@magnusandersson4615
It should be noted that ksgs is a also a time-averaged quantity, which is not clear in this video. In Sec. 7 (time = 32:21) the ksgs looks more like a snapshot plot. That is, you need to have a time-averaged of the subgrid viscosity, otherwise this would be misleading. This method also have limited utility and working poorly for certain problem where the subgrid models return very low sgs_viscosity; perioidic hills being one benchmark example showing +99% resolved for relatively coarse hex meshes using the WALE model.
@tanmuen7955
Thank you so much for the informative video! There is just a part that I wanted to clarify with anyone who knows, for the criteria of resolving at least 80% of k, how can I judge if it is changing across time?
@tomasprejda4165
Thanks for the lecture! I am going to simulate liquid metal's heat transfer to the wall - since the Prandtl number is very low, the Reynolds-Chilton analogy is no longer valid and that is something Fluent's RANS models depend on. I do not know of an efficient way to model this phenomenon, so I guess LES is going to be the way. Do you agree with this idea? :) I've read about the "algebraic turbulent heat flux model", but that is well beyond my UDF knowledge in Fluent. If that's even possible to implement, which I doubt. Thanks again!
@harilalv8284
Great video, Here u used RANS averaged equation for Resolved part in LES. But from i understanding, we cannot use RANS average equations instead of filtering for the resolved part. Please correct if am wrong...
@skhossenali5654
Thanks for the informatic video. While we learned that turbulence is usually a 3D chaotic flow, you showed in your presentation which resembles mostly 2D domain. Can you share some light how the analsys should be done for a 3D case. Is it very same as 2D or is there any difference?
@luliluli1471
Dear Aidan, this has been a master class about ditactics in teaching CFD Thank you so much for explaining so clearly the K energy concept and its components. Now a question please: how can we find the precise location of the vertical line which separates both energy regions in the spectrum, please? Would it be simply a matter os calculating the largest sub-grid length scale? Thanks again.
@ujjwalchetan4907
Very well explained but I didn't understand how to calculate mu_sgs?
@ahmedelhawary6866
Great Video. I need some help. How to create a new field in post processor for Ansys?
@amrragheb9415
Why do we say resolved turbulent K.E. (17:14) while u', v', and w' fields we get are due to the contribution of resolved large eddies AND the contribution of modeled smaller eddies by the subgrid model? The word resolved turbulent K.E. makes me think automatically about velocity' fields from DNS.
@SoumilSahu
30:40 I had a doubt regarding this modification: Does equation 20 apply only to cells adjacent to the wall or to all cells in the mesh? It seems to me that some unnecessary computation could be avoided by limiting this modification only to cells close to the wall.
@tirudesta
Thanks for such excellent and informative videos . In equation (18), is the turbulent viscosity (𝜐_sgs) that is calculated at each cell in Smagorinsky-Lilly model ? Thanks !
@tejesdas3896
Very well explained! there are different types of sub-grid scale model, I believe. Could we have a presentation on that topic and practical application of such models, please? what is about DES vs LES ?
@martinfruhwirth6681
Thank you for helping out with all of you content! Do you have a source for model equation for the sub-grid scale turbulent kinetic energy (eqn 18 at 26:36, k_sgs = (kinematic viscosity / length scale)² )? I have seen various approaches for k_sgs, but never this one.
@aleks7964
What about dynamic viscosity mu_sgs in the formula for sub-grid length scale? Should it be taken as dynamic viscosity of the fluid or turbulent viscosity or smth else?
@SaiPrasadBodapati
Thanks Mr.Aiden can you please make a vedio on Reynolds stress model (stress-omega sub model) and it's differences with k-omega model..
@krishanchand5842
Thankyou for sharing the method how to select grid size for LES. It would be great if you share the relevant papers and sources.
@entropies1
Good stuff. But I don't think a CFD code would solve instantaneous field values and have a statistic package though. Well if it's a DNS code and use statistics to get the mean values, then it's probably right. But if it's a RANS or LES, we then already use the averaged values in the equation/code and therefore just use these averaged mean values to model turbulent viscosity, aren't we? In fact, in OpenFOAM for example, I don't see any averages specifically done through some statistics package, I just saw the "turbulence->correct();" line deals the turbulence part. If one turn turbulence off and uses very small dx and dt, it should automatically perform as an DNS solver, isn't it?
@ghebat4716
techncally my master degree professor owes u his salary because i understood literally evertyhing about LES from this videos and zero from his lectures. Super Thank YOU!
@simeonskopalik8362
Very helpful, please keep the LES series coming. Also DES would be an interesting topic!
@lukasf2648
Great video(s)! If you use e.g. the LES turbulence model WALE you have to add this to your controlDict to let OpenFOAM (here Version v1912) write out k_sgs (denoted as k or turbulenceProperties:k in the code) for you. Here also U prime2Mean is calculated to allow the determination of k_res / k_sgs as well. functions { turbulenceFields1 { type turbulenceFields; libs ("libfieldFunctionObjects.so"); // Either field or fields entries fields (k); enabled true; writeControl writeTime; executeControl timeStep; executeInterval 1; } fieldAverage1 { type fieldAverage; libs (fieldFunctionObjects); writeControl writeTime; fields ( U { mean on; prime2Mean on; base time; } turbulenceProperties:k { mean on; prime2Mean off; base time; } ); }