Hey Dr. Aidan, I am also visiting three of your courses on Udemy and have to say thank you for all of that! You do not have just the knowledge about all of that, you also are able to explain it all the time in such a clear way, that everone who is interested can easily follow, understand and using it. Thank you very much!
I just want to comment on how your videos have help me to understand CFD concept better than the cfd class i've taken. Thank you and i hope your channel can grow in the future.
Aidan : *does a phd in the one of the best university in the world, weekly creates video about complex CFD concepts* Also Aidan : 'meme generator is my only bookmark lol' More seriously, very nice video once again (y) It is good to know/be remembered these stuff, but focus should really be put on few regions as in practice, a ''proper'' y+ check is never really applied, or is it? For example, a proper y+ check is not possible in turbomachinery. It would need to be done at all flow rates, and if you try different designs, it will take weeks. Keep up the good work, that's amazing!
Hi Lili An, yes well spotted 😂 you are right. Carrying out proper y+ checks is rarely done in practice and is particularly important for turbomachinery and external aero applications. Hopefully this video pushes everyone in the right direction for how to start their y+ and mesh checks.
Oh and for flow rates, you can always choose the high rpm/high speed case for your y+ checks, as the boundary layer will be thinnest and your y+ values will be largest (for a fixed mesh). As you drop the speed, the boundary layer thickens and y+ will reduce, improving your solution accuracy.
@@fluidmechanics101 hey I love your videos, they really allow me to grasp some concepts in a way no textbook can. Just to hear someone explain plainly and directly to me is so helpful. I wanted to ask how do external aerodynamics y+ is verified and how does it affect lift and drag predictions
The trick is to think of y+ as 'the height of the first cell off the wall'. So yes, y+ has been looked at extensively in the field of aerodynamics and it is generally accepted that y+ ~ 1 gives accurate results. However, you can still get errors from other sources .... mesh refinement isn't everything!
Yes, i realise this video is really a CFD basics video. But it is often overlooked and not really explained well for beginners. Hopefully this will be a useful resource for beginners to have a quick look at and set them on the right path. Thanks again for your support!
This is what exactly makes you different, being simple yet precise. CFD is currently an essential tool not only for graduate students, but also undergraduate engineering students, as well as practicing engineers. From this point of view, your contribution is unique, since it covers both basics and advanced topics. You are expected to make a lesson on the turbulence models, their performance on some prototype problems.
I have been learning the fundamentals of CFD from your channel and it's totally mind-blowing how you had exquisitely brought every flower of CFD into a bokeh.
Love your work! However I believe I've found a small mistake in slide 6, in the skin friction coefficient formula. The 6.5 should be 0,65, I think. Please keep continuing producing these high quality videos with high quality explanations! Thank you!
Im glad you found this channel too! My aim is to put out lots of extensive videos for all areas of CFD, as it is quite a difficult subject to get into, particularly for beginners :) just drop a comment if there is anything you are unsure about or need more information/explanation
@@fluidmechanics101 Thanks for your reply. Can you please make a video about boundary layer separation? Why does recirculation happen? why and when we should care about the separation of boundary? Does it matter in internal flows?
Sure. As a quick explanation, separation and recirculations lead to additional pressure drops in internal flows. This increases the pumping requirements to drive the flow through the system and can lead to unsteady loading/vibrations in the supporting structure. So generally, we would like to avoid unnecessary separations! I hope this helps
Thank you once again for your top quality content. As a follow-up to this video, will you ever consider to discuss best practices for adaptive mesh refinements (strategies, criteria, and useful thoughts)? They should pave the way to the best of both worlds - computational efficiency of coarse meshes and best accuracy of fine meshes.
Excellent work！Thanks for everything you've done on sharing and propagating CFD knowledge. More concepts and details on CFD are expected. You are the best!
Thank you so much for all the clear videos! I'm working on my master thesis and I'm using FLUENT, but your videos really make me confident about my understanding regarding all the background theory and such. You're a true lifesaver at the moment haha. :) Wish you the best!
Watching all the videos from series... They are too good and level of presentation is awesome... On this same video note, could you please make a video on Mesh Convergence Criteria as this is one of the guiding criteria for Mesh
Hi Prof. Aiden! Really good video, this is nice to view and remember the bases. Btw, there's a typo in your formula of Cf of Schlichting. It's 0.65 not 6.5. Continue the great work!
Hi Alan, i am just putting together the video on PISO and PIMPLE. It should be done next week. The third course is taking me a bit longer to put together, as it is a lot more challenging! I would expect it in the new year (2020). The courses are come as PDFs with excel and python source code, so you can get them anywhere in the world 😊 ive always wanted CFD to be available to everyone, no matter who they are or where they are from. So ive tried to make everything (the youtube videos and courses) so that everyone can get access to them. Its the only way we are going to do better CFD and build a better world 😊
Great video, Aidan! The part where you mentioned about the residuals and point convergence really sparked my curiosity in checking solution convergence. I've always considered comparing solutions from historical data to validate convergence/solution. Like I would only check if the Cl and Cd values match for an airfoil and never cared much about the other residuals Can you please add a video about solution convergence to the list of videos you're going to make? Thanks in advance!
Yes that is a really good idea. I can see where you are coming from. Particularly with aerofoil data, you are usually comparing to values in the literature and experiments so we usually use these to check for convergence. For other cases you may not have any literature or experimental data for checking and we need to use everything we have to check convergence (point probes, residuals and flux balances)
@@fluidmechanics101 Thank you and yes not always do we have data to compare our solutions to! Really appreciate you responding to each and every comment on your videos!
The first guess for supersonic flow in an airfoil is fine when assuming the cf formula showed in the video? I think there are better alternatives but you said this first step is a guess anyway, so does it really matter what formula for the shear stress I use?
As this is a first guess, it is probably fine. You are really just looking for the order of magnitude of the cell thickness. Should I use 1e-4, 1e-5 or 1e-6m?
Best thing to do is compare to experimental measurements. The closest agreement is probably the best model to choose. If you don't have any data, choose a test case that is similar to your simulation which does have experimental data available 👍
What if the case is a closed flow, for example, the internal walls of a pump impeller, is it still a valid procedure? What characteristic length should be used for defining Re? Ps: Love your videos, by far the best.
Hi Adrian! First of all, your videos are awesome and extremely detailed and well explained so thank you for that. Second of all, i am currently working on CFD simulations on aerofoils with surface roughness and i am having trouble to select the target first element length using the y+
Remember: these calculations are all estimates. If your CFD simulation says that y+ is much larger than 1, then you need to refine your mesh and try again! For example, if y+ = 5, then reduce your first cell height by a factor of 5 (and adjust the growth ratio and number of layers of course) and try again. Usually you can get the y+ you want with about 1 or 2 re-meshes, so it doesn't normally take that long
Thank you. This video was very useful for understanding the inflation layer. I hope you will disscuss other meshing variables(global and local mesh) as well. What should be the element maximum size and minimum size(its in global element seed in icem) ? Is there rule or formula for calculating them for different parts like runner of francis turbine or draft tube?
You have to choose a first cell height when you mesh the aerofoil. Then run the case and plot y+. Have a look at the y+ you have and then go back to your mesh and adjust the first layer height. Repeat the process until you have the y+ you are looking for. In the same way that CFD codes use iterative solvers, the process of obtaining the right y+ and right mesh is also iterative!
Hey Aiden. In the previous "what is the difference between y+ and y*" video, you mentioned that we should use y* to minimize the computational cost. Then in the real world task, should I do the initial y+ guess as you showed in this video and assume y+ =y*?
Hello Aidan, this is a great content. I have a wuestion. Once we set up a first layer height, how many inflation layer we should use, what should be the total height for sum of inflation layer until we proces witth lets Say with normalna hex mesh.
Just excellent and i am really impressed by the quality of your presentation ! , please i have question how to estimate the Y+ for natural convection case inside square cavity since we don't have velocity to calculate Reynolds number ! thank you so much again
Hi Leila, what i would do is make a really coarse grid first in 2D. Go with maybe 50 by 50 cells. It should run in a few seconds. Then look at the solution and see what the velocity and y+ are. You can use this as a starting point for your full grid. As a rule of thumb 10^-3 usually seems to give y+ of order 100, while 10^-5 seems to give y+ of order 5 for many flows. See how it goes 😊
@@fluidmechanics101 Thank you so much , please is it possible to give me your e-mail , i have some questions about convergence in 3D vented cavity (mixed convection) under turbulent regime with Fluent . thank you so much again !
Thank you for this wonderful video, i just have one question, what if we do not have an estimate of the velocity at the beginning of the calculation, how we do calculate the Reynolds number ?
Good question! It depends on your flow scenario. I would have a look in the literature and see if you can guess a reasonable flow rate for your case. Ultimately this is just a first guess and you will have to go back to your mesh and refine it anyway, so what would be a sensible flow rate? Internal flows are probably going to be less than 1-2 m/s, external flows are probably not going to be more than 10-15 m/s unless you have a very gusty wind and aerofoils /wings are usually going to be subsonic. Does that give you a good idea?
W8, does your online calc spit out the yp value (wall-centroid distance) or the cell height (2*yp)? Btw I ve been watching whole lotta videos of yours lately, keep up the good(great) work there. Really appreciate it man edit: I ve been wondering, why not fully resolve the b.layer by adding many grid layers around our body instead of modelling with wall functions? Is it just a matter of computational time?
Yep the calculator should give you both. Yes, in an ideal world we would always fully resolve the boundary layer but sometimes it is too expensive so we are forced to use wall functions
How do we fix the initial mesh size for any give problem? Is it dependent on some geometrical parameter or so? Do I start with some 1mm or 0.5mm ? Can you throw some light on this? Thanks in advance. 🙌🙌
Have a look at the geometry and look at your large and small features. If your large features are ~1m in size and your small features are 0.05m in size, then maybe go for a cell size of 0.02m. Usually it is better to start big and then go smaller. If you start small, you might generate way too many cells and freeze your computer!
why can't we use equation 4 to find the wall shear stress in the wall function video? (this would save us from the need to that kinematic viscosity linear approximation, or the need for an iterative scheme?). is it because this is just an initial guess?
Equation 4 is for the flow over a flat plate with zero pressure gradient. This is just an initial guess as most CFD cases will have different flow conditions near the wall and equation 4 is not exactly correct
The inflation later calculator also calculates yH, so it should do everything you need. There is a PDF guide at the bottom of the page which explains how it works
Amazing as always. If I have a number of elements limitation of 512k, and I am required to run a CFD simulation of a multi-element wing. How would you suggest targeting and achieving the y+ value? Where do I make compromises, on the virtual wind tunnel dimension? On the growth rate? Thank you for the super informative presentation.
You are going to have a tough time with a limit of only 512k! I would recommend trying a 2D simulation first, with a very fine mesh (all 512k) and then reduce the cells in your 2D simulation to see how innaccurate it gets. Once you have a level that you are ok with, then use those same settings on your 3D wing. You will probably need to reduce the upstream and downstream lengths, increase your y+ all the way up to ~200, increase the growth ratio to 1.2 and reduce the number of cells along the span. It will definitely be innaccurate but doing 2D first should get you a good idea of how innaccurate and you can quantify this limitation in your report 👍
i did the calculations on matlab, and the yh value i find is different to the one on the calclulator (1.5^10-4 vs 3.2^10-4 on the calculator of yours). Can you help me understand which way to go? By the initial guest of the calculator, or by the value i found from by the equations you explain on the video? Also you say initial height but the symbol is yp so whats really the number we get from your calculator? btw you are very helpful, glad i found your channel
Do the second and subsequent cell heights matter in getting the proper results? In other words can we keep a relatively large inflation ratio so that we satisfy the y+ condition also and keep the total cell count minimum?
The second and third cell heights do matter, as you need to make sure you have enough cells through the thickness of the boundary layer (I am making a video on this at the moment). The best thing to do is set your first cell height to get y+ and then try and keep your growth ratio less than 1.2 (1.1 is usually best), then you should be fine
If you have a 'difficult' flow, like separation over the back of an aerofoil, natural convection driven flow from a hot surface or an impinging jet, then go for y+ ~ 1. If your flow is close to flow over a flat plate or an isothermal pipe flow then you can be more relaxed and go for y+ ~ 30. In general, if you are unsure, try for y+~ 1.
Thanks for your tutorial! According to your vedio, I understands that the desire y plus should change until the result in post processing that are agree with experimental results. if I dont have experimental results, how should i choise for the best desire y plus?
If you dont have any experiments, try and get y+< 1. This is the most reliable. Y+ < 5 is usually fine as long as you dont have separation/impingement 👍
Hey! first of all thank you for all your work, it's helping me a lot ! And I have a question, I can't find your calculator for Y+ in your website, have you delete it since ?
Hi Aidan. Thanks for the great videos. I have almost gone though all of your videos. As a CFD specialist that also has worked in this field I still have problem in meshing and validation. After I receive a project I always ask: What is the efficient step by step procedure to mesh the domain? How to choose best initial resolution not to repeat meshing procedure several times ? How much to refine the mesh in critical regions? Should I care about y+ values in every surfaces? Where to care about high y+ values and where not (e.g. In very big geometries to care about this value increases the number of mesh dramatically). How to do mesh independence analysis in unstructured complex mesh? How to reduce the time I spend for mesh independence study? How to validate our simulation results when we do not have an experimental results? We can learn schemes and numerical methods from different sources. But meshing and validation in shorter time are still in question. Do you have an idea? or a source?
That's a lot of questions there .... All I can say is that a lot of it comes from experience with particular problems. Once you have meshed and solved a given application several times you get used to what the solution and mesh resolution should look like. So, maybe have a look at academic papers and see what they are using? If you have an idea of what the solution is like then you can often work backwards 👍 that's the best I can do in a paragraph 😄
@@fluidmechanics101 Thanks Adan for your response. I know there may not be a solid answer for all those questions. Just wanted to have your view and your experience. Or maybe you have seen a great reference that I am not aware of.
Take the chord length at 50% span as the length of the plate. The flat plate should do a pretty good job for getting you the right order of magnitude, as it is an external flow with a developing boundary layer 👍
In the case of a multiphase fluid, how do find the Reynolds number since you have more than one density, dynamic viscosity and maybe different velocities. Thanks for the excellent teaching
Yes you are right. Reynolds number will vary as the fluid properties vary throughout the mesh. What is the flow like near the wall? If it is mostly primary phase, I would use the viscosity and density of the primary phase. If both phases are present you can calculate a the local mixture properties using a volume fraction weighting and then use this to calculate the Reynolds number. I hope this helps!
the y* that comes from the post processing, its in a form of graf or a range of value (0-x). how can we extract a number out of the range or out of the graph, so we can then relate it to the initial guess ans then refine the mesh in order to match it ?
Hello , could you please explain to me how can I estimate the first cell height in a multiphase simulation, for example two fluids with differents viscosities? Lets say a boat in the water with a certain draft (interface bewtween air and water would be presented, the free surface). Thank you for the videos once again and if you could reply the question it would be awesome!
I would estimate the volume fraction in the cells adjacent to the wall. Use the volume fraction to calculate the dynamic viscosity and density in the cells (volume fraction average). Then estimate y+ using these values. Or just make a quick mesh, run it and see what y+ is after a few 100 iterations. Then refine your mesh 😊
@@fluidmechanics101 Thank you for the repply, I will try to do this. Just one thing, last week I send a e-mail to you, could you please say to me if you received it? If you can't suppoort me with what I asked that is no problem.
Thank you for your really nice content. I have confused on slide 12. Dose an initial cell height mean the yH or yp? Because you mentioned that yH is the height of the first layer in your content ([CFD] Inflation Layers / Prism Layers in CFD). When we make the inflation layers, height of first layer(yH) should set up double 1st Cell Height(yp), shouldn't it? Thanks.
Yes exactly! The first cell height is yH, but the centroid height is yP. Be careful with your mesh generator. Mesh generators normally specify yH, whereas the y+ calculations are based on yP.
Really great video, I also seen your previous videos on wall treatments, but I still have a confusion on the good targeted value of y+ itself, (or y* since you said in a previous video that Ansys Fluent uses y* instead of y+), so do I get from all the previous videos that the y+ (y* ?) value should always be under 1 ?
If you can get y+ and y* under 1 that is best. However, if your mesh quality is too low or your cell count is too high then you can relax this and choose to either have y+ < 5 or y+ > 30 👍
Great video, thanks Aidan! I still don't understand how to decide, if I should put my y+ in the viscous sub-layer or the log-law region? Wouldn't I always put it in the log-law region because it reduces the number of cells I need?
Hi there! If you can you should always try and put your cells in the viscous sub-layer (y+ < 5) as you will get a more accurate solution. The log law model gets less accurate under adverse pressure gradients, strong curvature and separation. However, we sometimes have no choice and cant put our cells in the viscous sub-layer as the mesh would either have too many cells or the aspect ratio/skewness is so high that we cant get a stable solution. Oy under these conditions should be we use a log law model. I hope this helps :)
Excellent video, Aidan! one question: when I use the formula in video to calculate "yp", my result is different from the result of your web calculator. and i find the Schlichting skin-friction empirical formula is different with what i find in the cfd-online y+ calculator. In Schlichting skin-friction correlation, there is a "0.65' rather than "6.5". and i love these videos, Thank you!
Hi there. I am a CFD noob and trying to learn the basics for some tasks at work to run simulations for airflow through ductwork systems and rooms. Do you offer one-on-one consulting sessions?
Can we use this for a nozzle(rect) as the Reynolds number limits are different. What will be the characteristic length for rect nozzle, is it length or equivalent Dia.
![[CFD] y+ for Laminar Flow](http://i.ytimg.com/vi/yfYr72Gc4S4/mqdefault.jpg)
![[CFD] y+ for Laminar Flow](/img/tr.png)
![[CFD] Residuals in CFD (Part 1) - Understanding Residuals](http://i.ytimg.com/vi/v9OnNeYH4Ok/mqdefault.jpg)
![[CFD] What are Wall Functions and How do they work?](http://i.ytimg.com/vi/fJDYtEGMgzs/mqdefault.jpg)
I'm not even skipping the ads inorder to give you a compliment for the service you are providing to the public.
Thank you so much 🙂 I really appreciate it
Just discovered your channel and I am impressed with the quality of the content. Keep up the good work!
Fantastic, thanks Lucas!
Honestly, one of the best videos about CFD I've ever seen.
Thank you so much 😊
i agreee....
Hey Dr. Aidan, I am also visiting three of your courses on Udemy and have to say thank you for all of that! You do not have just the knowledge about all of that, you also are able to explain it all the time in such a clear way, that everone who is interested can easily follow, understand and using it. Thank you very much!
PhD in FSI, just writing up my thesis, and I find the video very HELPFUL!!
I just want to comment on how your videos have help me to understand CFD concept better than the cfd class i've taken. Thank you and i hope your channel can grow in the future.
Thank you so much for your kind words 😊
Aidan : *does a phd in the one of the best university in the world, weekly creates video about complex CFD concepts*
Also Aidan : 'meme generator is my only bookmark lol'
More seriously, very nice video once again (y)
It is good to know/be remembered these stuff, but focus should really be put on few regions as in practice, a ''proper'' y+ check is never really applied, or is it?
For example, a proper y+ check is not possible in turbomachinery. It would need to be done at all flow rates, and if you try different designs, it will take weeks.
Keep up the good work, that's amazing!
Hi Lili An, yes well spotted 😂 you are right. Carrying out proper y+ checks is rarely done in practice and is particularly important for turbomachinery and external aero applications. Hopefully this video pushes everyone in the right direction for how to start their y+ and mesh checks.
Oh and for flow rates, you can always choose the high rpm/high speed case for your y+ checks, as the boundary layer will be thinnest and your y+ values will be largest (for a fixed mesh). As you drop the speed, the boundary layer thickens and y+ will reduce, improving your solution accuracy.
@@fluidmechanics101 hey I love your videos, they really allow me to grasp some concepts in a way no textbook can. Just to hear someone explain plainly and directly to me is so helpful. I wanted to ask how do external aerodynamics y+ is verified and how does it affect lift and drag predictions
The trick is to think of y+ as 'the height of the first cell off the wall'. So yes, y+ has been looked at extensively in the field of aerodynamics and it is generally accepted that y+ ~ 1 gives accurate results. However, you can still get errors from other sources .... mesh refinement isn't everything!
Man, you are an absolute hero.
I'm a final year mechanical engineering student and your channel is very helpful, thank you!
Thank you for your content
You just answered my crucial question in CFD!
God bless you Dr. Aidan
Excellent explanation again. Thank you for your contribution to the CFD basics.
Yes, i realise this video is really a CFD basics video. But it is often overlooked and not really explained well for beginners. Hopefully this will be a useful resource for beginners to have a quick look at and set them on the right path. Thanks again for your support!
This is what exactly makes you different, being simple yet precise. CFD is currently an essential tool not only for graduate students, but also undergraduate engineering students, as well as practicing engineers. From this point of view, your contribution is unique, since it covers both basics and advanced topics.
You are expected to make a lesson on the turbulence models, their performance on some prototype problems.
@@fluidmechanics101 hi sir
if we worked by the K Omega model, Do we need to define the Y + or no?
Thanks for creating such kind of videos. As a cfd engineer, I have to say that they are really beneficial.
I have been learning the fundamentals of CFD from your channel and it's totally mind-blowing how you had exquisitely brought every flower of CFD into a bokeh.
Really appreciate your valuable information...No one one you tube offers accurate lecturing on CFD like you.
Love your work! However I believe I've found a small mistake in slide 6, in the skin friction coefficient formula. The 6.5 should be 0,65, I think. Please keep continuing producing these high quality videos with high quality explanations! Thank you!
Yep, thanks for sporting this. A lot of people have noticed the typo. Guess I need to be a bit more careful with my equations ...
Thanks a lot for clear explanation about how get quality meshing in CFD :) Your videos are birlliant
Thanks for the videos. I am a beginner in CFD and I feel lucky to get to know about this channel.
Im glad you found this channel too! My aim is to put out lots of extensive videos for all areas of CFD, as it is quite a difficult subject to get into, particularly for beginners :) just drop a comment if there is anything you are unsure about or need more information/explanation
@@fluidmechanics101 Thanks for your reply. Can you please make a video about boundary layer separation? Why does recirculation happen? why and when we should care about the separation of boundary? Does it matter in internal flows?
Sure. As a quick explanation, separation and recirculations lead to additional pressure drops in internal flows. This increases the pumping requirements to drive the flow through the system and can lead to unsteady loading/vibrations in the supporting structure. So generally, we would like to avoid unnecessary separations! I hope this helps
Thank you once again for your top quality content. As a follow-up to this video, will you ever consider to discuss best practices for adaptive mesh refinements (strategies, criteria, and useful thoughts)? They should pave the way to the best of both worlds - computational efficiency of coarse meshes and best accuracy of fine meshes.
Excellent work！Thanks for everything you've done on sharing and propagating CFD knowledge. More concepts and details on CFD are expected. You are the best!
Thank you so much! Thats such lovely feedback
Nice idea to make a web site with all formulas. I hope there are summaries of all the other videos reliased later
Thank you so much for all the clear videos! I'm working on my master thesis and I'm using FLUENT, but your videos really make me confident about my understanding regarding all the background theory and such. You're a true lifesaver at the moment haha. :) Wish you the best!
Thanks Michelle. Best of luck with your masters thesis
@@fluidmechanics101 thanks!!
Some how youtube knew what I wanted to see and recommend it......tq YT...and these contents are awesome🔥..I just love Fluid mech...❤️
👏👏👏 you are a legend. Thank you 👍👍
Watching all the videos from series... They are too good and level of presentation is awesome...
On this same video note, could you please make a video on Mesh Convergence Criteria as this is one of the guiding criteria for Mesh
Very useful, and clear explanations!
7:10 There is a typo in equation for Cf. The formula should be (2*log10(Re)-0.65)^(-2.3)..
Yep!
THe best video and detailed explaination i was looking for my Engineering project. Thank you so much Aidan Wimshurst
Excellent ,short and clear video thanks
amazing work man
Thanks a lot for such good quality education sir
You are a beast bro! I´ll see all your videos
Hi Prof. Aiden! Really good video, this is nice to view and remember the bases. Btw, there's a typo in your formula of Cf of Schlichting. It's 0.65 not 6.5. Continue the great work!
I know! So annoyed about the typo 😂
Thanks Aiden. excellent video.
I would like a video about the SIMPLE, SIMPLEC and Coupled Differences
Do you have a tentative date for the 3rd book? I would like to buy them all together. do you ship internationally?
greetings from Argentina
Hi Alan, i am just putting together the video on PISO and PIMPLE. It should be done next week. The third course is taking me a bit longer to put together, as it is a lot more challenging! I would expect it in the new year (2020). The courses are come as PDFs with excel and python source code, so you can get them anywhere in the world 😊 ive always wanted CFD to be available to everyone, no matter who they are or where they are from. So ive tried to make everything (the youtube videos and courses) so that everyone can get access to them. Its the only way we are going to do better CFD and build a better world 😊
Great video, Aidan! The part where you mentioned about the residuals and point convergence really sparked my curiosity in checking solution convergence. I've always considered comparing solutions from historical data to validate convergence/solution. Like I would only check if the Cl and Cd values match for an airfoil and never cared much about the other residuals Can you please add a video about solution convergence to the list of videos you're going to make? Thanks in advance!
Yes that is a really good idea. I can see where you are coming from. Particularly with aerofoil data, you are usually comparing to values in the literature and experiments so we usually use these to check for convergence. For other cases you may not have any literature or experimental data for checking and we need to use everything we have to check convergence (point probes, residuals and flux balances)
@@fluidmechanics101 Thank you and yes not always do we have data to compare our solutions to! Really appreciate you responding to each and every comment on your videos!
The first guess for supersonic flow in an airfoil is fine when assuming the cf formula showed in the video? I think there are better alternatives but you said this first step is a guess anyway, so does it really matter what formula for the shear stress I use?
As this is a first guess, it is probably fine. You are really just looking for the order of magnitude of the cell thickness. Should I use 1e-4, 1e-5 or 1e-6m?
Great content. I would love to know more about how to chose the correct turbulent model for any given geometry
Best thing to do is compare to experimental measurements. The closest agreement is probably the best model to choose. If you don't have any data, choose a test case that is similar to your simulation which does have experimental data available 👍
Thank you Aidan, great Job!
Good work!! It really help to understand easily
What if the case is a closed flow, for example, the internal walls of a pump impeller, is it still a valid procedure? What characteristic length should be used for defining Re?
Ps: Love your videos, by far the best.
The chord length of the blades or perhaps the impeller diameter would probably be a good idea 👍
Great and accurate content. Appreciate the time and effort you put into these videos.
Thanks Nasim I really appreciate it 😊
excellent. thank you
It's very good video. I'm subscribed your channel. Thank you so much! With best regards from Russian naval engineer!
Thanks so much. this video is so interesting
thank you for useful data
I've just got to know about your videos. They're simply perfect and a great hep. Thank you so much.
Wow, What an explanation !!!
Hi Adrian! First of all, your videos are awesome and extremely detailed and well explained so thank you for that. Second of all, i am currently working on CFD simulations on aerofoils with surface roughness and i am having trouble to select the target first element length using the y+
Remember: these calculations are all estimates. If your CFD simulation says that y+ is much larger than 1, then you need to refine your mesh and try again! For example, if y+ = 5, then reduce your first cell height by a factor of 5 (and adjust the growth ratio and number of layers of course) and try again. Usually you can get the y+ you want with about 1 or 2 re-meshes, so it doesn't normally take that long
Thank you. This video was very useful for understanding the inflation layer. I hope you will disscuss other meshing variables(global and local mesh) as well. What should be the element maximum size and minimum size(its in global element seed in icem) ? Is there rule or formula for calculating them for different parts like runner of francis turbine or draft tube?
Thanks!
How to set y+ value for airfoil for k-omega case and k-Epsilon case
You have to choose a first cell height when you mesh the aerofoil. Then run the case and plot y+. Have a look at the y+ you have and then go back to your mesh and adjust the first layer height. Repeat the process until you have the y+ you are looking for. In the same way that CFD codes use iterative solvers, the process of obtaining the right y+ and right mesh is also iterative!
Thank you for sharing the information
No problem! Glad you found it useful 😊
Great content, well explained! Thanks
Hey Aiden. In the previous "what is the difference between y+ and y*" video, you mentioned that we should use y* to minimize the computational cost. Then in the real world task, should I do the initial y+ guess as you showed in this video and assume y+ =y*?
Yes!
An amazing Explanation
Great Video, great channel! Keep going! 👍👍
Hello Aidan, this is a great content. I have a wuestion. Once we set up a first layer height, how many inflation layer we should use, what should be the total height for sum of inflation layer until we proces witth lets Say with normalna hex mesh.
Great question. I have a video titled 'Inflation Layers' which should give you the answer to all your questions ☺️
Explained very well, thank you!
Thank you! It really helped me!!!
Your channel helps me a lot! Thank you so much!
So happy it's been helpful for you 😊
Just excellent and i am really impressed by the quality of your presentation ! , please i have question how to estimate the Y+ for natural convection case inside square cavity since we don't have velocity to calculate Reynolds number ! thank you so much again
Hi Leila, what i would do is make a really coarse grid first in 2D. Go with maybe 50 by 50 cells. It should run in a few seconds. Then look at the solution and see what the velocity and y+ are. You can use this as a starting point for your full grid. As a rule of thumb 10^-3 usually seems to give y+ of order 100, while 10^-5 seems to give y+ of order 5 for many flows. See how it goes 😊
@@fluidmechanics101 Thank you so much , please is it possible to give me your e-mail , i have some questions about convergence in 3D vented cavity (mixed convection) under turbulent regime with Fluent . thank you so much again !
Thank you for this wonderful video, i just have one question, what if we do not have an estimate of the velocity at the beginning of the calculation, how we do calculate the Reynolds number ?
Good question! It depends on your flow scenario. I would have a look in the literature and see if you can guess a reasonable flow rate for your case. Ultimately this is just a first guess and you will have to go back to your mesh and refine it anyway, so what would be a sensible flow rate? Internal flows are probably going to be less than 1-2 m/s, external flows are probably not going to be more than 10-15 m/s unless you have a very gusty wind and aerofoils /wings are usually going to be subsonic. Does that give you a good idea?
@@fluidmechanics101 thank you very much 😊
What will be the length scale for a rectangular tank with sloshing phenomenon?
Maybe the tank length or diameter? It doesn't really matter what length you choose as long as it is the same order of magnitude as the problem
Amazing content! Congratulations!
Hi! Thanks a lot for your teachings. I would like to know the theory behind overset grid. Can you kindly make a detailed lecture on it? Thanks a lot.
W8, does your online calc spit out the yp value (wall-centroid distance) or the cell height (2*yp)? Btw I ve been watching whole lotta videos of yours lately, keep up the good(great) work there. Really appreciate it man
edit: I ve been wondering, why not fully resolve the b.layer by adding many grid layers around our body instead of modelling with wall functions? Is it just a matter of computational time?
Yep the calculator should give you both. Yes, in an ideal world we would always fully resolve the boundary layer but sometimes it is too expensive so we are forced to use wall functions
very useful video thank you very much
Thank you for your effort
How do we fix the initial mesh size for any give problem? Is it dependent on some geometrical parameter or so? Do I start with some 1mm or 0.5mm ? Can you throw some light on this?
Thanks in advance. 🙌🙌
Have a look at the geometry and look at your large and small features. If your large features are ~1m in size and your small features are 0.05m in size, then maybe go for a cell size of 0.02m.
Usually it is better to start big and then go smaller. If you start small, you might generate way too many cells and freeze your computer!
@@fluidmechanics101 thank you for the explanation.
why can't we use equation 4 to find the wall shear stress in the wall function video? (this would save us from the need to that kinematic viscosity linear approximation, or the need for an iterative scheme?). is it because this is just an initial guess?
Equation 4 is for the flow over a flat plate with zero pressure gradient. This is just an initial guess as most CFD cases will have different flow conditions near the wall and equation 4 is not exactly correct
Great video. Just wanted to ask if the last term in the Schlichting equation should be 0.65 and not 6.5
Yep it is a typo
Thanks a lot! Nice work!
Great content! keep it up!
Very helpful, thank you and good luck :)
A calculator for the total layer thickness would be great... *g*
Yea, I am thinking about adding this!
I see that the calculator is not available right now. I see only inflation layer calculator. Can you tell me if im missing something?
The inflation later calculator also calculates yH, so it should do everything you need. There is a PDF guide at the bottom of the page which explains how it works
impressive lecture ，also well structured
Thank you so much, that is great and wonderful
Amazing as always. If I have a number of elements limitation of 512k, and I am required to run a CFD simulation of a multi-element wing. How would you suggest targeting and achieving the y+ value? Where do I make compromises, on the virtual wind tunnel dimension? On the growth rate? Thank you for the super informative presentation.
You are going to have a tough time with a limit of only 512k! I would recommend trying a 2D simulation first, with a very fine mesh (all 512k) and then reduce the cells in your 2D simulation to see how innaccurate it gets. Once you have a level that you are ok with, then use those same settings on your 3D wing. You will probably need to reduce the upstream and downstream lengths, increase your y+ all the way up to ~200, increase the growth ratio to 1.2 and reduce the number of cells along the span. It will definitely be innaccurate but doing 2D first should get you a good idea of how innaccurate and you can quantify this limitation in your report 👍
Mesh Y+? Y+ is just the boundary layer?
Yep, just for the first layer layer height in a boundary layer
i did the calculations on matlab, and the yh value i find is different to the one on the calclulator (1.5^10-4 vs 3.2^10-4 on the calculator of yours). Can you help me understand which way to go? By the initial guest of the calculator, or by the value i found from by the equations you explain on the video? Also you say initial height but the symbol is yp so whats really the number we get from your calculator?
btw you are very helpful, glad i found your channel
Do the second and subsequent cell heights matter in getting the proper results? In other words can we keep a relatively large inflation ratio so that we satisfy the y+ condition also and keep the total cell count minimum?
The second and third cell heights do matter, as you need to make sure you have enough cells through the thickness of the boundary layer (I am making a video on this at the moment). The best thing to do is set your first cell height to get y+ and then try and keep your growth ratio less than 1.2 (1.1 is usually best), then you should be fine
Great video! How you choose target y+?
If you have a 'difficult' flow, like separation over the back of an aerofoil, natural convection driven flow from a hot surface or an impinging jet, then go for y+ ~ 1. If your flow is close to flow over a flat plate or an isothermal pipe flow then you can be more relaxed and go for y+ ~ 30. In general, if you are unsure, try for y+~ 1.
Thanks for your tutorial!
According to your vedio, I understands that the desire y plus should change until the result in post processing that are agree with experimental results.
if I dont have experimental results, how should i choise for the best desire y plus?
If you dont have any experiments, try and get y+< 1. This is the most reliable. Y+ < 5 is usually fine as long as you dont have separation/impingement 👍
Hey! first of all thank you for all your work, it's helping me a lot ! And I have a question, I can't find your calculator for Y+ in your website, have you delete it since ?
It should be there. If you go to 'Tools' and look for 'inflation layer calculator' it should be there 👍
Hi Aidan. Thanks for the great videos. I have almost gone though all of your videos. As a CFD specialist that also has worked in this field I still have problem in meshing and validation. After I receive a project I always ask: What is the efficient step by step procedure to mesh the domain? How to choose best initial resolution not to repeat meshing procedure several times ? How much to refine the mesh in critical regions? Should I care about y+ values in every surfaces? Where to care about high y+ values and where not (e.g. In very big geometries to care about this value increases the number of mesh dramatically). How to do mesh independence analysis in unstructured complex mesh? How to reduce the time I spend for mesh independence study? How to validate our simulation results when we do not have an experimental results? We can learn schemes and numerical methods from different sources. But meshing and validation in shorter time are still in question. Do you have an idea? or a source?
That's a lot of questions there .... All I can say is that a lot of it comes from experience with particular problems. Once you have meshed and solved a given application several times you get used to what the solution and mesh resolution should look like. So, maybe have a look at academic papers and see what they are using? If you have an idea of what the solution is like then you can often work backwards 👍 that's the best I can do in a paragraph 😄
@@fluidmechanics101 Thanks Adan for your response. I know there may not be a solid answer for all those questions. Just wanted to have your view and your experience. Or maybe you have seen a great reference that I am not aware of.
Hi Dr. Aidan, In case of a 3d wing , can we use the Mean aerodynamic chord for the calculating the Reynolds number?
Yep 👍
is this applicable for hypersonic flow CFD simulation? the friction coefficient estimation
I don't think so!
I know the skin friction coefficient equation is for flat plates, but how would I approach it if I am simulating a 3D wing?
Take the chord length at 50% span as the length of the plate. The flat plate should do a pretty good job for getting you the right order of magnitude, as it is an external flow with a developing boundary layer 👍
@@fluidmechanics101 Thank you so much!
Does y+ depend on the choice of finite volume or finite element?
Good question. I haven't looked into the FEM for CFD, so I can't really say
In the case of a multiphase fluid, how do find the Reynolds number since you have more than one density, dynamic viscosity and maybe different velocities. Thanks for the excellent teaching
Yes you are right. Reynolds number will vary as the fluid properties vary throughout the mesh. What is the flow like near the wall? If it is mostly primary phase, I would use the viscosity and density of the primary phase. If both phases are present you can calculate a the local mixture properties using a volume fraction weighting and then use this to calculate the Reynolds number. I hope this helps!
the y* that comes from the post processing, its in a form of graf or a range of value (0-x). how can we extract a number out of the range or out of the graph, so we can then relate it to the initial guess ans then refine the mesh in order to match it ?
Hello , could you please explain to me how can I estimate the first cell height in a multiphase simulation, for example two fluids with differents viscosities? Lets say a boat in the water with a certain draft (interface bewtween air and water would be presented, the free surface).
Thank you for the videos once again and if you could reply the question it would be awesome!
I would estimate the volume fraction in the cells adjacent to the wall. Use the volume fraction to calculate the dynamic viscosity and density in the cells (volume fraction average). Then estimate y+ using these values.
Or just make a quick mesh, run it and see what y+ is after a few 100 iterations. Then refine your mesh 😊
@@fluidmechanics101 Thank you for the repply, I will try to do this. Just one thing, last week I send a e-mail to you, could you please say to me if you received it? If you can't suppoort me with what I asked that is no problem.
Thank you for your really nice content. I have confused on slide 12. Dose an initial cell height mean the yH or yp? Because you mentioned that yH is the height of the first layer in your content ([CFD] Inflation Layers / Prism Layers in CFD). When we make the inflation layers, height of first layer(yH) should set up double 1st Cell Height(yp), shouldn't it? Thanks.
Yes exactly! The first cell height is yH, but the centroid height is yP. Be careful with your mesh generator. Mesh generators normally specify yH, whereas the y+ calculations are based on yP.
Really great video, I also seen your previous videos on wall treatments, but I still have a confusion on the good targeted value of y+ itself, (or y* since you said in a previous video that Ansys Fluent uses y* instead of y+), so do I get from all the previous videos that the y+ (y* ?) value should always be under 1 ?
If you can get y+ and y* under 1 that is best. However, if your mesh quality is too low or your cell count is too high then you can relax this and choose to either have y+ < 5 or y+ > 30 👍
@@fluidmechanics101 okay thanks a lot
Great video, thanks Aidan! I still don't understand how to decide, if I should put my y+ in the viscous sub-layer or the log-law region? Wouldn't I always put it in the log-law region because it reduces the number of cells I need?
Hi there! If you can you should always try and put your cells in the viscous sub-layer (y+ < 5) as you will get a more accurate solution. The log law model gets less accurate under adverse pressure gradients, strong curvature and separation. However, we sometimes have no choice and cant put our cells in the viscous sub-layer as the mesh would either have too many cells or the aspect ratio/skewness is so high that we cant get a stable solution. Oy under these conditions should be we use a log law model. I hope this helps :)
yes! thank you very much!
Excellent video, Aidan!
one question: when I use the formula in video to calculate "yp", my result is different from the result of your web calculator. and i find the Schlichting skin-friction empirical formula is different with what i find in the cfd-online y+ calculator. In Schlichting skin-friction correlation, there is a "0.65' rather than "6.5".
and i love these videos, Thank you!
Ah yes, i think i made a typo here! Well spotted 😄
hi sir
if we worked by the K Omega model, Do we need to define the Y + or no?
Yep, you should probably check your values of y+ so you know what region of the boundary layer you are modelling / resolving 👍
Hi there. I am a CFD noob and trying to learn the basics for some tasks at work to run simulations for airflow through ductwork systems and rooms. Do you offer one-on-one consulting sessions?
Can we use this for a nozzle(rect) as the Reynolds number limits are different. What will be the characteristic length for rect nozzle, is it length or equivalent Dia.
Yes, you might need to use a different correlation. You should use the same length scale as that which is used in the correlation 👍
@@fluidmechanics101 thanks for answering, could you make a video on meshing for les, kolmogorov scales and how it translates to actual mesh.
Hi not quite sure which is right but in your skin coefficient you have -6.5 and in the calculator you have -0.65 which is correct?
Well spotted! This is such an unfortunate typo 😂 the value is -0.65, so the calculator is correct.