I cannot express how much of a service this is to the world, I'm doing a PhD and these videos are literally saving my ass. I come back and watch these videos over again every so often, every time I learn and understand a little more, every time I feel a sense of relief when the penny drops. Thank you so much.
I have been digging through papers for the last week or so and this explanation was very well laid out and could have saved me a lot of time on coming up to speed. Really really really well done!!! You are asset to the new folk in the CFD community, dont stop!!!
All, I can say is "I FOUND YOU, NuTILDA". This video has answered most of the questions with which i've been struggling far too long . Probably the most concise video on SA turbulence model on UA-cam. Couple of questions: 1) How would they have measured Turbulent Kinematic Viscosity experimentally , to obtain that profile? 2) When we speak about diffusion terms in the Momentum equation, What is getting diffused ? (Velocity?) How is diffusion related to viscosity? and how do we relate viscosity to concentration gradient intuitively ? 3) Can you differentiate how the SA Curvature Correction model is different from the Standard SA model? From , my experience the SA curvature correction model provides a much better correlation at higher AoA (i.e separation dominated flows), I would like to understand , why? Finally , thank you soo much for this one. P.S (Watching it on replay)
I have no words to describe how happy I am to have watched this video. Thank you a lot man. You could describe whats I have looking in just simple words that make me understand and increment my wish to keep going to the CFD world. Thank you a lot.
Thank you for the concise explanation. I am doing some work that requires me to understand this model and reading the original paper is sometimes. At least now I can read it with a clearer idea.
I think 'S' in the production term in the S-A model is the magnitude of vorticity which is related to the rotation tensor instead of the shear-rate tensor. Correct me if I am wrong. Thanks
Thank you for the very comprehensible explanation. One remark: comparing with my old cfd lecture manuscript, isnt there a c_mu missing in front of the right-hand expression in (25) ?
Great talk, thanks so much. Can you make a video about the Rotational corrections applied to the SA turb model (SA-RC) to capture streamline curvature in RANS simulations?
Hey buddy...big fan of your videos...would you mind doing a video on the LES and DES models as it is commonly used in CFD simulations...would really appreciate it..:)
For practical use, I understand SA is only applicable in exterior aerodynamics, and even there is less accurate that k-w SST. The only advantage it has is that it is computationally cheaper?
Hi Aidan, Just wondering where are the turbulent kinetic energy and other 2/3 term in the Reynolds averaged equation? Since they were present at the begging of k-epsilon model video. Waiting for your reply.
As quite an old model, I think they did not bother with the additional terms (the effect of turbulence is only accounted through adding nut to the Laplacian terms). You are of course correct, and the full divergence of the Reynolds Stress tensor is actually needed. I am looking into this at the moment because it seems to be a problem with a lot of turbulence model implementations
@@fluidmechanics101 Hello Dr Wimshurst and thank you as always for this awesome content you created. I think @NITESHKSAHU meant the difference between Equation 1 in the present video and Equation 40 in ua-cam.com/video/SVYXNICeNWA/v-deo.htmlsi=cg8wpmDK9IQqh5Qc In the Background slides there is no -2/3*rho*k*I anymore.
hi Dr.Aidan Thnak you for sharing this wonderful vedio. Could you please answer a stupid question from me. What happened if we apply Multiple reference Frame.?Which velocity are we going to use in Spalart-Allmaras transport equation? My opinion is to use a relative velocity in the LHS and all absoulte Velocity in the Rhs( production term,dissipate term and diffusion term).
Good question! I think in the MRF approach we use absolute velocity for everything, so I would go with that here. Maybe have another watch of my 'MRF video' ??
I sell them on my website for super cheap (£10 for 10 sets of slides). Its just a little bit of money that helps me pay the webpage upkeep and buy research papers to help with the videos. Or you can get them all on Patreon if you prefer
Great video, thanks! A question I had is: Why do you say that u_t = \tilde u f_{v1} has "quartic behavior"? It appears as (linear)*(cubic)/(cubic+C), which I am not sure how this is quartic?
I was wondering that there might be a typo in the momentum equation. abla \dot \tau = abla \dot ((\mu+\mu_t)(( ablaU)+( ablaU)^T)) + abla \dot (-2/3\mu( abla \dot U)I) = abla \dot ((\mu+\mu_t)(( ablaU)+( ablaU)^T)) + abla(-2/3\mu( abla \dot U)). Correct me if I am wrong. Thanks.
Great video! By the way, is it possible for you to make a lecture video on dealing with the pressure boundaries (Prgh and/or P), in both incompressible and compressible solvers? It should be very helpful to understand how OpenFOAM handles the term before choosing the Boundary/Initial conditions provided by OpenFOAM, like prgh, p, prghTotalHydrostaticPressure, prghTotalPressure, prghPressure, and many more out there. I know there are many discussions on the subject on the cfd online or other forums, but a comprehensive introduction is still absent. Anyone interested in this, please let me know. Thanks in advance.
Stupid question: In the momentum equations there should also be a term related to the Boussinesq approximation, namely -2/3 k delta_ij, right? If so, how can we find the turbulent kinetic energy k from nu_tilde? Could we also find the dissipation epsilon? Thank you for your help Aidan:)
The turbulent kinetic energy in Boussinesq approximation of shear stresses is usually neglected when using the SA model ( the term is simply droped out). However, some variants of the model exist were the KE is approximated (rarely used). You can find a good summary of all the models and the contribution of each variant to the original model in the below link: turbmodels.larc.nasa.gov/spalart.html. Hope this helps :)
Hello! thank you for your explanation. :D I've tried to run SA model in an airfoil with C-domain (structured mesh) I got y^+ below 1 but I couldn't get good results in Cd coefficient. I've tried to solve a flow of Re = 7e5 (Transitional) in OpenFOAM... I think that this Reynolds number isn't compatible with OpenFOAM's SA model formulation. What I mean is the SA model hasn't a transitional model incorporated. In numbers I've could get results with a difference around 5% and 10 % in Cl and Cm coefficients in relation to experiment, but a difference of 60% in Cd. Could you give me your opinion of what happens whit this? Best regards!
I found the answer. Effectively the OpenFOAM formulation hasn't the ft2 term, it has the same formulation that you explain in the video. I could code the standard version of SA model in OpenFOAM having good results. The formulation that you explain is only compatible with high Re cases!.
Hi, nice videos, solved many doubts about basics of CFD. I have one request/suggestion. You should also make videos on complete simulation of few standard problems. For example, I am working on vortex shedding about circular cylinder in a rectangular duct( 300 mm wide and 25 mm depth). I can found videos on the basics or the shallow simulations. But no videos are there that explain simulation of shedding phenomena using correct fundamentals of CFD and Fluid Mechanics. Please make a video on vortex shedding , Reynolds number 3000-15000 considering all basic details correctly. It will help a lot in practice/implementation of fundamentals.
Yes, i am going to start doing some example problems on this channel. The difficulty is getting hold of some good meshing software, as CFD products and licenses tend to be very expensive! Also vortex shedding is one of the most difficult flow phenomena to simulate accurately using CFD, so dont worry if you are having difficulties. It is very hard!
Yep, if you are going to run SA, try and get y+ = 1. Remember that aerofoils have adverse pressure gradients and the SA model was derived from flat plate boundary layer results with no pressure gradients. ... besides having y+ = 1 always gets reviewers/approvers/professors to leave you alone and stop being picky about the mesh ... 😏
![[CFD] The Transition SST (gamma - Re_theta) model](http://i.ytimg.com/vi/5htknS9uVEk/mqdefault.jpg)
![[CFD] The Transition SST (gamma - Re_theta) model](/img/tr.png)
![[CFD] The k - epsilon Turbulence Model](http://i.ytimg.com/vi/fOB91zQ7HJU/mqdefault.jpg)
![[CFD] Turbulence Intensity for RANS](/img/n.gif)
I cannot express how much of a service this is to the world, I'm doing a PhD and these videos are literally saving my ass. I come back and watch these videos over again every so often, every time I learn and understand a little more, every time I feel a sense of relief when the penny drops. Thank you so much.
I have been digging through papers for the last week or so and this explanation was very well laid out and could have saved me a lot of time on coming up to speed. Really really really well done!!! You are asset to the new folk in the CFD community, dont stop!!!
Thanks Bryan 😊 much appreciated!
All, I can say is "I FOUND YOU, NuTILDA". This video has answered most of the questions with which i've been struggling far too long . Probably the most concise video on SA turbulence model on UA-cam.
Couple of questions:
1) How would they have measured Turbulent Kinematic Viscosity experimentally , to obtain that profile?
2) When we speak about diffusion terms in the Momentum equation, What is getting diffused ? (Velocity?) How is diffusion related to viscosity?
and how do we relate viscosity to concentration gradient intuitively ?
3) Can you differentiate how the SA Curvature Correction model is different from the Standard SA model? From , my experience the SA curvature correction model provides a much better correlation at higher AoA (i.e separation dominated flows), I would like to understand , why?
Finally , thank you soo much for this one.
P.S (Watching it on replay)
I used to be super confused with the value nuTilda in OpenFOAM. Thank you so much for the explanation!
As usual, concise yet precise, crystal clear presentation! Great job
I have no words to describe how happy I am to have watched this video. Thank you a lot man. You could describe whats I have looking in just simple words that make me understand and increment my wish to keep going to the CFD world. Thank you a lot.
Splendid as usual. Excellent explanation
All that you have explained is quite lucid and informative. It gives a good overview to understand what actually happens when we use this model!
A FANTASTIC VIDEO! I finally start to understand some of the concepts. Thank you so much!
Awsome video! Very clear and informative.
I was waiting for this! Thank you.
Fantastic! I knew a few of you were looking for this video. It was a tricky one but well worth it 😊
Thank you for the concise explanation. I am doing some work that requires me to understand this model and reading the original paper is sometimes. At least now I can read it with a clearer idea.
Thank you! Very clear and informative =)
Wao! Great Job. Thanks for sharing.
Good job bro congratulations
thank you so much! you're explainations are the best. it is helping in understanding CFD a lot. never stop making videos.
👍👌😊
Please do a video on LES/DES models
Good lecture.
Thank you very much, that was very helpful, keep it up!
These videos are great !!!!
just fantastic!!!!! easy way to learn
I think 'S' in the production term in the S-A model is the magnitude of vorticity which is related to the rotation tensor instead of the shear-rate tensor. Correct me if I am wrong. Thanks
Thank you for the very comprehensible explanation. One remark: comparing with my old cfd lecture manuscript, isnt there a c_mu missing in front of the right-hand expression in (25) ?
Thanks Dr.Aidan
Isn't it for external flow in the main aerofoils etc
Nice episode. Could you please give a little more explanation about why nut is proportional to (y+)^4 in the viscous sublayer ?
Im not too sure on this. I think it is from experimental measurements (DNS) of the viscous sub-layer. Let me get back to you on this one!
Fantastic! Can‘t wait to see more about turbulence modeling...maybe v2-f or reynolds-stress models? 😉👌
🤔 hmmmm which one to do next ...
Maybe the v2-f-model to complete the linear Eddy viscosity models 😉
Hey...could u make a lecture video on Moving Reference Frame method!
Oooh yes, thats a great idea! Will add it to my list!
Great talk, thanks so much. Can you make a video about the Rotational corrections applied to the SA turb model (SA-RC) to capture streamline curvature in RANS simulations?
Hey buddy...big fan of your videos...would you mind doing a video on the LES and DES models as it is commonly used in CFD simulations...would really appreciate it..:)
I will start looking at LES soon. It is quite difficult, so i want to make sure i get it right 🙃
For practical use, I understand SA is only applicable in exterior aerodynamics, and even there is less accurate that k-w SST. The only advantage it has is that it is computationally cheaper?
Hi Aidan, Just wondering where are the turbulent kinetic energy and other 2/3 term in the Reynolds averaged equation? Since they were present at the begging of k-epsilon model video. Waiting for your reply.
As quite an old model, I think they did not bother with the additional terms (the effect of turbulence is only accounted through adding nut to the Laplacian terms). You are of course correct, and the full divergence of the Reynolds Stress tensor is actually needed. I am looking into this at the moment because it seems to be a problem with a lot of turbulence model implementations
@@fluidmechanics101
Hello Dr Wimshurst and thank you as always for this awesome content you created.
I think @NITESHKSAHU meant the difference between Equation 1 in the present video and Equation 40 in ua-cam.com/video/SVYXNICeNWA/v-deo.htmlsi=cg8wpmDK9IQqh5Qc
In the Background slides there is no -2/3*rho*k*I anymore.
hi Dr.Aidan
Thnak you for sharing this wonderful vedio.
Could you please answer a stupid question from me. What happened if we apply Multiple reference Frame.?Which velocity are we going to use in Spalart-Allmaras transport equation? My opinion is to use a relative velocity in the LHS and all absoulte Velocity in the Rhs( production term,dissipate term and diffusion term).
Good question! I think in the MRF approach we use absolute velocity for everything, so I would go with that here. Maybe have another watch of my 'MRF video' ??
@@fluidmechanics101 would be any extra source term?
Thanks for the awesome videos! Are the slides that you use free available?
I sell them on my website for super cheap (£10 for 10 sets of slides). Its just a little bit of money that helps me pay the webpage upkeep and buy research papers to help with the videos. Or you can get them all on Patreon if you prefer
@@fluidmechanics101 Thanks
Great video, thanks! A question I had is: Why do you say that
u_t = \tilde
u f_{v1} has "quartic behavior"? It appears as (linear)*(cubic)/(cubic+C), which I am not sure how this is quartic?
Thx
I was wondering that there might be a typo in the momentum equation.
abla \dot \tau =
abla \dot ((\mu+\mu_t)((
ablaU)+(
ablaU)^T)) +
abla \dot (-2/3\mu(
abla \dot U)I) =
abla \dot ((\mu+\mu_t)((
ablaU)+(
ablaU)^T)) +
abla(-2/3\mu(
abla \dot U)). Correct me if I am wrong. Thanks.
Ausome job
Great video! By the way, is it possible for you to make a lecture video on dealing with the pressure boundaries (Prgh and/or P), in both incompressible and compressible solvers? It should be very helpful to understand how OpenFOAM handles the term before choosing the Boundary/Initial conditions provided by OpenFOAM, like prgh, p, prghTotalHydrostaticPressure, prghTotalPressure, prghPressure, and many more out there. I know there are many discussions on the subject on the cfd online or other forums, but a comprehensive introduction is still absent. Anyone interested in this, please let me know. Thanks in advance.
Yep, i agree. Sounds like a good introductory video is needed!
Hi Aidan sir, I think equation 8 is not correct. should we not use del . (-2/3 meu.....)????? in equation 8
Yep, I think there might be a typo here!
Stupid question: In the momentum equations there should also be a term related to the Boussinesq approximation, namely -2/3 k delta_ij, right?
If so, how can we find the turbulent kinetic energy k from nu_tilde? Could we also find the dissipation epsilon?
Thank you for your help Aidan:)
The turbulent kinetic energy in Boussinesq approximation of shear stresses is usually neglected when using the SA model ( the term is simply droped out). However, some variants of the model exist were the KE is approximated (rarely used). You can find a good summary of all the models and the contribution of each variant to the original model in the below link: turbmodels.larc.nasa.gov/spalart.html. Hope this helps :)
Awesome! Thanks George
Hello! thank you for your explanation. :D I've tried to run SA model in an airfoil with C-domain (structured mesh) I got y^+ below 1 but I couldn't get good results in Cd coefficient. I've tried to solve a flow of Re = 7e5 (Transitional) in OpenFOAM... I think that this Reynolds number isn't compatible with OpenFOAM's SA model formulation. What I mean is the SA model hasn't a transitional model incorporated. In numbers I've could get results with a difference around 5% and 10 % in Cl and Cm coefficients in relation to experiment, but a difference of 60% in Cd. Could you give me your opinion of what happens whit this?
Best regards!
I found the answer. Effectively the OpenFOAM formulation hasn't the ft2 term, it has the same formulation that you explain in the video. I could code the standard version of SA model in OpenFOAM having good results. The formulation that you explain is only compatible with high Re cases!.
Hi, nice videos, solved many doubts about basics of CFD. I have one request/suggestion. You should also make videos on complete simulation of few standard problems. For example, I am working on vortex shedding about circular cylinder in a rectangular duct( 300 mm wide and 25 mm depth). I can found videos on the basics or the shallow simulations. But no videos are there that explain simulation of shedding phenomena using correct fundamentals of CFD and Fluid Mechanics. Please make a video on vortex shedding , Reynolds number 3000-15000 considering all basic details correctly. It will help a lot in practice/implementation of fundamentals.
Yes, i am going to start doing some example problems on this channel. The difficulty is getting hold of some good meshing software, as CFD products and licenses tend to be very expensive!
Also vortex shedding is one of the most difficult flow phenomena to simulate accurately using CFD, so dont worry if you are having difficulties. It is very hard!
Is S-A model suitable for rotor stator interaction under MRF and under Sliding mesh.
Maybe
Yes, it is. At least it is so for most centrifual and axial compressors.
@@thortt515 thanks a bunch man
S-A runs fast too right? Like faster than Ke. What are the y+ needs? 5》30》200
Yep, if you are going to run SA, try and get y+ = 1. Remember that aerofoils have adverse pressure gradients and the SA model was derived from flat plate boundary layer results with no pressure gradients.
... besides having y+ = 1 always gets reviewers/approvers/professors to leave you alone and stop being picky about the mesh ... 😏
@@fluidmechanics101 Thanks !!!!!
best