Test Cases for NASA’s Revolutionary Computational Aerosciences Technical Challenge Part 1: Standard Test Cases for Separated Flows Part 2: Standard Test Cases for Propulsion Jim DeBonis Chris Rumsey & Mujeeb Malik Inlets and Nozzles Branch Computational AeroSciences Branch NASA Glenn Research Center NASA Langley Research Center Cleveland, Ohio Hampton, Virginia 216‐433‐6581 757‐864‐2165 [email protected] [email protected] [email protected] AIAA SciTech, Kissimmee, FL, January 2015 Introduction • Revolutionary Computational Aerociences (RCA) Technical  Challenge*: – Identify and down‐select critical turbulence, transition, and numerical  method technologies for 40% reduction in predictive error against  standard test cases for turbulent separated flows, evolution of free shear  flows and shock‐boundary layer interactions on state‐of‐the‐art high  performance computing hardware. – Timeframe: by 2017 • Discussions held with AIAA Turbulence Modeling Benchmark  Working Group (TMBWG) and others – Ideas for good (representative) cases • Simple enough to be useful (avoid complex geometries that introduce  uncertainties) • Possess the relevant flow physics – Ideas for evaluation metrics – By defining “common” cases and metrics, unbiased evaluation of future  model improvements will be easier * see: https://www.aeronautics.nasa.gov/fap/aeronautical_sciences.html 2 Introduction, cont’d • Issue 1: – There are too many cases out there, and everyone has their own favorite • Solution 1: – Highlight top 2‐3 representative “primary” cases for each category, but list many  others as optional • Issue 2: – Many cases exist for which RANS “fails” but LES/DNS/hybrid “succeeds” • Solution 2: – Interpret differently: For RANS, look for 40% improvement in results; for  LES/DNS/hybrid, look for 40% improvement in time‐to‐solution for those cases  where the prediction is already good • Issue 3: – Impossible to agree on metrics; they are either too simplistic or too difficult to  define • Solution 3: – Define some relevant metrics for the primary cases, but allow some leeway and  use judgment 3 Separated Flow Cases 4 2‐D NASA Hump Greenblatt et al • Rationale for: excellent high‐quality reference experimental  data set; good 2‐D characteristics; includes both baseline and  flow control; RANS known to do poorly; eddy‐resolving methods  have been shown to do well; well‐vetted in previous workshop • Rationale against: endplates introduced some blockage M=0.1 Re =0.936 million c - Greenblatt, D., Paschal, K. B., Yao, C.-S., Harris, J., Schaeffler, N. W., Washburn, A. E., “Experimental Investigation of Separation Control Part 1: Baseline and Steady Suction,”AIAA Journal, vol. 44, no. 12, pp. 2820-2830, 2006. - Rumsey, C. L., “Turbulence Modeling Resource,” http://turbmodels.larc.nasa.gov (data posted online), and “CFD Validation of Synthetic Jets and Turbulent Separation Control,” 5 http://cfdval2004.larc.nasa.gov (data posted online). 2‐D NASA Hump (cont’d) Baseline case: Typical RANS 35% error in bubble size (overprediction) 6 2‐D NASA Hump (cont’d) Baseline case: Typical RANS turbulent shear stress too small in magnitude in separated shear layer of bubble 7 2‐D NASA Hump (cont’d) • General validation metrics: – Separation and reattachment locations – Turbulent shear stress profiles – Velocity profiles – Surface pressure and skin friction coefficients • Specific validation metrics: ((x / c) 0.665) / 0.665 sep ((x / c) 1.1) /1.1 reattach (uv/U2 ) 0.020/ 0.020   ref min@x/c0.8 – From first two, can find bubble size relative to experiment – Current typical RANS for baseline case:    (x / c) (x / c)  0.435 / 0.435  35%   reattach sep (uv /U2 )  0.020/ 0.020  45%   ref min@x/c0.8 8 Axisymmetric Transonic Bump Bachalo & Johnson • Rationale for: includes shock‐induced separation; widely‐used  dataset for many years; axisymmetry removes 2‐D questions;  RANS OK for some aspects but poor for others • Rationale against: old experiment; no experience yet with  eddy‐resolving methods M=0.875 Re =2.763 million c - Bachalo, W. D., Johnson, D. A., “Transonic, Turbulent Boundary-Layer Separation Generated on an Axisymmetric Flow Model,” AIAA Journal, vol. 24, no. 3, pp. 437-443, 1986. - Rumsey, C. L., “Turbulence Modeling Resource,” http://turbmodels.larc.nasa.gov (data posted online). 9 Axisymmetric Transonic Bump (cont’d) Typical RANS: Cp can be reasonable, but 20-30% error in bubble size (overprediction) 10