Walls are not the only pitfall for understanding this shift, but as engineering and technology gain the ability to supply validated satisfying results for countless engineering applications, the need rises to understand the phenomenology and technicalities which are somewhat coupled in the case of wall sensitive models, and by that I mean the k-ε family of turbulence models, for which of-course the Jones-Launder (1972) is the first officially closed 2-equation model which performed quite well, and became the basis for models such as essentially all the k-ω family (besides that of David Wilcox which is mostly impractical due to its extreme sensitivity at the edge of the boundary layer for free-stream values from the inlet).
The Intention of the Webinar is to expose the outreaching failure in improving RANS capabilities, or in other words, Companies like ANSYS and others (but mainly the latter. Propose WORLD WEBINAR to expose the “NEW PARADIGM” in turbulence modeling, CFD simulation engineers seem not to care to much as the vendor offers them an ease of use like no other and results in real-time.
Besides opening with a full clear history of the evolvement in turbulence modeling from Rodi (1952) to Jones-Launder (1972) and Spalding (1974), David Wilcox (1980), Florian Menter (1992), through the year 2020 there has been no significant value nor much validation from the abundance of new models proposed: LCTM, SAS, EARSM, GEKO, SBES, Monolithic (DES, DDES, SDES, IDDES…), Zonal (no clear advancement whatsover in grey area issues), etc’…
All “NEW PARADIGMS IN TURBULENCE MODELING”, which have not advanced the field of engineering CFD by more than a tiptoe.
“The New Paradigm Phenomena”
There were a few models ANSYS introduced in the past two decades, every time naming them something like “A New Paradigm”. The k-w SST is Menter’s, but it’s before his ANSYS days, and I claim most people don’t understand it. Then there is Volker-Menter LCTM transition. Now if there is any subject I’m certain I know the most about it’s transition and it’s the vaguest among all regimes. If you sometime may find the time you may listen to my transition zoom. LCTM is not a menter concept and because Fluent can not do non-Local operations they did some odd manipulation, such that they claim to “predict transition’ while Parviz Moin (director of stanford’s turbulence center) and his team are working on understanding this vague topic with the best DNS could provide yet still it is incomplete work, There are still debates on the mechanisms themselves. Then there was Menter and Egorov Scale Adaptive Simulation – I will dedicate another post on SAS just to describe why it is not possible to get higher moments in validation with a model which in essence is an EVM that models ALL turbulence. I bet my life you have never used it, The pictures are not even so impressive Q-Factor wise, and the von-karman length scale has nothing to do with some specific wave number content just because of the mesh resolution. ANSYS contributed only finding a problem in Spalart first monolithic formulation of DES, the “grid induced separation”, so they offered DDES, SDDES, IDDES, and still the model functions well only when there is a global instability, and switches oddly between RANS and LES, never passing a LES higher moment validation,. Their contribution to zonal LES is even worse! It’s one generation to have a vortex generator at the inlet (where you may also recycle instead), but they simply have not found any way to handle the grey area. A LES zone get’s no turbulent content or following Davidson offer (not Menters) using SAS to generate turbulent content in the shift from a RANS zone to a LES zone is exactly like adding a vortex generator, except there is a difference when you do that in the entrance where a LES content has the opportunity to evolve then simply laying it in the grey area, where a) it doesn’t pass LES higher moments validation. b)returns to the RANS zone with a wrong and uncontrollable eddy viscosity value. About LES there is simply nothing to talk about, it’s a deep subject with tons of theory from the 50’s till now and in it’s an academic one not a commercial vendor problem. They do sell: “SIMULATION MADE IS EASY” with Discovery Live (which is crap) and a similar one in a collaboration with CREO. In my eyes, the improvements in RANS from Jones-Launder, with a little Wilcox and Menter (let me remind that the BSL-SST document is a NASA 1992, and is called: “Improvement of 2-eq models for AERODYNAMIC flows”). The bradshaw hypothesis and SST concept solved one simple issue and it’s the fact that it more sensitive to adverse pressure gradient in AERODYNAMIC flows. It’s certainly is not known to be better in anything else, notwithstanding the fact that from RANS you get almost no data except average properties and forces due t pressure. As you know the linear boussinesq hypothesis is a constitutive relationship between the symmetric part of the velocity gradient tensor – shear. In other words rotation as the antisymmetric part of the tensor is non-existent, So they have added the curvature correction menter both SST and GRKO (as a free parameter) – but even this correction has got nothing to do with ANSYS (funny as they call it CCurv in GEKO) – It’s Spalart-Shur correction for the SA model. What’s so outstanding to me is that even some very experienced practitioners don’t understand that.
Join me to an hour of voyage which promises to change your views about the CFD simulation market:
tomer avraham is inviting you to a scheduled Zoom meeting.
Topic: From Jones-Launder (1972) to ANSYS Turbulence Modules 2020 Team – True Story Told
Time of broadcast: Jul 20, 2020