The present paper is focused on the solution of differential-algebraic equation systems (DAE), which arise in large viscoelastic deformations within the quasi-static finite element context. For this purpose linearly implicit methods of Rosenbrock-type are used, which avoid completely the solution of non-linear equations. This article investigates a possible treatment of the new global approach with respect to expense and achievable accuracy.
Constitutive ModelIn this investigation an overstress-type model of finite strain viscoelasticity with an extention to a poly-convex hyperelasticity relation is applied, see [3] and [4]. The model is based on the multiplicative decomposition of the deformation gradient F into an elastic and a viscous part, F = F e F v . The total deformation, represented by the right Cauchy-Green tensor C = F T F, defines the equilibrium stress partT eq = 2dψ(J,C) dC , which is carried out by an additive decomposition of the strain-energy function ψ(J, C) = U (J) + w eq (C) into a volume-changing and a volume-preserving part, where the volume-changing part depends on J = det F and the volume-preserving part on the invariants of the unimodular right Cauchy-Green tensor C = (det C) −1/3 C. The over-stresses are given by a Neo-Hooke type model w ov (C e ) = µ(I Ce − 3) which are formulated by quantities relative to the inelastic intermediate configuration:C e = (detC e ) −1/3 C e defines the unimodular elastic right Cauchy-Green tensor withstands for the 2nd Piola-Kirchhoff tensor, where T is the Cauchy-stress tensor. In the case of viscous (inelastic) deformations the internal variables develop according to the associated flow ruleĊ v = 4µ η, described by a process-dependent viscosity η = η 0 exp(−s(CT ov ·T ov C) 1/2 ) and the viscous right Cauchy-Green tensor
Application of Rosenbrock-type methodsThe numerical treatment of quasi-static problems with constitutive equations of evolutionary type, is currently explained as a solution procedure of DAEs related to the consistent application of the vertical method of lines , see [1] and the literature cited therein. In view of this global DAE-approach the calculation of the reaction forces are involved using the method of the Lagrange multipliers, see [5]. This extends the DAE-system with a constraint equation to the form
F(t, y(t),ẏ(t))where λ ∈ R n p are the Lagrange multipliers interpreted as the reaction forces. In contrast to the classical "displacementdriven" formulation the discretized weak form g a = {g,ĝ} T depends on all nodal displacements u a ∈ R n dof . Here the decomposition u a = {u,û} T , into displacements u ∈ R n u at those nodes where no constraints are given and the displacementŝ u ∈ R n p where the prescribed displacements u ∈ R n p apply, is used. q ∈ R n Q is the vector of all internal variables occurring at all Gauss-points. In [5] diagonally implicit Runge-Kutta methods (DIRK), which embraces the classical Backward-Euler method as a particular case, are applied to (1). The integration procedure yield a coupled system of no...