SAP2000 - nonlinear static analysis 1

[ECStruct]

Hi folks,

SAP2000 V9

I´m trying to do a nonlinear static analysis with P-Delta (no material nonlinearity yet).
I need some help.

During analysis and in the '.log' file SAP2k V9 presents some information not reported or explained in the manuals.

Here they are (resumed to show only the 'non-self-explained' ones):

N O N L I N E A R S T A T I C A N A L Y S I S

1) TYPE OF GEOMETRIC NONLINEARITY = P-DELTA

2) MAXIMUM NUMBER OF ITERATIONS PER STEP = 10
3) RELATIVE FORCE CONVERGENCE TOLERANCE = 0.000100
4) RELATIVE EVENT TOLERANCE = 0.010000


5) TOTAL NUMBER OF CONVERGED STEPS SAVED = 0
6) TOTAL NUMBER OF CONVERGED STEPS NOT SAVED = 1
7) TOTAL NUMBER OF NULL STEPS = 5
----------
8) TOTAL NUMBER OF ALL STEPS = 6

9) TOTAL NUMBER OF ITERATIONS CONVERGED = 2
10) TOTAL NUMBER OF ITERATIONS DISCARDED = 5
11) FOR CONVERGED STEPS ONLY:
12) AVERAGE NUMBER OF ITERATIONS PER STEP = 2.00
13) MAXIMUM NUMBER OF ITERATIONS PER STEP = 2


Unbalance/tolerance = xx.xxx , Iteration 2 converged
Saved as Output Step 0

What is this? What are the variables SAP2000 compares?


14) We see in the analysis screen:

saved Null Total Interaction Relative Curr Step Curr Sum Max Sum
steps Steps Steps this step Unbalance Size of steps of Steps


Now the questions:

a) in (5) to (10) SAP2k is telling us that no step is saved? i can´t understand what is
going on! See (5): SAP2k shows that no converged steps was saved? Is this right? I see the results on screen!..

Topic (14):

b) "Null Steps" = steps not saved?

c) "Interaction this step" = substeps? As in Ansys?

d) "Relative unbalance". Here the major doubt.
Sometimes SAP2k considers that the analysis is converged but the Relative "unbalance value" is
near unit. Is it right? Is this variable a measure of the quality of the results? Relative is a comparision of two values, right? So I can ask.... what is SAP2k comparing
(measuring relative error)? Displacement? Forces?

Every time this value was near unit I´ve compared joint reactions vs. applied loads but never got any errors!

e) "Curr step Size" = I´ve learned that when SAP2000 reduce too much this value and "Max Sum of Steps" don´t change there is a problem in analysis / model. Is this right?

f) Convergence is archieved when "Max Sum of Steps" equals unit. So I fix eyes on "Relative unbalance" and "Max Sum of Steps". Sometimes "Relative unbalance" < 10e-5
(less than RELATIVE FORCE CONVERGENCE TOLERANCE, above) but the analysis / step did not converge. (???)

Right now the analysis screen shows:

Relative unbalance = 20.688152
Curr step Size = 4.69E-121 (!)
Max Sum of Steps = 0.250000

I´m in trouble! HELP!

Thanks in advance.
Marco

 

[arigato]

Hmm maybe I am missing something, but unless you have hinges or nonlinear links in your model, your going to get elastic behavior out of your structure, and should result in a linear pushover curve. Is this what you wanted?

The unbalance check comes from the solution trying to converge to a value that is within a certain tolerance. In dynamic analysis this means that a solution can not be reached and you have force imbalances in the solution perhaps due to numerical instability.