[CFD] Residuals in CFD (Part 3)

No video

[CFD] Residuals in CFD (Part 3) - Normalisation

Рет қаралды 15,545

Күн бұрын

Part 3 of the lecture series on residuals in CFD, focusing on mass imbalance, normalisation and how normalisation is different to scaling. The talk is divided into the following parts:
0:00 Introduction
1:14 An example problem
3:07 Introducing the mass imbalance
6:52 Manual scaling the mass imbalance
9:45 Why the mass imbalance is different to energy and momentum residuals
13:55 There is no matrix equation for the mass imbalance
17:16 Normalisation
24:12 Why is the maximum value taken from the first 5 iterations?
26:03 Limitations of Normalisation
31:01 Summary
33:34 Outro
#residuals #fluidmechanics101 #cfd
============================================
Some useful references
============================================
1. Ferziger & Peric, 'Computational Methods for Fluid Dynamics', 3rd Edition, Springer-Verlag, Berlin, Heidelberg, New York, 2002.
www.amazon.co.uk/Computationa...
2. ANSYS Fluent 12.0 User manual (26.13.1 Monitoring Residuals)
www.afs.enea.it/project/neptu...
3. OpenFOAM: User Guide v2012. Residual Normalisation
www.openfoam.com/documentatio...
============================================
Want to learn more?
============================================
Grab a copy of my CFD Fundamentals Course (for beginners):
www.udemy.com/course/computat...
Learn how to write your own CFD code in MATLAB and python (for intermediates):
dr-aidan-wimshurst-s-school.t...
Learn how I draw my figures and diagrams in Inkscape (for everyone):
dr-aidan-wimshurst-s-school.t...
============================================
Did you like the video?
============================================
Download the lecture slides from my website:
www.fluidmechanics101.com/pag...
Buy me a coffee to say thanks:
www.buymeacoffee.com/NKEZrhvg0
Support the channel on Patreon (and get useful extras for your CFD studies):
/ fluidmechanics101
============================================
Donations:
============================================
1) PayPal
www.paypal.me/fluidmechanics101
==================================
Disclaimer
==================================
The methods, algorithms, equations, formulae, diagrams and explanations in this talk are for educational and demonstrative purposes only. They should never be used to analyse, design, accredit or validate real scientific / engineering / mathematical structures and flow systems. For such applications, appropriate trained, qualified and accredited (SQEP) engineers / scientists should be consulted along with the appropriate documentation, procedures and engineering standards. Furthermore, the information contained within this talk has not been verified, peer reviewed or checked in any way and is likely to contain several errors. It is therefore not appropriate to use this talk itself (or any of the algorithms, equations, formulae, diagrams and explanations contained within this talk) as an academic or technical reference. The reader should consult the original references and follow the verification and validation processes adopted by your company / institution when carrying out engineering calculations and analyses. Fluid Mechanics 101 and Dr. Aidan Wimshurst are not accountable or liable in any form for the use or misuse of the information contained in this talk beyond the specific educational and demonstrative purposes for which it was intended.

Пікірлер: 52

@naldofrasco1467 2 жыл бұрын

My gosh; your presentation method, media quality and level of detail are always excellent. I don't really understand why you do it with no compensation, but I really appreciate your efforts being made available to all for free. I look forward to every video. Thank you.

@pawankhanal8472 2 жыл бұрын

you are genius man. I just found your channel and stuck here for hour and hour. Thanks a lot buddy, keep making interesting videos about CFD.

@user-zc5gy9sb1n 2 жыл бұрын

Finally I understand why in some simulations only the continuity residuals have larger values comparing to other residuals. Really helpful! Thanks!

@Tyrim 2 жыл бұрын

27:20 OH MY GOD... I was stuck for weeks on my bachelor thesis (it was a topology optimization problem for heatsinks with buoyancy-driven natural flow) because i tought my solution didn't converge.... it was a year ago by now, but oh man if this video came out a year earlier you would've saved me a lot of headaches... But well, the video is out now, so let's hope future students find it and they won't have these same headaches! keep up the good content.

@fluidmechanics101 2 жыл бұрын

😂 I will try to be a year early next time! Let's hope this at least helps future students

@Fioritoful 3 ай бұрын

I am performing a CFD analysis on a static refrigerator, and in my Master Qualification Exam, one of the professors asked me about the level of continuity residuals from my simulation. At that time, I knew it was difficult to get low residuals due to the natural convection nature of the problem (lower air velocity inside the case), but I couldn't give a solid answer. But with your explanation I will have much more basis to explain in detail for my future master's defense. Thank you, Dr. Aidan!

@jessbuildstech 2 жыл бұрын

Thank you Dr Aidan for posting part 3, I've been waiting for this!

@annabecker3294 Жыл бұрын

Thank you very mutch for this very well explained and useful talk. I was really desperate with my problem until I found your talk and made me truly understand what residuals are. Turned out I used the wrong kind of residuals to judge convergence and now I also understand why. Thanks for you very high quality content. I always enjoy your talks.

@mackr88 2 жыл бұрын

This is soooo good. You make everything so clear that some may find CFD actually a very simple tool :D

@MartijnvDorst 2 жыл бұрын

Great videos man! Looking forward to Part 4

@denyriva1894 2 жыл бұрын

Helpful as always Prof ! Thank you very much!

@matchris2284 2 жыл бұрын

Happy new year! Nice and helpful talk, looking forward to part 4. I would also like to watch a talk over URF. Thanks in advance!

@jamalalamri8282 2 жыл бұрын

thank you very much, it was valuable lecture, we are looking forward the part 4 ❤❤❤

@francisseka4727 2 жыл бұрын

Thank you very much .... You do a very good job

@ilhantalih9949 2 жыл бұрын

Thanks sir, waiting for final lecture.

@sharkracingteam1986 Жыл бұрын

Great videos. Your method is so... SIMPLE!! Congrats

@fluidmechanics101 Жыл бұрын

😂😂

@ayoubgarich8726 2 жыл бұрын

a Like for this video even if I have not played it yet

@yao8085 2 жыл бұрын

Hi Dr. Aidan I really wish more detail about the particle trajectory theory!!!! Look forward to more video about these case. And maybe something about DEM as well～

@gerhardbekker646 2 жыл бұрын

Enlightening. I thought scaled and normalised residuals were basically synonymous... I was wrong. Thanks, Aidan!

@jasonchu852 Жыл бұрын

amazing

@EclecticVibe 2 жыл бұрын

Thank you for another well explained video! You are like Feynman of CFD;)

@fluidmechanics101 2 жыл бұрын

I wish I was! Feynman is a legend

@EclecticVibe 2 жыл бұрын

@@fluidmechanics101 You explain complex topics ina relatable and easy manner. I would call it explained in "Feynmanic" way;). Dr.Feynman is legendary!

@sergniko 2 жыл бұрын

It was amazing talk! Going to part 4! But I think at 6:54 the 0.02/2 is not equal to 4%.

@fluidmechanics101 2 жыл бұрын

That's embarrassing 😅 well spotted

@xuxu119 2 жыл бұрын

Thank you x 100000

@mohamedshaaban2920 3 ай бұрын

thanks

@bernardbayangbe2745 4 ай бұрын

Thanks

@engineerjohn7362 2 жыл бұрын

Amazing and very helpful talk. Thank you very much. We are waiting the last talk on the residuals. Can I ask you a question far from this topic? How to know what solver type to use for my simulation case(pressure or density based)? and how to choose correctly the turbulence model (k-w-sst or k-eps,...)? and their coefficients? Thank you in advance.

@fluidmechanics101 2 жыл бұрын

Generally the density based solver is good for compressible flows. Otherwise, pressure based with k omega SST with the default coefficients is fine for most cases. If you want to go further, do some comparisons against some experimental measurements and see what works best

@liangjungong4430 2 жыл бұрын

I like this video so much. It's very helpful for CFD beginners. If you don't mind, I would like to ask your permission to share this video to the other website in China such as Bilibili. For some reason KZfaq is blocked from accessing in China. Of course, I will give sources of the original website. Thank you very much.

@fluidmechanics101 2 жыл бұрын

Drop me an email and I will see what I can do: fluidmechanics101@gmail.com

@svenw8781 2 жыл бұрын

First of all let me thank you for your great channel and your skilful and pedagogical way of explaining things. Really appreciated. I wonder if you could touch the topic of under relaxation factors and their influence on residuals. How do they influence a transient solution and its convergence. And why can some under relaxation factors be lowered more than others?

@fluidmechanics101 2 жыл бұрын

Yes! My next series is going to be on under relaxation and pseudo transients. I'm just putting together the lectures at the moment. Hopefully they will be finished in the next month or so

@svenw8781 2 жыл бұрын

@@fluidmechanics101 Excellent!!!

@rogerhubert9989 2 жыл бұрын

Hi Aidan, I was wondering if your course Fundamentals 3 contains an example of how to conduct simulation like in the 2D-diffusion-structured-grid case but with an unstructured grid instead ? Thanks !

@fluidmechanics101 2 жыл бұрын

Yes! The course shows you how to write code for an unstructured mesh (rather than a structured mesh). The only thing that isn't covered is the non-orthogonal corrector, which is really complicated so I left it out of the course

@abhijitpal7202 2 жыл бұрын

Can you please do a video on LBM

@hulala6097 2 жыл бұрын

Hi Aidan. So, unlike the scale factor which got its value updated for every iteration, the normalization factor is a constant value (max absolute value of the first five iteration residual) through the whole simulation, correct?

@fluidmechanics101 2 жыл бұрын

Yep!

@NotDrJeff 2 жыл бұрын

Thank you for the great videos. Might be a stupid question, but why can't we just use the average of the (absolute) mass flux through each cell face (Either on cell-by-cell basis or throughout the whole mesh)? This would have given us the scaling factors you use in the examples at 6:53, so why wouldn't be useful in a more general case? Why is normalisation better? Also, from the user guide you linked to, OpenFOAM seems to be a bit different here. Am I correct in saying they normalise all residuals not just mass imbalance. Also, their normalisation is very different. They seem to use the average of the residual vector, which would negate the point you made about choosing a value within the first few iterations to prevent masking a divergent solution. They also calculate the difference between the residual and the average value instead of using division. How would this impact my interpretation of residuals in OpenFOAM?

@fluidmechanics101 2 жыл бұрын

Hi Jeffrey, yes I like your approach for normalising the mass flux. You could take the magnitude through each of the cell faces. Actually you could probably take whatever approach you want, as long as it is just gives a good indication to the user of convergence. My only thought is that your method might be quite expensive. Maybe the reason that the CFD authors chose their method is it is fast and cheap? I suppose we will never know as we can't really ask them 😅 As for the OpenFOAM approach, I haven't looked into it too much (other than reading the user manual), so your guess is as good as mine

@NotDrJeff 2 жыл бұрын

@@fluidmechanics101 Wow, thanks for the swift response!

@fluidmechanics101 2 жыл бұрын

No problem. Always happy to help ☺️

@atifali3941 2 жыл бұрын

Why is this so addictive?

@arjunputhusseri2545 2 жыл бұрын

video 4 is hidden 🙄

@gurubasavaraj2560 2 жыл бұрын

It would be better if you increase your voice a little bit. Even in full volume, your voice is hardly reaching its peak. The issue is the same in all videos. (PS: My ears and the devices -(apple) are in good condition).

@fluidmechanics101 2 жыл бұрын

Thanks for the feedback. I have the sound increased to a max of -5dB in my latest videos. If this is too quiet I will see if I can amplify further

@gurubasavaraj2560 2 жыл бұрын

@@fluidmechanics101 in this it's good. But other videos it's little less.

@gurubasavaraj2560 2 жыл бұрын

@@fluidmechanics101 I really appreciate your effort and knowledge share thank you very much.

@fluidmechanics101 2 жыл бұрын

Yea, maybe it was some of my old videos before I learned how to fix the audio 😂