Learning Forum

[DoodlerD]

Hi,

Pardon me, I am a new Ansys user moving over from Nastran. I'm modeling the elastic compensation method where I must update the elemental values of Young's modulus after running an iteration and re-run the analysis iteration till convergence is reached. Convergence is set when [Sy - Svm(i)] < 5 units of stress. Is this possible in Workbench? I appreciate any leads or hints.

The equation is E(k+1) = E(k)*Sy/Svm(k),

k = iteration

E = Young's modulus of the element

Sy = Yield Stress

Svm = unaveraged maximum elemental Von Mises Stress

Thank you.

[peteroznewman]

nI am an experienced Ansys user whose new job uses Nastran most of the time. Please say more about the purpose of this elastic compensation method. What is it for? What is the practical benefit of changing the stiffness of the material? Is this done element by element? It sounds like an optimization process.nAnsys has Topology Optimization where a large block of material is meshed, then small faces on the block are used to apply loads or supports. The solver decides which elements are needed to carry the load and those retain the full density and stiffness, while elements that are carrying little load have the density and stiffness dialed down. Elements above a density threshold are kept and those elements define the new shape of a part that carries the load but is much lighter, because most of the material has been removed.n

[DoodlerD]

@peteroznewmannI'm self learning the limit load analysis as per ASME, which is different from modern topology optimization algorithms. I assume a value of load and then run an iteration as mentioned in the picture above.nHere's a sample paper: http:/citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.1079.1846&rep=rep1&type=pdfnI was unable to get any information on programming equations directly into Ansys Workbench. I assume this will require emodif and some sort of a conditional loop. I'm notasking for code, I just want to know whether it can be done, and where I should be searching to get the right information.

[DoodlerD]

For unknown reasons the previous comment was deleted. nThis is for a limit load analysis as per ASME, which is different from topology optimization. nA reference paper: 'Simple Bounds on Limit Loads by Elastic Finite Element Analysis' is available via a google search.nHow is this to be programmed into ansys Workbench? Can it be done and what would be a good reading source?Thank you! n

[DoodlerD]

@peteroznewman nI will further elaborate on the method: nFor example, a cylindrical shell clamped on its ends with an arbitrarily chosen pressure P, I run a first analysis. nI compare the unaveraged VonMises stress with the yield stress and then calculate a new Young's modulus based on the equationnI repeat the analysis by updating the Young's modulus of each element nI test for convergence when the Von Mises stress of the last iteration is close to the yield stressn

[peteroznewman]

nThe cylindrical shell clamped on its ends is best done in an Axisymmetric model. Furthermore, there is a plane of symmetry at the center of the length of the cylinder.nAxisymmetric models are constructed using a surface on the XY plane. The Y axis is the axis of rotation. Draw a rectangle with the bottom corner at X = R, Y = 0 and the top corner at X = R+T and Y = L/2 where R is the internal radius of the shell, T is the thickness of the shell and L is the length of the shell. Slice that thin rectangle into five segment that can each have its own material.nIn the Engineering Data, create five materials called can1, can2, can3, can4 and can5. On each one, check the Parameter box next to Young's modulus.nIn a Static Structural model, assign the five materials to the five segments, apply the pressure load to the inside face of the can, a Fixed support on the top edge of the can and a Y = 0 symmetry BC on the bottom edge of the can. Request five Stress plots, set to Unaveraged, and check the box for the Maximum Result to be a Parameter.nUse Design Optimization and create five constraints that the stress in each segment has a target value of the yield stress.n

[DoodlerD]

@peteroznewmannI think what you have suggested is an alternative and viable approach. If I get both methods to work, I will post them here for others to learn.