Abstract:
Second-order monotonicity-preserving schemes represented by MUSCL (Monotone Upstream-Centered Schemes for Conservation Laws) have dominated engineering computational fluid dynamics due to their strong shock-capturing capability and high computational efficiency. However, the limiter functions employed in such schemes severely degrade resolution in smooth regions, causing accuracy deterioration at extrema and low fidelity in separated flow simulations, which remains an important topic of concern in fluid dynamics. In this paper, within the HWCNS (Hybrid Cell-Edge and Cell-Node Weighted Compact Nonlinear Schemes) framework and based on the quasi-rational (QR) non-polynomial interpolation method, a highly robust third-order scheme is developed. The QR interpolation shares the same three-point stencil as the MUSCL method, achieves third-order accuracy in smooth regions without extrema, preserves monotonicity on monotonic stencils, and guarantees boundedness on stencils containing extrema. The ADR (Approximate Dispersion Relation) method is employed to analyze the spectral properties. The Sod shock tube, Leblanc shock tube, and astrophysical jet problems are used to verify the scheme's strong robustness. Engineering validation of aerodynamic forces and surface heat flux is performed on the three-dimensional double-ellipsoid and HB-2 standard models. Results demonstrate that the QR scheme exhibits significantly lower numerical dissipation than the MUSCL-minmod scheme, computes stably in extreme cases under the present grid, CFL number and flux-splitting settings without additional positivity-preserving treatment, and achieves more accurate aerodynamic force and heat flux predictions at a computational cost comparable to that of MUSCL-minmod on engineering standard models. This scheme provides a limiter-free non-polynomial high-accuracy approach for engineering CFD. As the QR scheme shares the same stencil as the mainstream MUSCL method, it can be readily integrated into existing mature CFD software.