In my recent work with the Adflow solver, I aimed to calculate the lift and drag coefficients for the S809 airfoil, intending to compare these results with experimental data. However, I encountered significant challenges, particularly when utilizing different turbulent models.
Initially, I employed the Spalart-Allmaras (SA) turbulent model. Unfortunately, the results obtained did not align with the expected outcomes. To address this discrepancy, I decided to explore an alternative turbulent model. However, I faced a new issue: the alternative model failed to converge to a solution, even after increasing the maximum iterations to 50,000.
During my attempts, I observed that while using the other turbulent model, the iterations continued until reaching the maximum cycle limit. In contrast, with the SA turbulent model, the iterations shifted to AK, and I was unable to exceed the set limit of 50,000 iterations. This divergence in behavior raises concerns about the stability and reliability of the alternative model.
Given these challenges, I seek guidance on how to effectively utilize the alternative turbulent model to achieve results comparable to those obtained with the SA model. Understanding the underlying issues and receiving assistance in this matter would significantly enhance my research and analysis.
Description
In my recent work with the Adflow solver, I aimed to calculate the lift and drag coefficients for the S809 airfoil, intending to compare these results with experimental data. However, I encountered significant challenges, particularly when utilizing different turbulent models.
Initially, I employed the Spalart-Allmaras (SA) turbulent model. Unfortunately, the results obtained did not align with the expected outcomes. To address this discrepancy, I decided to explore an alternative turbulent model. However, I faced a new issue: the alternative model failed to converge to a solution, even after increasing the maximum iterations to 50,000.
During my attempts, I observed that while using the other turbulent model, the iterations continued until reaching the maximum cycle limit. In contrast, with the SA turbulent model, the iterations shifted to AK, and I was unable to exceed the set limit of 50,000 iterations. This divergence in behavior raises concerns about the stability and reliability of the alternative model.
Given these challenges, I seek guidance on how to effectively utilize the alternative turbulent model to achieve results comparable to those obtained with the SA model. Understanding the underlying issues and receiving assistance in this matter would significantly enhance my research and analysis.
Steps to reproduce issue
log_cfd.txt | All Modified ADFLOW Options: | +----------------------------------------+ {'ANKCharTimeStepType': 'VLR', 'CFL': 1.0, 'CFLCoarse': 0.5, 'L2Convergence': 1e-05, 'NKSwitchTol': 0.0001, 'acousticScaleFactor': 0.01, 'gridFile': 's809.cgns', 'lowSpeedPreconditioner': True, 'monitorVariables': ['resturb', 'resrho', 'cl', 'cd', 'yplus', 'mach'], 'nCycles': 50000, 'outputDirectory': 'cfd_solver', 'smoother': 'Runge-Kutta', 'surfaceVariables': ['p', 'vx', 'vy', 'vz', 'cf', 'ch', 'ptloss', 'cp', 'mach', 'yplus'], 'turbResScale': 10000.0, 'useNKSolver': True, 'writeTecplotSurfaceSolution': True}