boldsymbol{\phi}_k(\mathbf{x},{\bf a}(t)),
\quad \boldsymbol{\varepsilon}_h(\mathbf{x},t) = \sum_{k=1}^{n_\varepsilon} \boldsymbol{\varepsilon}_k(t) \boldsymbol{\zeta}_k(\mathbf{x},{\bf a}(t)),
\quad
\boldsymbol{p}_h(\mathbf{x},t) = \sum_{k=1}^{n_p} p_k(t) \psi_k(\mathbf{x},{\bf a}(t)),
\end{align*}
where ${\bf a}(t)$ denotes the element coordinates (determined by displaced vertices) in time, $n_w$, $n_\varepsilon$ and $n_p$ denote, respectively, the number of degrees of freedom for the discrete velocity, strain and pressure fields.
Substitution of the Galerkin approximations in the weak form results in the following semi-discrete algebraic system:
\begin{subequations} \label{mvpe_semidisc}
\begin{align}
& \rho \frac{d}{dt} (M_w {\underline w}) + S_{vis} {\underline w} + S_{el} {\underline \varepsilon} - D^T {\underline p} = {\underline b}_w, \\[1ex]
%& \frac{d}{dt} (M_{\varepsilon} {\underline \varepsilon}) + N({\underline w},{\underline \varepsilon}) = - \underline{b}_G, \\[1ex]
&
\frac{d}{dt} (M_{\varepsilon} {\underline \varepsilon}) - B \underline{w} + N({\underline w},{\underline \varepsilon}) + \alpha M_{\varepsilon} \underline{\varepsilon} = \underline{0}, \\[1ex]
& D {\underline w} + \kappa L {\underline p} = %{\underline 0}.
\underline{b}_p.
\end{align}
\end{subequations}
Here, vectors $\underline{b}_w$ and $\underline{b}_p$ account for the right hand side terms in Eqs. \eqref{wf_w} and \eqref{wf_p} respectively, $M_w$ and $M_\varepsilon$ the (block) mass matrices, $D$ denotes the divergence matrix, $S_{el}$ and $S_{vis}$ the elasticity and viscous stiffness matrices, $L$ the Laplace matrix and matrix $B$ and vector $N(\underline{w},\underline{\varepsilon})$ arise from the discretization of the central term in \eqref{wf_varepsilon}, see Remark \ref{rem:wf_eps}.
Let $\Delta t$ be the time step and denote by $\tau$ the current time level (i.e., $t^\tau = \tau \Delta t$). After application of the backward Euler method to the semi-discrete problem (\ref{mvpe_semidisc}), one obtains
\begin{subequations}\label{discrmorphporo1}
\begin{align}
& \Big( \! \rho M_w^{\tau} + \Delta t S_{vis}^{\tau} \! \Big) {\underline w}^{\tau} + \Delta t S_{el}^{\tau} {\underline \varepsilon}^{\tau} - \Delta t (D^{\tau})^T {\underline p}^{\tau} = \rho M_w^{\tau-1} {\underline w}^{\tau-1} + \Delta t {\underline b}_w^{\tau}, \\[1ex]
& (1+\Delta t) M_\varepsilon^{\tau} \underline{\varepsilon}^{\tau} - \Delta t B^{\tau} \underline{w}^{\tau} + \Delta t \ N({\underline w}^{\tau},{\underline \varepsilon}^{\tau}) = M_\varepsilon^{\tau-1} {\underline \varepsilon}^{\tau-1}, \\[1ex] %G^{\tau}, \\[1ex]
& D^{\tau} {\underline w}^{\tau} + \kappa L^{\tau} {\underline p}^{\tau} = \underline{b}^\tau_p,
\label{discrmorphporo1_p}
\end{align}
\end{subequations}
which contains time-dependent matrices and vectors. Note that Eq. (\ref{discrmorphporo1}b) yields a nonlinear equation for the strain $\underline{\varepsilon}$. We adopt a fixed point iteration, on each time level, to solve the resulting nonlinear problem. % $A(z)z = b(z)$ We use a fixed point iteration to solve the resulting system
As we shall see in the numerical results, non-physical oscillations may appear in the pressure field for certain cases. Next we analyze the conditions that lead to these oscillations in the one-dimensional case.
\subsection{Monotonicity requirements} \label{sec:monotonicity}
We consider the one-dimensional problem and, for notational convenience, we drop the $\tau$ superscripts on the matrices and vectors at time $t^\tau$. After algebraic manipulation of Eqs. (\ref{discrmorphporo1}a) and (\ref{discrmorphporo1}c) we obtain that the pressure $\underline{p}$ must satisfy the following equation
\begin{equation}
A \underline{p} = \Delta t D P^{-1} S_{el}\underline{\varepsilon} + \tilde{\underline{b}}_p,
\end{equation}
where
\[
A = \kappa L + \Delta t D P^{-1}D^T, \qquad P = \rho M + \Delta t S_{vis} = \rho M + \Delta t \mu_{vis} L,
\]
with $\mu_{vis} = \mu_1 + \mu_2$, and
\[
\tilde{b}_p = -\rho D P^{-1}M^{\tau-1} \underline{w}^{\tau-1} - \Delta t D P^{-1} \underline{b}_w.
\]
If $A$ is an $M$-matrix then the pressure field satisfies a discrete maximum principle, which prohibits the occurrence of non-physical oscillations. Here, $P^{-1}$ is approximated by using fundamental solutions \cite{Vermolen2022}, which results in the following approximation
\begin{subequations} \label{approx}
\begin{align}
& (P^{-1})_{ij} \approx (\tilde{P}^{-1})_{ij} = \frac{\nu}{2 \rho \cosh(\nu)} \bar{p}_{ij} \\
& \bar{p}_{ij} =
\begin{cases}
\sinh(\nu(1-x_i+x_j)) + \sinh(\nu(1-x_i-x_j)) , & \text{if $j \le i$,} \\[1ex]
\sinh(\nu(1+x_i-x_j)) + \sinh(\nu(1-x_i-x_j)) , & \text{if $j > i$,}
\end{cases}
\end{align}
\end{subequations}
where $x_i$ represents the nodal point $i$ and
\[
\nu := \sqrt{\dfrac{\rho}{\Delta t \mu_{vis}}}.
\]
In \cite{Vermolen2022} it has been proved that $| (P^{-1})_{ij} - (\tilde{P}^{-1})_{ij} | = \mathcal{O}(h^{3/2})$, where $h$ denotes the size of the finite-element mesh.
We note that the analysis can be done for a non-uniform, however, for the sake of illustration, we give the analysis for a uniform mesh.
After some tedious algebraic manipulations, see Section 4.4 in \cite{denBakker2020}, it follows that
\begin{equation*}
(D P^{-1} D^T {\underline p})_i = \frac{h}{4 \Delta t \mu_{vis} } (p_{i-1} + 2 p_i + p_{i+1}) + \mathcal{O}(h^{7/2}),
\end{equation*}
since the coefficients of $D$ are of order $h$. Neglecting higher order terms in the previous construction we obtain
\begin{align*}
(A {\underline p})_i & = \left( (\kappa L + \Delta t DP^{-1}D^T) p \right)_i \\
& \approx \left( \frac{h}{4\mu_{vis}} - \frac{\kappa}{h} \right) p_{i-1} + 2\left( \frac{h}{4\mu_{vis}} + \frac{\kappa}{h} \right) p_i + \left( \frac{h}{4\mu_{vis}} - \frac{\kappa}{h} \right) p_{i+1}.
\end{align*}
Recall that for a matrix to be an $M$--matrix (for monotonicity of the finite-element solution), the off-diagonals need to be non--positive. %, hence we need
\begin{thm}\label{Theorem1}
{The matrix $A$ based on approximation (\ref{approx}) for $P^{-1}$ is an $M$-matrix if
\begin{equation} \label{eq:h_constrain}
h \leq 2\sqrt{\mu_{vis}\kappa}
\end{equation}
}
We remark that this condition is sufficient to remove spurious oscillations for the approximate system. If $h \rightarrow 0$, that is, $h$ is small, then the above result becomes sharper. The assertion displays an indication that (large) violation of this condition leads to spurious oscillations. We will see this in the numerical results section.
\settowidth{\leftmargin}{{\Large$\square$}}\advance\leftmargin\labelsep
\itemsep8pt\relax
\renewcommand\labelitemi{{\lower1.5pt\hbox{\Large$\square$}}}
\end{thm}
Note that this constraint affects only the mesh size and not the time step. As an intuitive interpretation, one could argue that the distance that particles in the porous medium travel over one time-step should be bounded by a number that is proportional to the mesh size. This is motivated by the following argument: Let the distance that a particle has traveled over one time-step $\Delta t$ be given by $v \Delta t$ be smaller than the mesh size, then we have $v \Delta t \le h$. Darcy's Law stipulates that $v = \kappa \frac{\partial p}{\partial x} \approx \kappa \frac{\Delta p}{h} = \frac{\tilde{\kappa}}{\mu_{vis}} \frac{\Delta p}{h}$, where $\tilde{\kappa}$ denotes the real permeability with unit $m^2$. This implies that
$\frac{\tilde{\kappa}}{\mu_{vis}} \frac{\Delta p}{h} \Delta t \le h$. This gives
$h^2 \ge \tilde{\kappa} \frac{\Delta p \Delta t}{\mu_{vis}}$. Since $\frac{\Delta p \Delta t}{\mu_{vis}}$ is a dimensionless number, we can indeed argue that the condition in the above theorem represents the necessity that over one time-step particles are not allowed to travel over more than a distance that is proportional to the mesh size.
If the viscosity and/or permeability are low, then the mesh resolution has to be very large. For these cases, stabilization may be attractive. On the other hand, note as well that for $\kappa \longrightarrow \infty$ the above criterion will always be satisfied.
\begin{rem}
{As $ \kappa \rightarrow \infty$, then the morphoelastic model is recovered, which always gives oscillation-free finite element approximations.}
\end{rem}
%\\[2ex]
\subsection{Stabilized finite-element method}
We follow the stabilization technique used by Aguilar et al. \cite{Aguilar2008} and Rodrigo \cite{RodrigoETAL2016}, which is based on the perturbation of the pressure equation. Note that the result that we are deriving in this section holds for the one-dimensional case. In our case, we introduce this perturbation for the discrete system of equations. Thus, we replace Eq. (\ref{discrmorphporo1}c) by
\begin{equation} \label{discrmorphporo}
D^{\tau} {\underline w}^{\tau} + (\kappa + \beta) L^{\tau} {\underline p}^{\tau} = \beta L^{\tau-1} \underline{p}^{\tau-1} + \underline{b}^\tau_p,
\end{equation}
where $\beta>0$ is the stabilization parameter. For the sake of presentation, we drop the superscript $\tau$. We look for an appropriate choice of $\beta$ such that
\[
B = (\kappa + \beta)L + \Delta t D P^{-1}D^T
\]
is an $M$--matrix. Using the approximation to $P^{-1}$ introduced in Section \ref{sec:monotonicity} we have, for the one-dimensional problem, that
\begin{align*}
(B {\underline p})_i & = \left( \Big((\kappa+\beta) L + \Delta t DP^{-1}D^T\Big) p \right)_i \\
& \approx \left( \frac{h}{4\mu_{vis}} - \frac{\kappa+\beta}{h} \right) p_{i-1} + 2\left( \frac{h}{4\mu_{vis}} + \frac{\kappa+\beta}{h} \right) p_i + \left( \frac{h}{4\mu_{vis}} - \frac{\kappa+\beta}{h} \right) p_{i+1}.
\end{align*}
The approximate matrix is an $M$--matrix if and only if
\[
\frac{h}{4\mu_{vis}}-\frac{\kappa + \beta}{h} \leq 0
\qquad \iff \qquad
\frac{h^2}{4\mu_{vis}} - \kappa \leq \beta.
\]
\begin{thm}\label{Theorem4}
{The matrix $B$ based on approximation (\ref{approx}) for $P^{-1}$ is an $M$-matrix if
\begin{equation}
\beta \ge \beta^* = \frac{h^2 }{4 \mu_{vis} }-\kappa,
\end{equation}
A sufficient condition for this is
$\beta \ge \dfrac{ h^2 }{4 \mu_{vis} }$.
}
\settowidth{\leftmargin}{{\Large$\square$}}\advance\leftmargin\labelsep
\itemsep8pt\relax
\renewcommand\labelitemi{{\lower1.5pt\hbox{\Large$\square$}}}
\end{thm}
The above condition is based on small $h$, and in the limit, warrants monotonicity of the finite element solution regardless of the permeability parameter $\kappa$ and the time step $\Delta t$.
We expect a similar condition for higher dimensionality in the future. However, in the next section we will consider a two dimensional test problem.
\section{Numerical results}
In this section, we test the numerical solution on the morpho-viscoporoelastic model \eqref{morphoporoelas} in the unit square, $\Omega(0) = [0,1]^2$, subject to homogeneous boundary conditions
\begin{subequations}
\begin{empheq}[left=\empheqlbrace]{align}
& {\bf w} = \mathbf{0}, \quad \kappa \nabla p \cdot \boldsymbol{n} = \mathbf{0} , & \text{on $\Gamma_1(t) \times (0,T]$,} \\[1ex]
& \boldsymbol{\sigma}\cdot\boldsymbol{n} = \mathbf{0}, \quad p = 0, & \text{on $\Gamma_2(t) \times (0,T]$,}
\end{empheq}
\end{subequations}
where $\Gamma_1 = \{{(x, y) : x = 0, y \in [0, 1]\}}$ and $\Gamma_2 = \partial \Omega \setminus \Gamma_1$, and initial conditions given in \eqref{mpve_ic1}-\eqref{mpve_icw}. We consider the problem with constant mechanical parameters
\[
2\mu = \lambda = 1, \quad \mu_v = \lambda_v = 1,
\]
density of the material $\rho = 1$ and the morphoelastic coefficient $\alpha=1$. In addition, the material is subjected to a time-dependent body force ${\bf f}_{\bf u}$, see Eq. \eqref{mvpe_momentum}, given by
\begin{equation}
{\bf f}_{\bf u}(t,{\bf x}) =
\begin{pmatrix}
0 \\
e^{-t} \sin(2\pi t)
\end{pmatrix}.
\end{equation}
\subsection{Pressure profiles and stabilization}
Figure \ref{fig:pressurefield_non_stabilized} shows the pressure field after one time step, $\Delta t = 0.1$, for different pairs of permeability $\kappa$ and mesh size $h$ (the diameter of the triangular elements). We can appreciate that the pressure field for $\kappa=10^{-2}$ and $h=0.14$ remains smooth (Fig. \ref{fig:non_stabilized_1}). This is to be expected as the chosen permeability and mesh size fulfill Eq. \eqref{eq:h_constrain}, $h \leq 2\sqrt{(\mu_1+\mu_2)\kappa}$. However, the pair $\kappa=10^{-6}$ and $h=0.07$ does not fulfill Eq. \eqref{eq:h_constrain} and spurious oscillations are visible in the pressure field (Fig. \ref{fig:non_stabilized_2}).
In order to obtain a monotonic pressure field, we could increase the grid resolution to meet Eq. \eqref{eq:h_constrain} (which yields $h\leq 0.0028$). We could also apply the stabilization technique proposed in the previous section. Theorem 5.2 establishes that the approximated system will stabilize for $\beta \geq 6.25\times 10^{-4}$. Figure \ref{fig:pressurefield_stabilized} shows the pressure profiles for different values of the stabilization parameter $\beta$. We observe the correspondence of the numerical results with the theoretical estimate.
\begin{figure}
\centering
% First row
\begin{subfigure}[b]{0.495\textwidth}
\centering
\includegraphics[width=\textwidth]{Non_stabilised_case_1.png}
\caption{$\kappa=10^{-2}$ and $h=0.14$}
\label{fig:non_stabilized_1}
\end{subfigure}
\hfill
\begin{subfigure}[b]{0.495\textwidth}
\centering
\includegraphics[width=\textwidth]{Case3_beta=0.png}
\caption{$\kappa=10^{-6}$ and $h=0.07$}
\label{fig:non_stabilized_2}
\end{subfigure}
\caption{Pressure fields for different pairs of $\kappa$ and $h$.}
\label{fig:pressurefield_non_stabilized}
\end{figure}
\begin{figure}
\centering
% First row
\begin{subfigure}[b]{0.32\textwidth}
\centering
\includegraphics[width=\textwidth]{Case3_beta=1e-5.png}
\caption{$\beta = 10^{-5}$}
\label{fig:stabilized_1}
\end{subfigure}
\hfill
\begin{subfigure}[b]{0.32\textwidth}
\centering
\includegraphics[width=\textwidth]{Case3_beta=6.25e-4.png}
\caption{$\beta = 6.25 \times 10^{-4}$}
\label{fig:stabilized_2}
\end{subfigure}
\hfill
\begin{subfigure}[b]{0.32\textwidth}
\centering
\includegraphics[width=\textwidth]{Case3_beta=1e-3.png}
\caption{$\beta = 10^{-3}$}
\label{fig:stabilized_3}
\end{subfigure}
\caption{Pressure fields for $\kappa = 10^{-6}$ and $h=0.07$ and different values of the stabilization parameter $\beta$.}
\label{fig:pressurefield_stabilized}
\end{figure}
\subsection{Total Variation (TV) for pressure field}
In order to have an objective measure for the amount of variation of the numerical solution, we consider the total variation over the domain of computation. In our case $\Omega$ is a rectangular domain. Let the numerical solution of the pressure be given by $p_h$ and suppose that the horizontal and vertical sides of the rectangular triangular elements have lengths $\Delta x$ and $\Delta y$, respectively. The discrete function $p_h$ is defined on this grid with $p_{i,j}$ being the value associated with a gridpoint ${{\bf x}_{i,j}}$. We use the definition for the Total Variation (TV) of Goodman and LeVeque \cite{Goodman1985}, which is given by:
\begin{equation}
\text{TV}_{\tau}(p_h) = \sum_{(i,j)\in {\tau}} \left( \Delta y |p_{i+1,j} - p_{i,j}| + \Delta x |p_{i,j+1} - p_{i,j}| \right).
\label{TV_def}
\end{equation}
This definition is commonly referred to as anisotropic TV in the literature, where $\tau$ is the set of all mesh points.
This formula approximates the given continuous total variation functional for pressure field given by
\begin{equation}
TV(p) = \int_\Omega |p_x| + |p_y| \, dxdy.
\end{equation}
We subsequently experimentally demonstrate that our stabilization technique is Total Variation Diminishing (TVD) with increasing $\beta$-values i.e. $$0 \le TV_{\tau}(p_{\beta_{k+1}}) \leq TV_{\tau}(p_{\beta_{k}}), ~ \forall ~ \beta_{k+1} \ge \beta_k.$$ As mentioned earlier, the current problem is solved on the unit square, $\Omega(0) = [0,1]^2$.
We list the total variation of the numerical solutions for the pressure field using equation (\ref{TV_def}) in Table \ref{Table_TV_vs_beta}. We observe that the total variation of the solutions over the whole domain monotonically decreases with increasing values of the stabilization parameter, which is to be expected (see also Figure \ref{TV_beta}). Note that the total variation converges to a non-zero limit since the actual numerical solution is not constant over the domain of computation.
\begin{table}
\centering
\caption{\it Total variation values for the pressure fields for different values of stabilization parameter $\beta$.}
\label{Table_TV_vs_beta}
\begin{tabular}{|c|c|c|}
\hline
stabilization parameter ($\beta$) & Total Variation ($TV(p_h)$) \\
\hline \hline
$0$ & 23.6096 \\
\hline
$10^{-5}$ & 12.1596 \\
\hline
$10^{-4}$ & 8.3987 \\
\hline
$3.12\times10^{-4}$ & 7.5860 \\
\hline
$6.25\times10^{-4}$ & 7.2501 \\
\hline
$10^{-3}$ & 6.9645 \\
\hline
\hline
\end{tabular}
\end{table}
\begin{figure}[t]
\centering
\includegraphics[width=10cm]{TV_plot.png}
\caption{\textit{TV for pressure fields for $\kappa = 10^{-6}$ and $h=0.07$ versus different values of the stabilization parameter $\beta$ for one time step $\Delta t=0.1$}.}
\label{TV_beta}
\end{figure}
\section{Conclusions}
We have studied a morpho-viscoporoelastic model that combines the porous, elastic and viscous structure of tissues with microstructural changes that may lead to growth or shrinkage of the tissue. The model has been analyzed in terms of the symmetry of the strain tensor and linear stability. Linear stability was analyzed for both the continuous and semi-discrete systems. Stability conditions have been derived. Besides stability, monotonicity of the numerical solution for the pressure can be warranted as long as the mesh size does not exceed a critical number that depends on the permeability of the porous medium. Furthermore, for larger mesh sizes, stabilization is required. This stabilization is based on the addition of a Laplacian operator on the pressure equation, and it has been found that the stabilization parameter should not be smaller than a threshold that incorporates the mesh size, viscosity, and permeability. The numerical simulations that we show confirm the stability bounds and bounds for the stabilization that we derived theoretically.
Our study was carried out for a one-dimensional case and we want to extend this to higher dimensionality. For this purpose, we need expressions for the inverse of Laplace matrices and matrices that depend affinely on Laplace matrices for higher dimensions. We recently found some closed-form expressions for multi-dimensional matrix polynomials of Laplace matrices (with various boundary conditions) \cite{Asghar2025}, and we will study whether we can apply our recent findings in the current context.
\section*{Acknowledgements}
This work was supported by Research England under the Expanding Excellence in England (E3) funding stream, which was awarded to MARS: Mathematics for AI in Real-world Systems in the School of Mathematical Sciences at Lancaster University. The work of E. Javierre is supported in part by the Spanish project PID2022-140108NB-I00 (MCIU/AEI/FEDER, UE), and by the DGA (Grupo de referencia APEDIF, ref. E24$\textunderscore$17R). Further, this work was supported by a fellowship awarded to S. Asghar by the Higher Education Commission (HEC) of Pakistan in the framework of project: 1(2)/HRD/OSS-III/BATCH-3/2022/HEC/527.
\begin{thebibliography}{00}
%% For authoryear reference style
%% \bibitem[Author(year)]{label}
%% Text of bibliographic item
\bibitem[Taylor(2020)]{Taylor2020}
M.R. Taylor, W. Zaw, W.A. Patrick, Large-deformation strain energy density function for vascular smooth muscle cells, J. Biomech. 3 (2020) Paper No. 110005. https://doi.org/10.1016/j.jbiomech.2020.110005
\bibitem[Simon(1990)]{Simon1990}
B.R. Simon, Poroelastic Finite Element Models for Soft Tissue Structures,
In: Hukins, D.W.L. (eds) Connective Tissue Matrix, Topics in Molecular and Structural Biology, Palgrave, London 1990. https://doi.org/10.1007/978-1-349-09865-1
\bibitem[Biot(1941)]{Biot1941}
M.A. Biot, General theory of three dimensional consolidation, J. Appl. Phys. 12 (1941) 155-–164. https://dx.doi.org/10.1063/1.1712886
\bibitem[Humphrey(2002)]{Humphrey2002}
J.D. Humphrey, K.R. Rajagopal, A constrained mixture model for growth and remodeling of soft tissues, Math. Models Methods Appl. Sci. 12 (3) (2002) 407--430. https://doi.org/10.1142/S0218202502001714
\bibitem[Mow(1980)]{Mow1980}
V.C. Mow, S.C. Kuei, W.M. Lai, C.G. Armstrong, Biphasic creep and stress relaxation of articular cartilage in compression: theory and experiments, J. Biomech. Eng. 102 (1) (1980) 73--84. https://doi.org/10.1115/1.3138202
\bibitem[BaxterJain(1989)]{BaxterJain1989}
L.T. Baxter, R.K. Jain, Transport of fluid and macromolecules in tumors. I. Role of interstitial pressure and convection, Microvasc. Res. 37 (1) (1989) 77--104. https://doi.org/10.1016/0026-2862(89)90074-5
\bibitem[NettiETAL(1995)]{NettiETAL1995}
P.A. Netti, L.T. Baxter, Y. Boucher, R. Skalak, R.K. Jain, Time-dependent behavior of interstitial fluid pressure in solid tumors: implications for drug delivery, Cancer Res. 55 (22) (1995) 5451--5458.
\bibitem[JainETAL(2014)]{JainETAL2014}
R.K. Jain, J.D. Martin, T. Stylianopoulos, The role of mechanical forces in tumor growth and therapy, Annu. Rev. Biomed. Eng. 16 (2014) 321--346. https://doi.org/10.1146/annurev-bioeng-071813-105259
\bibitem[HeldinETAL(2004)]{HeldinETAL2004}
C.H. Heldin, K. Rubin, K. Pietras, A. Ostman, High interstitial fluid pressure - an obstacle in cancer therapy, Nat. Rev. Cancer. 4 (10) (2004) 806--813. https://doi.org/10.1038/nrc1456
\bibitem[Mak(1986)]{Mak1986}
A.F. Mak, The apparent viscoelastic behavior of articular cartilage---the contributions from the intrinsic matrix viscoelasticity and interstitial fluid flows, J. Biomech. Eng. 108 (1986) 123--130. https://doi.org/10.1115/1.3138591
\bibitem[BergerETAL(2015)]{BergerETAL2015}
L. Berger, R. Bordas, D. Kay, S. Tavener, Stabilized lowest-order finite element approximation for linear three-field poroelasticity, SIAM Journal on Scientific Computing, 37 (5) (2015) A2222--A2245. https://doi.org/10.1137/15M1009822
\bibitem[MuradLoula(1994)]{MuradLoula1994}
M.A. Murad, A.F.D. Loula, On stability and convergence of finite element approximations of Biot's consolidation problem, Int. J. Numer. Methods Eng. 37 (4) (1994) 645--667. https://doi.org/10.1002/nme.1620370407
\bibitem[RodrigoETAL(2016)]{RodrigoETAL2016}
C. Rodrigo, F.J. Gaspar, X. Hu, L.T. Zikatanov, Stability and monotonicity for some discretizations of the Biot's consolidation model, Comput. Methods Appl. Mech. Eng. 298 (2016) 183--204. https://doi.org/10.1016/j.cma.2015.09.019
\bibitem[RodrigoETAL(2018)]{RodrigoETAL2018}
C. Rodrigo, X. Hu, P. Ohm, J.H. Adler, F.J. Gaspar, L.T. Zikatanov, New stabilized discretizations for poroelasticity and the {S}tokes' equations, Comput. Methods Appl. Mech. Eng. 341 (2018) 467--484. https://doi.org/10.1016/j.cma.2018.07.003
\bibitem[EgbertsStab(2021)]{EgbertsStab2021}
G. Egberts, F.J. Vermolen, P.P.M. van Zuijlen, Stability of a one-dimensional morphoelastic model for post-burn contraction, J. Math. Bio. 83 (24) (2021). https://doi.org/10.1007/s00285-021-01648-5
\bibitem[Hall(2017)]{Hall2017}
S.N. Menon, C.L. Hall, S.W. McCue, D.L.S. McElwain, A model for one-dimensional morphoelasticity and its application to fibroblast-populated collagen lattices, Biomech. Model. Mechanobiol. 16 (2017) 1743-–1763. https://doi.org/10.1007/s10237-017-0917-3
\bibitem[Hall(2008)]{Hall2008}
C.L. Hall, Modelling of some biological materials using continuum mechanics,
PhD Thesis at the Queensland University of Technology, 2008.
\bibitem[Rodriguez(1994)]{Rodriguez1994}
E.K. Rodriguez, A. Hoger, A. McCulloch, Stress-dependent finite growth in soft elastic tissues, J. Biomech. 27 (1994) 455--467. https://doi.org/10.1016/0021-9290(94)90021-3
\bibitem[Goriely(2015)]{Goriely2015}
A. Erlich, T. Lessinnes, D. Moulton, A. Goriely, A short introduction to morphoelasticity: the dynamics of growing tissues. In: Bigoni (Ed) Extremely deformable structures, CISM International Centre for Mechanical Sciences, 562, Springer, Vienna 2015.
\bibitem[Fletcher(1998)]{Fletcher1998}
C.A.J. Fletcher, Computational techniques for fluid dynamics 1. Springer, Berlin 1998.
\bibitem[Asghar1(2025)]{Asghar12025}
S. Asghar, F.J. Vermolen. Stability of Equilibria in a One Dimensional Model for Morpho–Poroelasticity for Soft Tissues. In: Numerical Mathematics and Advanced Applications ENUMATH 2023, Lecture Notes in Computational Science and Engineering, 154 (2) Springer, Cham 2025. https://doi.org/10.1007/978-3-031-86169-7
\bibitem[Reynolds(1903)]{Reynolds1903}
O. Reynolds, Papers on Mechanical and Physical Subjects. The Sub-Mechanics of the Universe. Cambridge: Cambridge University Press, 3 (1903) 12–-13.
\bibitem[Aguilar(2008)]{Aguilar2008}
G. Aguilar, F.J. Gaspar, F.J. Lisbona, C. Rodrigo, Numerical stabilization of Biot's consolidation model by a perturbation on the flow equation, Int. J. Numer. Methods Eng. 75 (11) (2008) 1282--1300. https://doi.org/10.1002/nme.2295
\bibitem[Vermolen(2022)]{Vermolen2022}
F.J. Vermolen, D. den Bakker, C. Vuik, On the fundamental solutions-based inversion of Laplace matrices, Results Appl. Math. 15 (2022) Paper No. 100288. https://doi.org/10.1016/j.rinam.2022.100288
\bibitem[denBakker(2020)]{denBakker2020}
D. den Bakker, Analysis of stabilized finite element methods for a morpho-poroelastic model applied to tumor growth. Master Thesis at the Delft University of Technology, Delft Institute of Applied Mathematics, 2020.
\bibitem[Goodman(1985)]{Goodman1985}
J.B. Goodman, R.J. LeVeque, On the accuracy of stable schemes for 2d scalar conservation laws, Math. Comput. 45 (171) (1985) 15--21. https://doi.org/10.2307/2008046
\bibitem[Asghar(2025)]{Asghar2025}
S. Asghar, Q. Peng, F.J. Veremolen, C. Vuik, On the Inversion of Polynomials of Discrete Laplace Matrices, (2025).
https://doi.org/10.48550/arXiv.2511.19724.
\end{thebibliography}
\end{document}
\endinput
%%
%% End of file `elsarticle-template-num-names.tex'.