4 point bending test of RC Beam — Convergence Issues

- February 27, 2024 at 1:22 pm
  
  Rommel Jr. Landicho
  Subscriber
  
  Hello everyone! I am trying to model an RC Beam subjected to a 4-point bending test using ANSYS 2023 R2 but I am encountering two major problems:
  
  (1) The results (force-deflection curve) are overestimated. I know because I am comparing the results with another study, and
  (2) The solution does not converge.
  
  Here are the specifics of my model:
  
  > Quarter of RC beam with steel reinforcement; Symmetry regions are applied at both faces as shown in the pictures
  > Contact surfaces: Frictionless; Pure Penalty with Normal Stiffness Factor of 0.01 (I have read that this helps with convergence but affects accuracy); Interface Treatment is adjusted to touch
  > Boundary Conditions: Displacement at the bottom face of the rectangular support (Pin - UX UY UZ are 0), symmetry faces (UX and UZ are 0 while UY is free), and a Remote Displacement set at 8mm at the upper portion of the circular loading (I am unsure what the difference of remote displacement and displacement is).
  > Load steps: (initial and minimum - 100, maximum 1000)
  > I used the following material models:
  Concrete - Menetrey-Willam Model w/ Exponential Softening
  Steel - Bilinear Isotropic hardening
  Supports/Loading - Linear isotropic hardening
  (NOTE: the dilatancy angle and softening parameters are obtained from various related literature while the rest are given from the study)
  
  I modelled these by slowly introducing the parameters. I initially modelled the concrete as Menetrey-Willam WITHOUT softening parameters, and the model converged (but the force reaction are still overestimated). However, as I introduced the input parameters, the model did not converge anymore.
  
  Results:
  > The beam deforms as expected but the force reactions are overestimated. It reaches up to 200 kN even though it should only reach ~55 kN.
  > If the normal stiffness factor is set at program controlled or at 1, the force convergence is very far from the criteria. However, even if I set it at 0.01, the force convergence plot is near the criteria but still fails to converge.
  > The beam deforms only until 5.8 mm even though I entered a displacement of 8 mm.
  > I also tried using Normal Lagrange contact formulation to lessen the penetration of the circular loading. It works in lessening penetration but it still does not converge and is farther away from the convergence criteria.
  > I have checked the residual forces (as shown in the images) but I do not know how to interpret them as they are at random spots of the beam. Sometimes, they are near the supports, sometimes they are the sides of the beam.
  
  Other questions
  > I have seen some youtube tutorials wherein the displacement condition (UX & UZ =0, UY = Free) is put at the bottom face of the beam instead of the symmetry face of the beam. I have tried both and they yield very different results. How would I know which is correct?
  > In checking my model for convergence, I only look at penetration, deformed shape, and residual forces . Can you please suggest how else I can determine the problems of my model?
  > I would also appreciate if someone could explain the softening parameters of the Menetrey-Willam model.
  > I am also wondering how I could manually choose the element type of each component. When I was studying ANSYS, I initially chose SOLID185 for concrete but I do not know where I can select it. In the Solution Output however, I have read that SOLID186 was automatically used.
  
  I would really appreciate any kind of help for me to model this for my study. Thank you very much!
- February 28, 2024 at 9:21 am
  
  Erik Kostson
  Ansys Employee
  
  Hi
  I would really try and simplify as musch as possible - use symmetry, remove blocks and contacts and apply boundary conditions instead - see here for a good description of a workflow for RC concrete.
  
  /forum/forums/topic/how-can-we-model-failure-in-reinforced-concrete-rc-slabs/
  (For softening parameters the user needs to come up with those values as it is a user material input)
  All the best
  Erik
  - March 2, 2024 at 9:59 am
    
    Rommel Jr. Landicho
    Subscriber
    
    Hi! Thank you so much for your response. I have tried your suggested workflow and I have the following concerns:
    (1) I removed the support and cylindrical loading block and replaced them with remote displacements. Can you verify if what I did is correct?
    Here are some pictures of my quarter beam (with symmetry applied as shown in my main post). The loading (small face at top) and pin support conditions (edge near bottom) are shown.
    I initially removed the softening parameters of the Menetrey Willam concrete just to check for the values and it converged. However, when I put the softening parameters back, the solution does not converge and it has a high residual force at the pin support. It also terminates at very few iterations so I do not get much values for comparison.
    
    The mesh at the pin support and at the loading also became distorted.
    I have tried increasing the substeps to 100 (min) and 1000 (max) but there are no changes. Can you suggest other alternatives for modelling this?
    (2) I have put the command snippets suggested in my concrete. However, the first line
    keyopt, typeids(1),2,3 causes an error in my solution output so I removed it. Would that be okay?
    (3) Also, can you please tell me what the numbers correspond to here in your command snippet?
    TBDATA, 1,2,1,1e-3,100,10,1e-20
    Thank you!
  - March 14, 2024 at 4:49 am
    
    Rommel Jr. Landicho
    Subscriber
    
    Hi again sir! My model is now converging thanks to the command snippet
    TB,CONCR,matid,,,MSOL
    TBDATA,1,2,1,1e-3,100,10,1e-20
    However, I am not sure what these values correspond to. Can someone please explain what this command does?
    Thanks!