TL;DR
This paper introduces a novel "weighted" discrete-velocity-direction model (DVDM) for simulating gas flows, incorporating internal molecular degrees of freedom. This weighted approach significantly improves the accuracy of macroscopic quantities, particularly heat flux, and the model is shown to be efficient and accurate across various flow regimes, from continuum to rarefied.
What Problem Does This Paper Address?
Simulating rarefied gas flows accurately is computationally challenging due to the high dimensionality of kinetic equations. While the Bhatnagar-Gross-Krook (BGK) model offers a simplification, its numerical solution remains expensive. Previous discrete-velocity-direction (DVDM) models, while reducing dimensionality, struggled with accurately recovering macroscopic quantities, especially heat flux, and often lacked a robust way to handle internal molecular degrees of freedom. This paper aims to address these limitations by introducing a weighted DVDM that improves accuracy and incorporates internal molecular dynamics.
Methods
The authors build upon their previous DVDM framework by introducing two key innovations:
- Internal Molecular Degrees of Freedom: The model is extended to include internal molecular degrees of freedom (rotation, vibration), allowing for more realistic fluid properties and a more accurate specific heat ratio.
- Weighted Macroscopic Recovery: A crucial novelty is the introduction of a "weight function" ($|\xi|^{D-1}$) when recovering macroscopic quantities (density, momentum, energy) from the discrete distributions. This weighting is inspired by coordinate transformations and is shown to be essential for accurate results, particularly for heat flux.
Three spatial-time submodels are then developed by treating the continuous velocity variable ($\xi$) using:
- Discrete-Velocity Method (DVM): Discretizes the velocity space into a set of nodes.
- Gaussian-Extended Quadrature Method of Moments (Gaussian-EQMOM): Approximates the distribution function as a sum of Gaussian functions.
- Hermite Spectral Method (HSM): Expands the distribution function in a series of Hermite polynomials.
These submodels are then implemented using appropriate numerical schemes (e.g., IMEX-RK, WENO, Godunov flux) and boundary conditions (diffuse-scattering, bounce-back).
Datasets
Not discussed in the paper. The paper focuses on numerical simulations and comparisons with theoretical solutions or existing benchmark data.
Key Insights and Ideas
- The Power of Weighting: The introduction of the $|\xi|^{D-1}$ weight function is a critical insight. It dramatically improves the accuracy of macroscopic quantities, especially heat flux, which was a significant issue in previous models. This suggests that how one "averages" or "recovers" macroscopic information from discrete velocity distributions is as important as the discretization itself.
- Internal Degrees of Freedom Matter: Including internal molecular degrees of freedom leads to more physically realistic simulations, particularly for phenomena where these properties are significant.
- DVDM as a Versatile Framework: The DVDM framework proves to be flexible, allowing for the integration of various methods (DVM, EQMOM, HSM) to create different spatial-time models, each with its own strengths and weaknesses.
- Efficiency of Optimization: The use of optimization techniques (like BFGS) to find discrete equilibria is shown to be numerically efficient, especially for a large number of discrete directions.
Key Results
- Improved Accuracy: The weighted DVDM submodels (especially DVD-DVM and DVD-HSM) demonstrate significantly improved accuracy in capturing flow discontinuities (shocks, rarefaction waves) and macroscopic profiles (density, velocity, energy, pressure) compared to unweighted models.
- Accurate Heat Flux: The weighted models correctly predict near-zero heat flux in continuum limits, a result that was not achieved by unweighted models.
- Good Agreement with Benchmarks: The simulations for 1D and 2D Riemann problems, Couette flow, and Lid-Driven Cavity flow show good agreement with theoretical solutions and existing benchmark data (DSMC, reference solutions).
- Regime Versatility: The models perform well across a range of flow regimes, from continuum to rarefied, and for different Knudsen numbers.
- DVD-DVM for Rarefied Flows: The paper suggests that the DVD-DVM is particularly well-suited for rarefied flows.
Limitations and Future Work
- DVD-EQMOM Accuracy: The DVD-EQMOM, particularly with a small number of Gaussian functions (M=2), showed less accuracy in some cases, potentially due to the lower-order approximation in velocity space or the upwind scheme used.
- DVD-HSM Limitations: The DVD-HSM also exhibited greater errors in temperature profiles, possibly due to approximations in the Hermite expansion or the upwind scheme.
- Higher-Order Schemes: The authors suggest that better results could be obtained by using higher-order numerical schemes for spatial-time models.
- 3D Flow Simulations: The paper mentions that simulating 3D flows is an ongoing project.
- Boundary Conditions for DVD-HSM: Further boundary conditions for the DVD-HSM are left for future work.
- More Directions/Nodes: For some cases, the authors note that more directions and discrete velocity nodes might be needed to reduce discrepancies.
Summary updated