An ultraweak DPG method for viscoelastic fluids

Brendan Keith,Philipp Knechtges,Nathan V. Roberts,Stefanie Elgeti,Marek Behr,Leszek Demkowicz

DOI: https://doi.org/10.1016/j.jnnfm.2017.06.006

2017-06-27

Abstract:We explore a vexing benchmark problem for viscoelastic fluid flows with the discontinuous Petrov-Galerkin (DPG) finite element method of Demkowicz and Gopalakrishnan [1,2]. In our analysis, we develop an intrinsic a posteriori error indicator which we use for adaptive mesh generation. The DPG method is useful for the problem we consider because the method is inherently stable---requiring no stabilization of the linearized discretization in order to handle the advective terms in the model. Because stabilization is a pressing issue in these models, this happens to become a very useful property of the method which simplifies our analysis. This built-in stability at all length scales and the a posteriori error indicator additionally allows for the generation of parameter-specific meshes starting from a common coarse initial mesh. A DPG discretization always produces a symmetric positive definite stiffness matrix. This feature allows us to use the most efficient direct solvers for all of our computations. We use the Camellia finite element software package [3,4] for all of our analysis.

Numerical Analysis,Fluid Dynamics

What problem does this paper attempt to address?

The paper attempts to solve several key problems in the simulation of viscoelastic fluid flow, specifically including:

1. **Numerical stability**: Due to the convective nature of the non - linear viscoelastic constitutive law in the viscoelastic fluid model, stabilization is a ubiquitous problem. Many literatures have reported differences in numerical results and final numerical failures, which may be because the stabilization introduces spurious effects. Therefore, it is crucial to find a method that does not require additional stabilization terms.

2. **Resolution of small - scale features**: In certain parameter ranges, the solutions of viscoelastic fluid flow exhibit small - scale, high - contrast features, such as boundary layers and internal layers. These features cause the values of stress components to change rapidly by several orders of magnitude in local regions. Accurately resolving these small - scale features is crucial for understanding fluid behavior.

3. **Verification of benchmark problems**: Due to the complexity of viscoelastic fluid equations, there are few known analytical solutions, so the verification of numerical methods depends on benchmark problems. Even for seemingly simple problems, differences in numerical results and loss of convergence are common and controversial.

To address these problems, the author adopts the discontinuous Petrov - Galerkin (DPG) finite - element method. This method has the following advantages:

- **Intrinsic stability**: The DPG method is intrinsically stable and can handle the convective terms in the model without stabilizing the discretized linearized problem.
- **A posteriori error estimator**: An intrinsic a posteriori error indicator has been developed for adaptive mesh generation.
- **Symmetric positive - definite stiffness matrix**: DPG discretization always produces a symmetric positive - definite stiffness matrix, which allows the most efficient direct solver to be used directly for calculation.
- **Generation of parameter - specific meshes from coarse meshes**: Intrinsic stability can be achieved on all length scales, and parameter - specific meshes can be generated from a common initial coarse mesh.

Through these characteristics, the DPG method can reduce numerical errors and adaptively resolve small - scale features through a series of refined meshes. In addition, this method also simplifies the analysis process, making it a powerful tool for solving viscoelastic fluid flow problems.