Hey Aidan, I have very recently started trying my hand in CFD in an industrial setup after a career in experimental combustion. Your videos, are almost always! on the pain points that I encounter and I am so glad to have come by them. I respect you for curating the content which in my view, is really kickass-->"no bullshit! approach" and valuing the viewer's time spent listening to your experience/solutions. Looking forward to many more. Cheers!
Awesome video. All your content is amazing and you're truly a lifesaver! It's hard to measure the impact you've created but I'm sure it's wider than you can possibly imagine. Thanks for the amazing work
Aidan as always with your content, this video is a great resource for those learning about CFD in general and meshing specifically. One teenie weenie small comment: a single polyhedral cell is a Polyhedr_on_, not polyhedr_a_ (same with the hexahedron etc)... But I might very well have been mishearing - in which case I apologize for picking up on this. Please continue to generate such high quality content - it's enormously valuable!!!!
Excellent video, as always. I just landed a career in this field and I am having such a blast! My employer is tasking me with hooking up 6 Nvidia A100s to run with the new Native GPU solver feature in Fluent 2023... It's going well, can't wait to have it fully setup
Awsome lecture. Thank you. I've been working on a LES (WALE) simulation of a flow over a flat plate, comparing results to a DNS solution, using openFoam (Foundation). Mesh was generated with blockMesh, so it was orthogonal (zero nonOrthogonal cells). I expected a good match with DNS, but it turned out to be far from it, particularly using stretched cells along the wall. Results start to get better (better matching to DNS) as the mesh is refined so that aspect ratio is close to 1. The problem is that the mesh gets really big, because the refinement close to the wall has to be extended to the whole domain. The only solution that I can figure for this is a Hexcore mesh. Let's see what happens next...
Hex meshing real world geometry while trying to keep the big volume transitions away from areas with high gradients is really difficult (impossible?). Is there any rule of thumb to get a good idea as to when tet meshing is generally better? Dynamic meshing would be a super interesting topic also!
My usual approach is to try structured hex first, then hexcore. If neither of these work then I will resort to tets. I tend to find the best approach is to use good hexes in the important areas and then do a small transition between them using tets. I will be showing some demos this year, so hopefully that will help!
Thank you so much for the unique and informative videos! Can you tell me, please, is the polyHexcore type of mesh is the only reason of appearance of hanging nodes in the mesh, or there can be another reasons?
I think this will depend on the settings in your mesh generator. Some mesh generators will produce a warning when hanging nodes are generated. It is hard to say in general though
Hi Dr Aidan! I am currently using a hexahedral core mesh, which is a good mesh type that greatly improves the convergence of the model. But in the more distant regions there will be numerical diffusion also yPlus will not satisfy the wallfunction.Will this affect the results in my relational region? (Assuming that the mesh in my region of interest is dense enough) Again no words can express my love!! :)
Try refining the mesh with a 'size field' / 'sphere of influence' / 'local refinement' in the region you are worried about and see if it affects the results 👍
Yup. Probably the mesh "push back" phase that causes this. Which is a crazy problem to have. Why not offset the surface for the boundary layer mesh, and then mesh to that offset? But ok to be fair, should just make a custom version of shm to do that, but don't have the time! :D
@@oskarelmgren it is what they are working on right now. Doing the layer first. I think (and it is only an opinion) that originally was done like that to add layers to meshes done in others meshers.
Hey Aidan are you planning to do a video un poly meshes? Lately I have seen so many people defending one or the other (hexa or poly meshes) would be great to have the other side of the coin and specifically a comparation... 😅 At the end of the day one is not entirely sure of where to go by listening contrary arguments
I work with CFD and discovering your channel was an amazing surprise! You do an excellent work with all these lectures, decomplicating CFD and making it more acessible to users. Thank you so much for you work!
![[CFD] Meshing Guide for Pipes and Ducts (O-grid, hexcore, polyhedra)](http://i.ytimg.com/vi/32cRwxAh2ys/mqdefault.jpg)
![[CFD] Meshing Guide for Pipes and Ducts (O-grid, hexcore, polyhedra)](/img/tr.png)
![[CFD] Aspect Ratio Warnings in CFD](http://i.ytimg.com/vi/oa3sxesYsKc/mqdefault.jpg)
![Tee Grizzley - Blow for Blow (feat. J. Cole) [Official Video]](http://i.ytimg.com/vi/wNLN5xsx5dc/mqdefault.jpg)
![[CFD] Pyramids, Prisms & Stair-Stepping](/img/1.gif)
Hey Aidan,
I have very recently started trying my hand in CFD in an industrial setup after a career in experimental combustion. Your videos, are almost always! on the pain points that I encounter and I am so glad to have come by them. I respect you for curating the content which in my view, is really kickass-->"no bullshit! approach" and valuing the viewer's time spent listening to your experience/solutions.
Looking forward to many more.
Cheers!
Awesome video. All your content is amazing and you're truly a lifesaver! It's hard to measure the impact you've created but I'm sure it's wider than you can possibly imagine. Thanks for the amazing work
Thank you so much. Your way of explaining makes things clear and understandable
Thank you Dr Aidan Wimshurst for taking the time and resources to make these very useful content.
Have a great day!
Love from India.
Aidan as always with your content, this video is a great resource for those learning about CFD in general and meshing specifically. One teenie weenie small comment: a single polyhedral cell is a Polyhedr_on_, not polyhedr_a_ (same with the hexahedron etc)... But I might very well have been mishearing - in which case I apologize for picking up on this. Please continue to generate such high quality content - it's enormously valuable!!!!
This is an amazing topic
thank you
Excellent video, as always. I just landed a career in this field and I am having such a blast! My employer is tasking me with hooking up 6 Nvidia A100s to run with the new Native GPU solver feature in Fluent 2023... It's going well, can't wait to have it fully setup
That sounds awesome! Well done on landing the position. I hope you really enjoy it
Thanks a lot for sharing such an insightful content on an important topic.
This is going to be good!
Awsome lecture. Thank you. I've been working on a LES (WALE) simulation of a flow over a flat plate, comparing results to a DNS solution, using openFoam (Foundation). Mesh was generated with blockMesh, so it was orthogonal (zero nonOrthogonal cells). I expected a good match with DNS, but it turned out to be far from it, particularly using stretched cells along the wall. Results start to get better (better matching to DNS) as the mesh is refined so that aspect ratio is close to 1. The problem is that the mesh gets really big, because the refinement close to the wall has to be extended to the whole domain. The only solution that I can figure for this is a Hexcore mesh. Let's see what happens next...
What a great video as usual, thanks a lot!
Amazing content
Many thanks 🙏
Very useful, now I know more.
Does the top-down meshing have similar idea with quadtree method?
Fantastic video. Is a video on polyhedral meshing on the works?
Perhaps 🙃
@@fluidmechanics101 cool. It will be interesting as I used polyhedral meshing for my dissertation, I want to see what mistakes I did :)
Hex meshing real world geometry while trying to keep the big volume transitions away from areas with high gradients is really difficult (impossible?). Is there any rule of thumb to get a good idea as to when tet meshing is generally better? Dynamic meshing would be a super interesting topic also!
My usual approach is to try structured hex first, then hexcore. If neither of these work then I will resort to tets. I tend to find the best approach is to use good hexes in the important areas and then do a small transition between them using tets. I will be showing some demos this year, so hopefully that will help!
What softwares are able to mesh hexcore?
Fluent mesher (T-Grid) and ANSA are the ones that I use
Thank you so much for the unique and informative videos! Can you tell me, please, is the polyHexcore type of mesh is the only reason of appearance of hanging nodes in the mesh, or there can be another reasons?
I think this will depend on the settings in your mesh generator. Some mesh generators will produce a warning when hanging nodes are generated. It is hard to say in general though
Hi Dr Aidan! I am currently using a hexahedral core mesh, which is a good mesh type that greatly improves the convergence of the model. But in the more distant regions there will be numerical diffusion also yPlus will not satisfy the wallfunction.Will this affect the results in my relational region? (Assuming that the mesh in my region of interest is dense enough) Again no words can express my love!! :)
Try refining the mesh with a 'size field' / 'sphere of influence' / 'local refinement' in the region you are worried about and see if it affects the results 👍
@@fluidmechanics101 Thanks a lot!!
Then one uses snappyHexMesh and finishes with 24 faced polyhedral 😂😂
Can we do multiregion meshing with snappy hex mesh? Like ev battery?
@@surajmulay6212 yes
Yup. Probably the mesh "push back" phase that causes this. Which is a crazy problem to have. Why not offset the surface for the boundary layer mesh, and then mesh to that offset?
But ok to be fair, should just make a custom version of shm to do that, but don't have the time! :D
@@oskarelmgren it is what they are working on right now. Doing the layer first. I think (and it is only an opinion) that originally was done like that to add layers to meshes done in others meshers.
@@francootaola9172 Oh, super exciting! shm could actually become a good mesher then!
Very informative, filling the lack of background knowledge on Hexcore mesh. Thank you for this brilliant explanation and video!
Amazing!
Thank You! Dr. Aidan
Love the video about mesh!! Hope there could be more!
this was a very usefull talk!
Hey Aidan are you planning to do a video un poly meshes? Lately I have seen so many people defending one or the other (hexa or poly meshes) would be great to have the other side of the coin and specifically a comparation... 😅 At the end of the day one is not entirely sure of where to go by listening contrary arguments
Yes, there is still plenty of heated discussion in the 'hexcore Vs poly' debate!
@@fluidmechanics101 I can see being a heated discussion around to try to not comment about but it would be awesome seeing your knowledge ❤️
If grid has hanging nodes, I think mostly finite volume approach is used. Are there any traditional way to employ finite difference method?
I work with CFD and discovering your channel was an amazing surprise! You do an excellent work with all these lectures, decomplicating CFD and making it more acessible to users. Thank you so much for you work!
Thank you sir
very useful !
Hello Dr. Can we use inflation in the laminar flow and how to calculate the first layer thickness or how method?
Thank you Dr Aidan Wimshurst for taking the time and resources to make these very useful content.
Have a nice day!
Love from Vietnam.