Abstract
We present a positivity-preserving method for multi-resolution simulations of compressible flows involving extreme conditions such as near vacuum and strong discontinuities. The novelty of this work is due to two aspects. First, we extend the positivity-preserving flux limiter of (Hu et al., J Comput Phys 242, 2013) to the multi-resolution framework by modifying the prediction operator, based on the same limiting concept. Second, we develop a positivity-preserving local time stepping scheme for adaptive time marching. Instead of using fixed hierarchical time steps, the local time stepping scheme dynamically adjusts the time steps of all multi-resolution levels to maintain positivity. The method is validated and its capabilities are demonstrated by a range of test cases.
Introduction
In computational fluid dynamics (CFD), high-order numerical schemes like Weighted Essentially Non-Oscillatory (WENO) methods are prized for their ability to capture fine-scale turbulent structures and sharp shock fronts. However, these schemes face a fundamental challenge in extreme physical regimes: numerical overshoots. In scenarios involving near-vacuum conditions or massive pressure gradients, the mathematical interpolation can predict negative values for density or pressure. Because these variables must be strictly positive to remain physically meaningful, such errors lead to non-physical sound speeds (imaginary numbers) and immediate simulation failure.
This stability issue is compounded in Multi-Resolution (MR) frameworks. MR methods optimize computational resources by applying a fine grid only where high gradients exist and a coarse grid elsewhere. The processes of grid refinement (prediction) and the use of Local Time Stepping (LTS)—where different grid levels advance at different temporal increments—introduce new opportunities for positivity violations that standard solvers cannot handle.
Methodology
This research introduces two primary innovations to ensure stability without sacrificing the high-order accuracy of the underlying solver.
1. Positivity-Preserving Prediction Operator
When an MR algorithm refines a cell, it uses a high-order prediction operator (interpolation) to determine values for the new sub-cells. To prevent this step from generating negative values, the authors modified the operator to include a limiting factor. If the high-order interpolation threatens to violate positivity, the algorithm adaptively blends the result with a "safe" first-order result. This ensures that the newly created grid points remain physically valid while preserving high-order accuracy in smooth, well-behaved regions of the flow.
2. Dynamic Local Time Stepping (LTS)
Traditional LTS schemes often use a fixed hierarchy, such as the fine grid taking exactly two steps for every one step of the coarse grid. However, in extreme flows, a shock wave may move faster than the fixed "clock" allows, leading to instability. The authors developed a dynamic LTS scheme that monitors flow conditions at every sub-step. By adjusting the time steps across all resolution levels on the fly, the scheme guarantees that the simulation remains within the stability limits required to preserve positivity.
These innovations were implemented using a 5th-order WENO scheme for spatial discretization and a 2nd-order Runge-Kutta framework for time integration.
Experimental Validation
The robustness of the method was evaluated using several grueling benchmark problems designed to stress-test numerical stability:
- 1D Double-Rarefaction: Evaluates the solver's performance in near-vacuum conditions.
- 1D Two Blast-Wave Interaction: Tests the handling of extreme pressure jumps and colliding discontinuities.
- 1D Le Blanc Problem: A notoriously difficult shock tube benchmark involving massive energy ratios.
- 2D Sedov Blast Wave: A simulation of a point-source explosion requiring high symmetry and stability.
- 2D Mach-2000 Jet: A high-speed gas jet entering a medium, characterized by extreme Mach numbers and complex vortical structures.
Key Insights and Results
The Interpolation Trap
A critical realization of this work is that even if a physics solver is inherently positivity-preserving, the auxiliary acts of grid management—specifically interpolation—can introduce instability. By baking safety checks directly into the grid-management logic, the researchers closed a significant loophole in MR stability.
Mathematical Rigor and Efficiency
The paper provides formal proofs (Theorems 2 and 3) demonstrating that the modified prediction and LTS operators are guaranteed to be both positivity-preserving and conservation-preserving. Crucially, the method maintains high "compression rates," achieving the resolution of a fine uniform grid while using significantly fewer computational cells. The method successfully completed the Mach-2000 jet and Le Blanc simulations, scenarios where standard MR methods typically crash.
Conclusion and Future Work
The proposed positivity-preserving MR method offers a robust solution for simulating compressible flows under extreme conditions. While highly effective, the current proof for the dynamic LTS is tailored to 2nd-order Runge-Kutta schemes. Future research will focus on extending these proofs to higher-order time integration (3rd and 4th order) and testing the framework on more complex physical systems, such as magnetohydrodynamics (MHD) and multi-phase flows.
Summary updated