SimScale CAE Forum

Step-by-step tutorial for Session 1 of CFD Master Class


#47

Thank you so much @jousefm for a quick reply.

I’m sure they’ll help me a lot

Best wishes!


#48

@AsadAli

What a great timeless video!!! I hope I can still ask questions :wink:

I know everyone ‘sees’ the boundary layer differently in their mind, I just want to make sure what I ‘see’ at a little more detailed level is still based somewhat on reality.

You mention the ‘parabolic’ shape of velocity increase in the BL at around 8:30 into your video.

I have always considered the y velocity profile to be more ‘asymptotic’ at both the lower 0 end and the upper freestream end. I think that regardless of how close the velocity gets to 0 or freestream, they never really get there after the freesteam passes over the flat plat. I have always considered that there would never be an air particle on the surface of the plate that would never move after a freestream was introduced over it (it may take a long time to resolve its movement, but I think it will eventually move). Any comments on what my brain ‘sees’? :slight_smile:

Also, with regards to X distance. With the assumptions used for CFD calculations of viscous drag, is there an X distance on a flat plate where the boundary layer will separate?

Thanks,
Dale


#49

Hi Dale (@DaleKramer)!

This is not Asad speaking in the video but my colleague Milad (@Milad_Mafi) - just to make that clear :wink:

The “shape” of the velocity profile really depends on which state you are talking about. Turbulent flows usually have a velocity profile that have a higher gradient at the wall due to increased momentum transport.


Figure 1

On the left side you see a laminar velocity profile and on the right side a turbulent velocity profile for Poiseuille flow. Another assumption that is made when drawing these profiles is that we actually look at the “mean” because in fact there is a fluctuating velocity field instantaneously which looks a bit chaotic as depicted below.


Figure 2

This becomes important when talking about Reynolds Averaging but that’s another topic.

Separation (same case for airfoils) requires an adverse pressure gradient to change the direction of the slow fluid to the other direction (roughly speaking). On a flat plate we assume constant pressure gradient thus no adverse pressure gradient.

P.S.: Actually good material to talk about - maybe I will do a video about such stuff in the future (if even necessary) :thinking:

Also keep in mind that Boundary Layer theory is nothing easy - just have a look at the book from Hermann Schlichting about Boundary Layer theory, have fun :wink:


Source:

Picture 2: Turbulence Modeling for CFD by David C. Wilcox


Best,

Jousef


#50

I was not questioning what the shape was but whether the shape was asymptotic to 0 and to the freestream velocity :slight_smile:

But there is an X pressure gradient towards 0 shown, so I was just wondering.

Dale


#51

Hi Dale,

we have to be careful that we are speaking of the same thing. A flat plate can also have a zero pressure gradient - inclination of the plate can either increase or decrease the pressure but can you tell me where the pressure gradient is shown to be zero? Could not find it in the video, maybe I skipped too much. Anyway if it is a zero pressure gradient we assume that “inertia” keeps the fluid moving although one might assume that if no pressure gradient then no driving force which is not the whole truth.

Also can you tell me exactly what you mean with asymptotic to zero at which position?

Best,

Jousef


#52

Here is the chart from video:
Xchange

All his boundary layer U magnitude as less at position x=X than at position x=0. I am just asking at some time WAY into the future what does the current assumptions of CFD show there?

For U magnitude, it would be asymptotic to 0 at y=0 and asymptotic to the freestream values for large y values.

Sorry, just started beer time and had to answer quick :wink:

Dale


#53

Hi Dale,

maybe to sum things up: In the future you will see the boundary layer reach a maximum thickness called \delta which is different for turbulent and laminar flow over a flat plate, namely Blasius solution. The velocity profile would not change in the case of laminar or turbulent but it would when we look at transitional behavior which is again another level because this will include phenomena like the lift-up effect etc.

Also quite late here, going to bed soon - hope I did not mix up “easy” explanations :wink:

Best,

Jousef


#54

Jousef,

Sorry I don’t think I am asking the question so that it is decipherable. I can live without an answer but I think I will continue to ‘see’ the BL as having asymptotic magnitude values approaching y=0 and y=BL thickness (ie I think asymptotic values would minimize magnitude discontinuities at the boundaries of the BL).

Thanks,
Dale


#55

Hi Dale,

maybe it’s a misunderstanding from my side :smiley: If you could upload a small sketch that would be of great help here :slight_smile:

Cheers!

Jousef


#56

in this CFD master class workshop i am done with first meshing but second meshing which content “WITH LAYER”
are face some problem in meshing.

due to limitation i cant go with higher memory run.
please help me.
thank you,
Rohit


#57

Hi Rohit,

please always share the project links in your posts as well, makes it way easier for us to immediately have a look at the project(s). You can decrease the level of the refinements by one and test that - I assume that you used the options given in the exercise?

Best,

Jousef


#58

project link
https://www.simscale.com/workbench/?pid=7648186729059263142&mi=spec%3A9d54980f-500f-480d-8307-f92fa1df9b68%2Cservice%3ASIMULATION%2Cstrategy%3A1


#59

Hi Rohit,

as previously mentioned reduce the refinement levels, that will do the job. What you can do is to start coarse, output a force or any other quantity and use that as some kind of convergence criterium, as long as no oscillating residuals can be observed that approach is the state of the art procedure in CFD to see when a mesh is sufficiently fine - FEA does the same but is not dependent on the residuals.

Let me know how things go!

Jousef


#60

I will try this.:grinning:


#61

HI SIR ,
i am follow the same step as you say
I was reduce the refinement level and its work :wink:
but there is problem in run time.

project link : https://www.simscale.com/workbench/?pid=7648186729059263142&mi=spec%3A269d5f5e-54f7-4282-83dc-9eff3f5331f7%2Cservice%3ASIMULATION%2Cstrategy%3A2

help me
thank you,
ROHIT


Where do i find topological entity set?
#62

hi everyone,
please help me to solve above issue :point_up_2:
project link:
https://www.simscale.com/workbench/?pid=7648186729059263142&mi=spec%3A269d5f5e-54f7-4282-83dc-9eff3f5331f7%2Cservice%3ASIMULATION%2Cstrategy%3A2&ps=domain

thank you,
rohit