New journal article in the Journal of Computational Physics:

This paper presents a new data-driven adaptive computational model for simulating turbulent flow, where partial-but-incomplete measurement data is available. The model automatically adjusts the closure coefficients of the Reynolds-averaged Navier–Stokes (RANS) k–ω turbulence equations to improve agreement between the simulated flow and the measurements. This data-driven adaptive RANS k–ω (D-DARK) model is validated with 3 canonical flow geometries: pipe flow, backward-facing step, and flow around an airfoil. For all test cases, the D-DARK model improves agreement with experimental data in comparison to the results from a non-adaptive RANS k–ω model that uses standard values of the closure coefficients. For the pipe flow, adaptation is driven by mean stream-wise velocity data from 42 measurement locations along the pipe radius, and the D-DARK model reduces the average error from 5.2% to 1.1%. For the 2-dimensional backward-facing step, adaptation is driven by mean stream-wise velocity data from 100 measurement locations at 4 cross-sections of the flow. In this case, D-DARK reduces the average error from 40% to 12%. For the NACA 0012 airfoil, adaptation is driven by surface-pressure data at 25 measurement locations. The D-DARK model reduces the average error in surface-pressure coefficients from 45% to 12%.

Li, Z., Zhang, H., Bailey, S. C., Hoagg, J. B., and Martin, A., “A Data-Driven RANS k-ω approach for modeling turbulent flows,” Journal of Computational Physics, vol. 345, 2017, pp. 111–131.
doi:10.1016/j.jcp.2017.05.009