Mesh Refinement and Best Practices - FEA using ANSYS - Lesson 5

[@chrisbenamati3415]

What a powerful tool to automatically converge the mesh! I'm used to manual mesh convergence studies in proprietary solvers / old school Nastran where you just have to do it over and over again. This has certainly come a long way (have been doing FEA for 15 years)

[@s.y.4676]

Hey man, Not sure if you're still on yourube. Ill be starting a new job tomorrow (my first job) where I'll have to use nastran for fea static, thermal and vibration simulations. I was wondering if you could give me some tips and tricks before I start the job. Thank you

[@MohammadAlzahraniii]

Ok, that's what you got from PHd holder. Explained the concept behind every single click on that software. Thank you.

[@Ans_aka_Inzi1073]

That's an amazing lecture! Everything was explained with great proficiency.

[@ShauryaRaikkonen]

What an awesome lecture! Thanks a lot from Wisconsin.

[@santiagobarragannoguera45]

Thank you so much for your video series it has been very helpful in finding my way around Ansys. I have only used Abaqus before and there the element type needs to be clearly selected during meshing and they have codes such as C3D8 for a linear brick element. In Ansys I am having a hard time figuring out what type of element it is because though I selected linear (so I assume there is only one integration point), the unaveraged results give a gradient within the element..

[@bradd8841]

Nice video sir!! Thanks

[@lucasauerdearaujo5234]

Excellent lesson! Congratulations and thank you so much!

[@StructuresProfH]

Thank you! I'm glad I could help.

[@kobidkunwar1807]

Thanks Dr. Brock Hedegaard

[@proticchobi9602]

Well explained. Just a small error, the solve doesn't occur at nodes it occurs at a point inside the element which is called integration point and then its extrapolated to the node.

[@StructuresProfH]

First, thanks for watching! It is a bit more nuanced than just saying everything happens at a node or integration point (I was trying to keep it simple). In the context of elasticity, the typical finite element formulation constructs the solution from nodal displacements that are interpolated over an element using shape functions. Integration points are used with these shape functions for the construction of the stiffness matrix, essentially turning a kind of energy minimization problem over a volume (requiring integration over that volume) into a set of algebraic equations. Once we reduce the problem to a system of equations in the form of Stiffness*Displacement = Force, we solve for the displacements, and these quantities reflect nodal displacements. However, the strains and stresses that we derive from these displacements (because they come from derivatives of displacement which depends on the particular choice of shape function) are NOT nodal values. This means it is correct to say, as you have done, that the stress and strain solutions are evaluated at the integration points and then extrapolated to the nodes, but at the same time the displacements themselves are nodal quantities that are interpolated throughout the element.

[@mohsenshamsi591]

Hi, I'm sorry. I had a question, if it's possible, please guide me. I have a circular geometry (pipe) inside the Ansys Machining software. I want to make a layered mesh, how is it possible?

[@fardin5059]

Hi profesor and thank you for the tutorial; I have a problem that you may know the solution; I want a .ans file to use in autodesk moldflow but I didn't figure out how to export a .ans file from ansys meshing workbench. May you help?

[@farhangymnastiar6967]

i got an error, "A failure occurred inside the mesh refinement module.". what i supposed to do?

[@fabioribeiro2480]

Still helpfull, Thanks :)

[@anilwins6415]

thankyou man

[@rezaahmadi1492]

hi prof, thank you for your helpful video, do you have any video for stress concentration factor calculation in ansys?

[@durgeshpandey7097]

hi

[@syl180ther]

Hi, in the version ANSYS 2024 R2, i uncunter those two errors ans the model don't work. -Warning:Unknown parameter name= CONC. A value of 7.888609052E-31 will be used. -error: 0 is not a valid type ID number. Type ID number must be a positive integer. The KEYOPT command is ignored. In the 2024 R1 i didnt have those errors Thanks for what you are doing thoses are nice explicative videos

[@StructuresProfH]

It's tough to diagnose what is wrong here just using your error message, but my guess is that you did not define CONC higher up in the model. It could also be something weird going on in 2024 R2 (I don't have that version, so I can't easily check that...) Anyway, try creating a material assignment in the Materials section of your model tree. Select whatever geometry you want to have the material for CONC for the scoping method, then give it some arbitrary material assignment. Preferably, for sake of organization, give it a unique material that you are not using somewhere else in the model, but it doesn't matter what the material properties actually are because we will change all that in the APDL script later. Finally, create a Command for that material assignment (right-click on the material assignment and Insert >> Commands). The only line of code you need in this command is... CONC = matid That should create the CONC variable and set it equal to the material ID of the material assignment you just created. Now when you use CONC in later commands, it knows that it should only apply to that specific material ID.

[@loveparfum]

this was super helpful! thank you! but i still haven't solved my problem ahhahaha

[@StructuresProfH]

Well, I'm glad it helped in some way, even if I didn't answer your question. Good luck with your modeling!