ICANEWS

Natural Sciences

MM-SUPG Method Developed for Time-dependent Convection-Dominated Convection-Diffusion Problems

A new research effort details the development of a Moving Mesh with Streamline Upwind Petrov-Galerkin (MM-SUPG) method. This method is specifically designed to

Moving Mesh with Streamline Upwind Petrov-Galerkin (MM-SUPG) Method Addresses Convection-Dominated Problems

A new research effort details the development of a Moving Mesh with Streamline Upwind Petrov-Galerkin (MM-SUPG) method. This method is specifically designed to tackle the complexities inherent in time-dependent convection-dominated convection-diffusion problems. The approach integrates residual-based SUPG stabilization with a metric-based moving mesh Partial Differential Equation (MMPDE) technique, presenting a novel strategy for computational modeling in such challenging scenarios.

Addressing Time-Dependent Convection-Dominated Convection-Diffusion Problems

The core focus of this research lies in phenomena characterized by time-dependent convection-dominated convection-diffusion problems. These are a class of mathematical problems that frequently arise in various scientific and engineering disciplines. Convection-dominated regimes, in particular, often pose significant numerical challenges due to the presence of sharp layers and the potential for spurious oscillations in numerical solutions.

The Interplay of Mesh Adaptation and Stabilization

A key characteristic highlighted in the developed MM-SUPG method is the intricate interaction between mesh adaptation and stabilization. This interaction is central to the method's functionality and performance. The evolving mesh plays a direct role in modifying both the direction and the magnitude of the SUPG stabilization. This modification occurs through its influence on the local element geometry. Simultaneously, the stabilization component of the method serves to improve robustness, particularly in the aforementioned convection-dominated regimes where numerical stability can be difficult to maintain.

This symbiotic relationship suggests that as the computational mesh dynamically adjusts to the solution's features, it actively informs the stabilization process. Conversely, the stabilization mechanism contributes to a more stable and accurate representation of the underlying physical processes, especially when strong convective forces are present.

Development of the MM-SUPG Method

The MM-SUPG method is constructed by a combination of two distinct, yet complementary, methodologies. The first is residual-based SUPG stabilization. The second is a metric-based moving mesh PDE (MMPDE) approach. The integration of these two techniques forms the foundation of the proposed method.

The residual-based SUPG stabilization is a well-established technique used to enhance the stability of finite element methods in convection-dominated problems. It introduces artificial diffusion aligned with the flow direction, thereby reducing oscillations without excessively smearing sharp features. The 'residual-based' aspect implies that the stabilization term is derived from the residual of the governing equation, indicating a direct link to the error in the solution.

The metric-based moving mesh PDE (MMPDE) approach, on the other hand, is concerned with dynamically adapting the computational mesh. This adaptation is typically driven by a mesh metric tensor, which dictates how the mesh elements are distributed in space to resolve solution features efficiently. By combining these two powerful concepts, the MM-SUPG method aims to leverage the strengths of each to address the complexities of the target problems.

Performance with Isotropic Diffusion

When applied to cases involving isotropic diffusion, numerical results provide insights into the effectiveness of the proposed MM-SUPG method. For these scenarios, the method demonstrates an ability to reduce spurious oscillations. Spurious oscillations are unphysical ripples or wiggles in numerical solutions that can arise from inadequate resolution or stability issues, particularly in the presence of sharp gradients.

Furthermore, the MM-SUPG method offers improved resolution of sharp layers. Sharp layers are regions where the solution changes rapidly over a small spatial extent, such as boundary layers or internal fronts. Accurately resolving these layers is crucial for capturing the correct physics of many problems.

"Numerical results show that the proposed method reduces spurious oscillations and provides improved resolution of sharp layers compared with fixed mesh methods, while yielding accuracy comparable to moving mesh finite element methods without SUPG in some cases."

This finding indicates a significant advantage over traditional fixed mesh methods, which often struggle with these aspects. Interestingly, for isotropic diffusion, the accuracy achieved by the MM-SUPG method is, in some cases, comparable to that of moving mesh finite element methods that do not incorporate SUPG stabilization. This suggests that the combined approach maintains the benefits of mesh adaptation while adding the stability advantages of SUPG.

Handling Anisotropic Diffusion

The research also extends to problems characterized by anisotropic diffusion. Anisotropic diffusion implies that the rate of diffusion varies depending on the direction. This makes the problem more complex than isotropic diffusion, where diffusion is uniform in all directions.

Constructing the Mesh Adaptation Metric for Anisotropic Cases

For anisotropic diffusion problems, the researchers introduced a weighted tensor. This tensor is designed to incorporate both the diffusion tensor and the convection field. The diffusion tensor describes how diffusion occurs in different directions, while the convection field describes the transport of properties by fluid motion. By combining these two crucial components into a single weighted tensor, the method prepares for more sophisticated mesh adaptation.

Following the introduction of this weighted tensor, a metric tensor is constructed via intersection. This constructed metric tensor serves a specific purpose: to guide mesh adaptation. In the context of moving mesh methods, a metric tensor defines the desired size, shape, and orientation of mesh elements. By creating such a tensor from the weighted combination of diffusion and convection information, the mesh can be adapted more effectively to the anisotropic nature of the problem, ensuring that mesh elements are appropriately aligned and sized to capture solution features even when flow and diffusion are not aligned.

Ensuring Monotonicity: The Discrete Maximum Principle (DMP)

A significant aspect of the research involves the analysis of the discrete maximum principle (DMP). The DMP is a desirable property for numerical schemes, particularly for diffusion-dominated or convection-diffusion problems, as it ensures that the discrete solution does not exhibit unphysical overshoots or undershoots beyond the range of actual solution values. Establishing conditions for the DMP is critical for the robustness and physical fidelity of a numerical method.

Sufficient Conditions for DMP

Under suitable assumptions pertaining to the elementwise tensors and the mesh geometry, the research establishes sufficient conditions for the discrete maximum principle (DMP) of the fully discrete scheme. These conditions are derived based on a rigorous analysis that includes quantitative bounds for the convection terms. This analytical step is fundamental, as it provides a theoretical underpinning for the method's behavior under specific circumstances.

Crucially, the analysis necessitates structural conditions regarding the alignment between the velocity field and the diffusion tensor. This means that for the DMP to hold, there must be a certain relationship or coherence between how the fluid is moving (velocity field) and how properties are diffusing (diffusion tensor). Such conditions are often encountered in the theoretical analysis of numerical methods for anisotropic problems, highlighting the detailed mathematical scrutiny applied to the MM-SUPG method.

Impact on Monotonicity and Accuracy

Numerical experiments were conducted to evaluate the performance of the proposed metric, particularly in the context of anisotropic diffusion and the DMP. These experiments demonstrate concrete benefits of the developed approach.

"Numerical experiments demonstrate that the proposed metric improves monotonicity properties and reduces undershoots, while maintaining overall accuracy."

The findings indicate that the introduced metric effectively improves monotonicity properties. Improved monotonicity ensures that the numerical solution does not produce non-physical oscillations. Furthermore, the experiments show a reduction in undershoots. Undershoots are a type of spurious oscillation where the numerical solution dips below the minimum physical value, often seen in regions with sharp gradients. By reducing these undershoots, the method produces a more physically realistic and stable solution.

Importantly, these improvements in monotonicity and the reduction of undershoots are achieved while the method successfully maintains overall accuracy. This suggests that the enhancements in solution quality (e.g., less oscillations, better monotonicity) do not come at the cost of reduced accuracy, which is a desirable characteristic for any numerical scheme.

Summary of Key Features and Benefits

The MM-SUPG method distinguishes itself through several key features. Its ability to enable interaction between mesh adaptation and stabilization allows the evolving mesh to dynamically modify the SUPG stabilization, thereby leveraging local element geometry. Concurrently, stabilization enhances robustness in convection-dominated regimes.

For isotropic diffusion, the method's advantages include the reduction of spurious oscillations and an improved resolution of sharp layers when compared to fixed mesh methods. Its accuracy for isotropic diffusion is also comparable to moving mesh finite element methods without SUPG in some instances.

In the more complex scenario of anisotropic diffusion, the introduction of a weighted tensor that combines the diffusion tensor and the convection field, followed by the construction of a metric tensor via intersection, is pivotal. This metric tensor is instrumental in guiding mesh adaptation effectively.

The theoretical analysis provides sufficient conditions for the discrete maximum principle (DMP) of the fully discrete scheme, relying on quantitative bounds for convection terms and specific structural conditions on the alignment between the velocity field and the diffusion tensor. These analytical results underpin the method's stability and reliability.

The numerical experiments further reinforce these theoretical findings, demonstrating that the proposed metric for anisotropic cases significantly improves monotonicity properties and reduces undershoots, all while successfully preserving the overall accuracy of the solution. This combination of theoretical grounding and empirical validation highlights the robustness and efficacy of the MM-SUPG method for time-dependent convection-dominated convection-diffusion problems.

Key Findings