Pushover "analyse failed, returned: -3 error flag" question

[NikosCV]

Dear All,

Say that I'm performing a displacement-controlled Pushover analysis of an RC structure.
I need to clarify the meaning of the "analyse failed, returned : -3 error flag" message, which occurs sometimes after several steps and stops the recording of the control node displacements.
Does it mean either that the structure has deteriorated before it reaches the target displacement of the control node or that there has been a numerical convergence failure?
Could it be either one of the above or the latter only?

Thank you in advance.

[fmk]

that message is coming from the analyze command .. so the analysis at the current step has failed .. if 1 step fails, you can of course do other things to move the analysis forward .. i suggest you look at the following basic pushover example:

http://opensees.berkeley.edu/wiki/index ... r_Analysis

the last message printed to the screen will tell you if success or failure

[NikosCV]

fmk wrote:
> that message is coming from the analyze command .. so the analysis at the current
> step has failed .. if 1 step fails, you can of course do other things to move the
> analysis forward .. i suggest you look at the following basic pushover example:
>
> http://opensees.berkeley.edu/wiki/index ... r_Analysis
>
> the last message printed to the screen will tell you if success or failure

Thank you for your concern.
I have implemented these analysis loops, but still I usually get this error message. Plotting the pushover curve until the last step recorded shows that the analysis usually fails within the horizontal branch of the curve.
What I meant previously is this: if the structure deteriorates before reaching the maximum displacement I set, will OpenSees stop the analysis and return a -3 error flag, or will it continue with plotting an horizontal branch on the pushover curve?

[fmk]

depends what you have in your script. in the exmaple i pointed you to, if the analysis fails and it does not achieve the desired displacement an error is retrnd and FAILURE is printed out. i suggest you look at the script again!

[NikosCV]

Thank you for your advice.

It is clear that if the analysis fails and it does not achieve the desired displacement, an error is returned and a "FAILURE" message is printed.
But, couldn't this message also suggest a numerical convergence failure (meaning that the whole response until collapse is not captured) ? And if yes, how can we distinguish these two?

Your help so far is much appreciated anyway