Title: “RANS and LES: Our Allies in the Quest for Turbulence”
Speaker: Philippe Spalart, Boeing
Date: Tuesday, March 28, 2017
Time: 2:00pm
Location: NASA/LaRC, IESB Bldg. 2102, Reid 2
Sponsor/Hosts: Mujeeb Malik/Christopher Rumsey
Abstract: We again take stock of the avenues for durable improvement in the prediction of turbulence in CFD, in the next two decades, with a priority on Aerospace problems. The contenders are Wall-Modeled LES (WMLES, wall-resolved LES not being practical at these Reynolds numbers), RANS, and hybrid RANS-LES approaches. We recall that Moore’s Law has finally lost its momentum thus delaying the prospects for the more expensive approaches, probably to infinity in the case of DNS. WMLES alone will not be practical for full-size wings due to the excessive ratio of boundary-layer area to the square of its properly-averaged thickness, often called Ncubes. In addition, even for research studies in apparently simple flows, lately WMLES has not been the costly-but-authoritative solution many observers expect after its success in simple channel-like flows. An example is the Bachalo-Johnson transonic bump, which so far has resisted WMLES while yielding to DNS. RANS continues to stagnate, with very few successful new ideas, and the implementation of Reynolds-Stress models in DLR and NASA codes not turning out the complex-but-authoritative solution many observers expect based on these models containing exact terms. Finally the hybrid methods including DES are practical at full Reynolds numbers and have definite Success Stories in massively-separated flows such as tandem cylinders and rocket aft-bodies. There is vibrant work, especially in Europe, and fresh ideas for instance to help the simulations develop resolved turbulence rapidly, starting from a layer in RANS mode. On the other hand, DES still lacks a clear basis in filtering, which makes systematic steps in improving the models difficult. Debate continues between proponents of seamless DES, which is more convenient, and Zonal DES, which gives better control especially in terms of initiating WMLES inside an attached boundary layer. The initial prediction of separation still depends on the RANS model. Best Practices are also difficult to establish, and mis-uses of DES and its robustness are common. The future of turbulent CFD is compound and thorny.
Bio: Philippe Spalart studied Math and Engineering in Paris, and obtained an Aerospace PhD at Stanford/NASA-Ames in 1982. Still at Ames, he then conducted DNS of transitional and turbulent boundary layers. After moving to Boeing in 1990, he created the Spalart-Allmaras one-equation RANS model. He wrote a review article and co-holds a patent on airplane trailing vortices. In 1997 he proposed the Detached-Eddy Simulation approach, blending RANS and LES to address separated flows at high Reynolds numbers at a manageable cost. Recent work includes RANS modeling, jet and airframe noise, and theoretical contributions in classical aerodynamics.