Computer Methods in Applied Mechanics and Engineering
Higher order stabilized finite element method for hyperelastic finite deformation
Introduction
Stabilized finite element methods (SFEMs) consist of adding mesh dependent terms to the usual Galerkin method (see, for example, the works of [1], [2]) in order to improve the stability of the solution in problems with known instabilities, for example in modeling incompressible material deformation. Without the added stabilization terms, Galerkin methods applied to incompressible or nearly incompressible material behavior in the setting of a mixed finite element method must fulfill the Ladyzenskaya–Babuska–Brezzi (LBB) condition to achieve unique solvability, convergence and robustness [3]. This places severe restrictions on the choice of the solution space. Without balancing the interpolation functions according to the LBB condition, large errors or oscillations may appear in the solution. Furthermore, the LBB condition complicates the use of p-adaptive techniques, where flexibility is needed in varying the interpolation order. The advantage of stabilized methods is that they do not impose constraints on the interpolation functions, thus allowing for greater flexibility in balancing efficiency and accuracy.
SFEMs, which avoid the limitations on the interpolation functions in the traditional Galerkin finite element method, are widely used now, especially in applications in fluid mechanics. The stabilization terms, which are added to the usual Galerkin formulation, are functions of the residuals of the Euler–Lagrange equations evaluated element-wise. From the construction, it follows that consistency is not affected since the exact solution satisfies both the Galerkin term and the additional terms (see, for example, the works of [1], [4], [5]). Recently, Klaas et al. [6] have applied a stabilized finite element method with linear elements to hyperelasticity, and the results show the effectiveness of the stabilized finite element method for large deformation problems. However, higher order stabilized finite elements were not implemented in that work because of the difficulty in computing the required higher order derivatives.
In this paper, a higher order stabilized finite element formulation in a Lagrangian reference frame for hyperelasticity is developed. A Petrov–Galerkin method, following that presented in [1], is used. This results in a term with the strong form of the equilibrium equation being added to the usual Galerkin method. This term involves the divergence of part of the stress tensor, which, in turn, depends on the deformation gradient field. In this work, a local reconstruction method, following the work presented for two-dimensional Navier–Stokes equations in [7], is used to compute the stress at the finite element nodal locations. Then the divergence of the reconstructed stress is computed from the nodal quantities and shape functions. Numerical examples with a non-linear hyperelastic constitutive law for rubber-like materials are given. Finally, conclusions are inferred and directions for future work are discussed.
Section snippets
Governing equations for Lagrangian description
Consider a three-dimensional reference domain B with boundary Γ. The boundary value problem for finite elasticity in the absence of body forces defined on the reference configuration is given as
Find a displacement field such thatwith boundary conditionswhere is the deformation gradient, is the second Piola–Kirchoff stress, is the outward unit normal and is the prescribed traction load on ΓN, and is the prescribed displacement on ΓD. Finally, for
Stabilized mixed displacement–pressure formulation
Following standard Galerkin procedure, the strong form in Eq. (1) is integrated with a kinematically admissible weighting function In this work, a simple Petrov–Galerkin formulation following that presented in [6] is used. Specifically, following the standard Galerkin method, the strong form of the equilibrium equation in Eq. (1) is integrated with weight function lying in the kinematically admissible space W=V×P:where is the standard part and the second
Linearization
Now Eq. (11), after the local reconstruction (Eq. (14)), can be rewritten in a short form asLinearization of Eq. (16), leads to the stabilized system equationswhere
Examples
The above stabilized formulation has been implemented into an object-oriented finite element framework, named Trellis, developed in the Scientific Computation Research Center at Rensselaer Polytechnic Institute, see [8]. In order to provide a quantitative assessment of the above stabilized, mixed finite element formulation and to demonstrate the behavior of the higher order stabilized mixed method, two examples are investigated: (1) Cook's example, and (2) plane strain extension of a
Conclusions
A higher order, stabilized, Lagrangian finite element formulation for nearly incompressible finite deformation hyperelasticity is presented. In the stabilized formulation, mesh dependent terms, which enhance stability, are added element-wise to the usual Galerkin method resulting in a Petrov–Galerkin formulation. A local reconstruction method is used to compute the higher order derivatives that arise in the stabilization terms, specifically derivatives of the stress tensor.
Numerical examples,
Acknowledgements
This work has been supported by the National Science Foundation through grant DMI-9634920 as well as by the US Department of Defense – Department of the Air Force, Wright Laboratory.
References (8)
Multiscale phenomena: Green's functions, the Dirichlet-to-Neumann formulation, subgrid scale models, bubbles and the origins of stabilized methods
Comput. Methods Appl. Mech. Engrg.
(1995)- et al.
A stabilized mixed finite element methods for finite elasticity
Comput. Methods Appl. Mech. Engrg.
(1999) - et al.
A better consistency for low-order stabilized finite element methods
Comput. Methods Appl. Mech. Engrg.
(1999) - et al.
A new finite element formulation for computational fluid dynamics: V. Circumventing the Babuska–Brezzi condition: A stable Petrov–Galerkin formulation of the Stokes problem accommodating equal-order interpolations
Comput. Methods Appl. Mech. Engrg.
(1985)
Cited by (63)
An accurate approach to simulate friction stir welding processes using adaptive formulation refinement
2023, Finite Elements in Analysis and DesignPyTorch-FEA: Autograd-enabled finite element analysis methods with applications for biomechanical analysis of human aorta
2023, Computer Methods and Programs in BiomedicineA kinematically stabilized linear tetrahedral finite element for compressible and nearly incompressible finite elasticity
2023, Computer Methods in Applied Mechanics and EngineeringAdaptive stabilized mixed formulation for phase field fracture modeling of nearly incompressible finite elasticity
2022, International Journal of Mechanical SciencesCitation Excerpt :From a numerical standpoint, the equal-order linear P1/P1 pair is highly attractive due to its simplicity and ease of adaptive meshing. To stabilize the P1/P1 formulation, numerous numerical methods have been developed, including polynomial pressure projections [56,57], variational multi-scale methods [58,59], mesh-dependent stabilization terms [60,61], etc. Recently, Gavagnin et al. [62] proposed a stabilized mixed formulation for the brittle fracture of porous media based on polynomial pressure projections.
Stress, strain and dissipation accurate 3-field formulation for inelastic isochoric deformation
2021, Finite Elements in Analysis and DesignA nonlinear plane strain finite element analysis for multilayer elastomeric bearings
2021, Applied Mathematics and ComputationCitation Excerpt :He gave the distribution of the stored energy function of the block as well as the stress concentrations at the corners of the border under compression. Maniatty et al. [16] investigated a centre-holed plate using plane strain model, with normal traction boundary conditions and gave stress distribution in the structure. Moon et al. [17] proposes a new type of bearing system for isolation.