different resuts with diffrnt dt in nonlinear dynmc analysis

silvia · Post by **silvia** » Thu Jan 29, 2009 1:33 pm

which means you might miss peaks in your ground motion, hence response.

imancivil · Post by **imancivil** » Thu Jan 29, 2009 9:35 pm

Dear all,

I did the nonlinear dynamic analyses for a big structure which has 841 nodes and 3200 elements. I used the Imperial Valley Earthquake that its time step is 0.01 sec with duration of 20 sec.
I assumed 3 different states with maximum duration of ground-motion analysis of 20 sec:
a) dt of analysis = 0.005, the number of time steps= 4000
b) dt of analysis = 0.01, the number of time steps= 2000
c) dt of analysis = 0.02 , the number of time steps= 1000
the results of analyses (displacement of a certain node) are compared below:
a) d=0.2987 m
b) d=0.29852 m
c) d=0.29643 m

would you please tell me which one is correct??

susan · Post by **susan** » Thu Jan 29, 2009 10:13 pm

Dear Frank,

1. you say that "the deltaT for the analysis and dT for the input motion do not have to be the same".
now, why do you use the same time step in one of the example manual (Ex2b.Canti2D.InelasticSection.EQ.tcl)? Is it wrong?

2. in your opinion, it is important that dt of the analysis should be not equal to dt of the input; whether dt of the analysis be less than dt of the input or not. right??

would you please tell me the clear reply about this topic? in the other words, which imancivil's results is true?
thank you.

silvia · Post by **silvia** » Thu Jan 29, 2009 10:22 pm

imancivil,
can you please also add the following time steps:
0.0025
0.0075

and, if you'd like, 0.0175

the time steps in the example may be the same, which is not the end of the world.

this is a numerical-simulation problem, not opensees-specific, so there must have been some studies done on time steps.

imancivil · Post by **imancivil** » Fri Jan 30, 2009 1:05 am

Sorry, I made a mistake to enter the results in the forum. I edited my previous comment. Also I added the time steps that silvia requested below:
dt of analysis = 0.0025 , disp.=0.29882 m
dt of analysis = 0.0075 , disp.=0.29832 m
dt of analysis = 0.0175 , disp.=0.29686 m

the results show that there is not a lot of difference between various the time steps.
Any opinion?

silvia · Post by **silvia** » Fri Jan 30, 2009 8:09 am

excellent, you manage to capture the peak.

ebe79442114 · Post by **ebe79442114** » Mon Feb 02, 2009 8:35 am

Hi all

I found from most articles, that it is better to assume dt for integration 0.1 times of deltaT (input motion time step). my quesion is:

I want to obtain absolute acceleration of output,how can I sum input and output with different length ? ( output is 10 times of input motion). is there a simple way? or I must for example do this in post process?:
load both of input and output and then add input and output(0

10)?

please excuse me for my poor english

silvia · Post by **silvia** » Mon Feb 02, 2009 8:41 am

in the recorders you can specify dT, that way you have consistent results.

silvia · Post by **silvia** » Mon Feb 02, 2009 8:42 am

also, dt=0.1 times input time step may be pushing it, the analysis will take forever!!!