Abstract
Computational fluid dynamics (CFD) for transonic flight often relies on the Full Potential Equation (FPE) for its efficiency, despite its inherent assumption of isentropic flow. This assumption leads to significant inaccuracies, including shock misplacement and non-physical multiple solutions. This paper presents an entropy correction method for the 2D unsteady FPE that bridges the gap between potential flow and Euler simulations. By modifying the density calculation using Rankine-Hugoniot relations, the proposed method successfully eliminates non-physical solutions and improves the prediction of shock strength and position.
Introduction
In aerodynamic design, engineers must balance computational speed with physical accuracy. The Full Potential Equation (FPE) is a computationally efficient model for simulating transonic flows. However, the FPE assumes isentropic flow, meaning entropy remains constant throughout the field. In reality, strong shock waves generated during transonic flight cause discrete jumps in entropy.
Ignoring these entropy changes leads to two primary failures in standard FPE solvers:
- Shock Misplacement: Shocks are typically predicted to be stronger and located further downstream than observed in physical experiments.
- Non-physical Multiple Solutions: Under certain transonic conditions, the FPE can yield multiple valid mathematical solutions for the same input, such as predicting lift on a symmetric airfoil at zero angle of attack—a phenomenon known as "ghost lift."
This research introduces a method to incorporate entropy physics into the potential flow framework, maintaining computational efficiency while approaching the accuracy of higher-fidelity Euler simulations.
Methodology
The proposed entropy correction method for the 2D unsteady full potential equation involves several key technical innovations:
Modified Density Formulation
The core of the correction lies in the density ($\rho$) calculation. Instead of the standard isentropic density, the authors scale the density by a factor of $e^{-\Delta s/R}$. The entropy jump ($\Delta s$) is calculated using the Rankine-Hugoniot relations based on the local Mach number upstream of the shock. This modification ensures that the density correctly reflects the thermodynamic state change across the shock.
Flux-Biasing and Numerical Schemes
To handle the sharp discontinuities of shock waves without numerical instability, an Osher-type flux-biasing scheme is employed. This provides the necessary artificial viscosity to capture shocks cleanly. The equations are solved using an implicit approximate factorization scheme, which decomposes multidimensional problems into simpler one-dimensional steps for efficiency.
Dynamic Grid Interpolation
For unsteady simulations, such as pitching airfoils, the authors developed a dynamic grid interpolation method. Rather than regenerating the mesh at every time step, the method interpolates between extreme grid positions. This approach was found to be significantly more accurate than simple grid rotation, which can introduce non-physical pressure distribution errors.
Results and Discussion
The method was validated against standard benchmarks, including the NACA 0012 and NLR 7301 airfoils, and compared with experimental wind tunnel data and Euler solutions.
Elimination of Multiple Solutions
A critical success of the entropy correction is the elimination of the multiple-solution phenomenon. In the NACA 0012 test case at Mach 0.84, standard potential theory incorrectly predicts an asymmetric flow and non-zero lift. The entropy-corrected method forces the solution to a physically correct symmetric state, aligning with high-fidelity Euler results.
Improved Shock Characteristics
The inclusion of the entropy jump acts as a physical constraint on the shock wave. By accounting for the energy loss across the shock, the predicted shocks are weaker and move forward on the airfoil surface. This shift significantly reduces the discrepancy between numerical predictions and experimental observations from Landon (1982) and Zwaan (1982).
Conclusion and Future Work
This research demonstrates that "smart" modifications to simpler fluid models can capture essential physics typically reserved for more complex equations. The entropy-corrected FPE provides a robust "middle ground" for transonic flow simulation.
However, limitations remain. The model is currently inviscid and does not account for boundary layer thickening or pressure recovery effects caused by viscosity. Future work will focus on extending this entropy correction to 3D wing geometries and integrating viscous-inviscid interaction models to further enhance predictive accuracy for practical aircraft design.
This work was documented under NASA Technical Memorandum 87769 at the Langley Research Center.
Summary updated