Learning Forum

[Nedas]

Hello,

I have been working with ANSYS mechanical apdl for few months now and I'm modeling reinforced concrete beam with two predifined cracks. Beam length - 2,4 m, heigth - 0,244 m, width - 0,112 m. Reinforcement area - 3,2 cm². Beam has two supports, first constrained in y and x directions, second only in y direction. Beam loaded with two concentrated loads at 0,8 and 1,6 from beginning of the beam. (I attached beam scheme)

I chose 2D plane model and I'm using PLANE182 element type.

For concrete I chose cast iron material model, where I putted stress-strain relationship for compression and for tension. (attached stress-strain relationship diagrams)

For steel I chose multilinear isotropic hardening model. (attached diagram).

I modeled cracks by deleting specific number of elements (attached picture, crack heigth I get from analytical calculation).

For analysis options I chose small displacement static, 20 substeps with aytomatic time stepping, convergence criteria by default.

When I run nonlinear diagnostics, my calulation doesn't converge and I get problem in concrete tension zone, between cracks (attached Newton-Raphson resudual force diagram). in elements events, status of failure criteria in elements, it shows numbers in plastic strain.

What can I do to remove plastic strain limits or anything elso to help to converge solution? Thanks in advance.

[John Doyle]

Are you expecting the concrete structure to develop cracks, but not catastrophically fail? Could it be possible that the nonconvergence you are experiencing is actually indicative of a physical instability leading to catastrophic failure? If that is the case, you will not be able to solve this beyond the point where the structure becomes unstable without adding more stabilization damping or solving with arc-length method. Or perhaps run this as a full transient analysis with damping.nYou mention that you ....modeled cracks by deleting specific number of elements... I assume you are using EKILL based on certain strain limits in /POST and executing restarts? If not, that would be the recommended approach as opposed to just deleting elements.nAlso, it looks like one of your plasticity material models is BISO or BKIN with near zero tangent stiffness. Is this associated the r-Bar? Are the elements associated with this material model going plastic also? If so, perhaps it would help to increase the tangent modulus to make this material model more forgiving in terms of convergence once plastic strains begin to develop.

[Nedas]

I have a data from reinforced concrete beam bending experiment. Beam was loaded be steps from 5 kNm, 6 kNm, 8 kNm ..... 26 kNm. The data collected from experiment was top of the beam concrete maximum strain and stress value. At 26 kNm beam collapsed. The there is analytical calculation method, which uses real concrete stress-strain relationship to calculate sections with or without cracks. So we calculated crack height, stress and strain values of this section and we want to compare results with numerical calculation, using FEM analysis with ANSYS.nSo we have:concrete stress-strain relationship, steel yield stress, geometry and calculated cracks height.nIn my model reinforcement is recalculated into specific heigth layer to match its area with rebars used in experiment. In experiment was used 2Ø14 mm rebars so I recalculated them into 3 mm height and 112 mm width (beam width) layer.nAs you mentioned I changed from deleting elements to using EKILL command, solution remain the same. I use EKILL command on specific number of elements to get precalculated crack height prof analytical calculations.nArc-length method also didn't give me solution, transient analysis either or maybe I didn't use right parameters for transient analysis.nI changed steel material model with increased tangent mpdulus. Also I added one more point in concrete strass-strain relationship that after it reaches its maximum, for example 25,6 MPa it goes further with some tangent (attached picture). But this also doesn't help for solution to converge.nAt the first step bending moment is 5 kNm, crack height at this moment is 85 mm. I recalculated bending moment into to concentrated forces (as I show in beam scheme). Should I do more loadsteps from 1 kNm, 2 kNm, 3 kNm till 5 kNm or this doesn't help to converge the solutions and if I try to write loadstep files it gives me error that LSWRITE can't be used with EKILL.nMaybe one of the problems is that there are no slip opportunity between rebar and concrete and when rebar deforms, concrete also deforms the same, because they have same nodes, which move the same. Hope you understand what I want to say.n

[Nedas]

Are you expecting the concrete structure to develop cracks, but not catastrophically fail? Could it be possible that the nonconvergence you are experiencing is actually indicative of a physical instability leading to catastrophic failure? If that is the case, you will not be able to solve this beyond the point where the structure becomes unstable without adding more stabilization damping or solving with arc-length method. Or perhaps run this as a full transient analysis with damping.You mention that you "....modeled cracks by deleting specific number of elements..." I assume you are using EKILL based on certain strain limits in /POST and executing restarts? If not, that would be the recommended approach as opposed to just deleting elements.Also, it looks like one of your plasticity material models is BISO or BKIN with near zero tangent stiffness. Is this associated the r-Bar? Are the elements associated with this material model going plastic also? If so, perhaps it would help to increase the tangent modulus to make this material model more forgiving in terms of convergence once plastic strains begin to develop./forum/discussion/comment/108876#Comment_108876

I have a data from reinforced concrete beam bending experiment. Beam was loaded be steps from 5 kNm, 6 kNm, 8 kNm ..... 26 kNm. The data collected from experiment was top of the beam concrete maximum strain and stress value. At 26 kNm beam collapsed. The there is analytical calculation method, which uses 'real' concrete stress-strain relationship to calculate sections with or without cracks. So we calculated crack height, stress and strain values of this section and we want to compare results with numerical calculation, using FEM analysis with ANSYS.nSo we have: concrete stress-strain relationship, steel yield stress, geometry and calculated cracks height.nIn my model reinforcement is recalculated into specific heigth layer to match its area with rebars used in experiment. In experiment was used 2Ø14 mm rebars so I recalculated them into 3 mm height and 112 mm width (beam width) layer.nAs you mentioned I changed from deleting elements to using EKILL command, solution remain the same. I use EKILL command on specific number of elements to get precalculated crack height prof analytical calculations.nArc-length method also didn't give me solution, transient analysis either or maybe I didn't use right parameters for transient analysis.nI changed steel material model with increased tangent mpdulus. Also I added one more point in concrete strass-strain relationship that after it reaches its maximum, for example 25,6 MPa it goes further with some tangent (attached picture). But this also doesn't help for solution to converge.nAt the first step bending moment is 5 kNm, crack height at this moment is 85 mm. I recalculated bending moment into to concentrated forces (as I show in beam scheme). Should I do more loadsteps from 1 kNm, 2 kNm, 3 kNm till 5 kNm or this doesn't help to converge the solutions and if I try to write loadstep files it gives me error that LSWRITE can't be used with EKILL.nMaybe one of the problems is that there are no slip opportunity between rebar and concrete and when rebar deforms, concrete also deforms the same, because they have same nodes, which move the same. Hope you understand what I want to say.n