1 (edited by sasan.dolati2 2020-05-29 04:27:40)

Topic: Failing sooner than expected

Hi dear all
I am modeling a column and I applied 5 step axial load first and then 10 step for monotonic lateral displacement. Almost 15 step. I check the LD diagram with experimental and all matches but the problem is that according to experimental it should not fail sooner that step 15 but mine fail step 13 and get the error that "The execution is killed due to violation of stop iteration criteria” I check the 2.4.2 of Troubleshooting and I check conditional break but I did not understand what should I do with my numbers. And I think if it fixes, the answer will be get. ( WHAT CAN I DO SO IT WILL FAIL LATER?)
I try other things and check 2.1.19 Troubleshooting and use Stiffness type: Elastic predictor, Update stiffness/Assemble Stiffness: Each step and the thing that happen is that column will not fail and at the end I received and error criteria not satisfied and I add 3 more step but It did not fail again and as I said I expect from experimental that it fails step 15 or 16. I think if I add 2 other more step it will not fail again
As I said I don’t have any idea about what number should be for error tolerance.

Re: Failing sooner than expected

Dear sasan.dolati2,
1. only having 10 steps from 0 to expected failure means unusually long loading steps. I suggest to generate 20-100 steps to the expected failure displacement, and add 50-100% additional steps (see also Point 2 below).

2. The usual approach in nonlinear FEM (as well as experiments) is to apply more displacement than the expected failure (or make sure the machine range is larger than the expected failure displacement).

3. Please check the material properties and the boundary conditions.
If you are not sure, you can follow Troubleshooting, 2.1.1 to send us your model along with sketches etc.

4. I also recommend to see Troubleshooting, 2.2.22 How to detect a structural collapse or ultimate load capacity.

5. Moreover, I suggest to follow Troubleshooting, 2.2.18 My analysis results do not match the experiment/expectations. How can I improve my model?


3 (edited by sasan.dolati2 2020-06-19 04:02:43)

Re: Failing sooner than expected

Hi dear Dobromil.
As you said about increasing step,

A-what is your idea about number of iterations? I should use for example 50 step or 40 steps with 40 iterations or 100 iterations or more?

B-I understand this item that when I receive this error :convergence criteria not satisfied( line search) after some steps, I receive the error that kill due to convergence criteria, but when I did not get this error, it will continue until the end and at the last step of the end, I received analysis is complete but it is written convergence criteria not satisfied. Is it right? For example I define 20 step with 100 iteration with line search 10 it continue step 19 but with 200 iteration with line search 10 fail at 2 step.

C- I applied 200 mm and I know that it should fail 160 mm as lateral displacement in 20 steps with 100 iteration with line search 10 it complete but when I apply 200 mm with 100 step and 100 iteration with line search 10 , Model start receiving error (line search) and fail step 40. The more step should be better why it fail?
what is number for step and attrition I should use? Can we say the more step, the less iteration?

D-Only to be sure, when I mesh micro-element, reinforcement will be considered in meshing?

Re: Failing sooner than expected

Dear sasan.dolati2,
ad A-C: 1. please first see ATENA Troubleshooting, 2.1.19 Problems reaching convergence and understanding ATENA convergence parameters and 2.4.2 I get the message "The execution is killed due to violation of stop iteration criteria", what does it mean?

2. For the Line Search message, please see Troubleshooting, 2.4.7 I get the warning "Convergence criteria (within line search) not satisfied due to eta limit, continuing super-ordinate iteration" during analysis, what does it mean?

3. According to our experience, 0-3 iterations make sense for Line Search, not more.

Ad D.: I am not sure if I understand, but I guess you can find related information in Troubleshooting, 2.1.21 When I model the reinforcement, do the bars and stirrups have to be in geometric connection (with common intersection points) or they can be modeled according to their axis so that they do not touch one the other in geometric sense?