calculate damp and freq from free vibration

ebe79442114 · Post by **ebe79442114** » Wed Jul 08, 2009 11:56 am

Hi all,

I want to calculate frequency and damping of my model with free vibration analysis.

I have multiple choices for input motion and I examined these:

1- rectangular time series
2- sin time series
3- sinusoidal wave as an uniform support

my question:

1- In case of 1 and 2 the output is absolute? If not, can I calculate frequency and damping from relative output?

2- I can calculate frequency and damping with acceleration and displacement time history, which one is better? I guess displacement, but it seems no different?

3- if I want to calculate transfer function(fourier amplitude of top/ fourier amplitude of base motion), I must record the input in case of 1 and 2 (rigid base motion)

silvia · Post by **silvia** » Wed Jul 08, 2009 12:34 pm

none of the 3 you mentioned is free vibration, they are all forced vibration.

for free vibration you first pull/push your model to a specific displacement, then you remove that load and you do a transient analysis.

ebe79442114 · Post by **ebe79442114** » Wed Jul 08, 2009 1:28 pm

I know.

I apply load for short time (for example 3 seconds) and let to vibrate for 50 seconds.

1- how can I pull the node? with initial displacement in OS?

silvia · Post by **silvia** » Wed Jul 08, 2009 1:29 pm

i'd do a pushover (you already have that, likely), then remove the load pattern and do a dynamic analysis with no loads.
try it and let me know, i've never done it.

ebe79442114 · Post by **ebe79442114** » Wed Jul 08, 2009 1:54 pm

I have been done that, but in some way it is give me different damping.

is damping calculated from ABS or REL is different?

ebe79442114 · Post by **ebe79442114** » Thu Jul 09, 2009 5:39 am

I calculate freq and damp with three ways:

***************************************************
1- rectangular motion:
***************************************************

*input motion is:
pattern Plain 1 "Rectangular 0 [expr 20*$deltaT]" {
load 2042 1000 0 0
}

pic: freeimagehosting.net/uploads/c28a706a06.jpg

&results:
------------------------------------
xi is: 0.00791896607077
Wn(Hz) is: 0.532405951651
------------------------------------
***************************************************
2- sin wave:
***************************************************

*input motion is:
pattern UniformExcitation 1 1 -accel "Sine 0 2 0.77 -factor 2.943"

pic:freeimagehosting.net/uploads/718693baa8.jpg

&results:
------------------------------------
xi is: 0.00680474652142
Wn(Hz) is: 0.533258996841
------------------------------------

***************************************************
3- uniform excitation (earthquake sin wave from file):
***************************************************

*input motion is:
pattern UniformExcitation 1 1 -accel "Series -factor 9.81 -filePath /home/ebi/Desktop/realSsi/GMdir/testResult/freeVib/freeVib -dt 0.033"

pic: freeimagehosting.net/uploads/f76736d289.jpg

&results

------------------------------------
from relative displacement:
xi is: 0.00379020819018
Wn(Hz) is: 0.532665913278
------------------------------------

------------------------------------
@absolute Acceleration:
xi is: 0.0073
wn is: 0 .531
------------------------------------
@relative Acceleration:
xi is: 0.0072
wn is: 0.531
------------------------------------

***************************************************
from EXP result
***************************************************
xi is: 0.0049
wn is: 0.523

from these results found that relative displacement is close to the Experimental results,

and another ways are approximate one answer.

please excuse me for my poor english.

silvia · Post by **silvia** » Thu Jul 09, 2009 6:34 am

it is relative, yes.
the rectangular load history should be what you want, but make sure you change the analysis type at the right time.
remember, keep the displacements elastic.

ebe79442114 · Post by **ebe79442114** » Thu Jul 09, 2009 10:35 am

one another question:

in rectangular and sin input motion, what is the dt for analysis? where can i say to OS to apply dt for input motion?

is that in analysis command?

my Structure is elastic, but my soil not:

freeimagehosting.net/uploads/1b26b6915e.png

silvia · Post by **silvia** » Thu Jul 09, 2009 12:28 pm

the analysis dt is specified with the integrator.

ebe79442114 · Post by **ebe79442114** » Fri Jul 10, 2009 12:00 pm

in uniform excitation we must specify dt for input motion, in rec and sin how can i say to OS to apply load in dt=0.033 sec?

thanks.

silvia · Post by **silvia** » Fri Jul 10, 2009 12:15 pm

you specify dt only when you only-specify y values.

you specify dt for analysis in the integrator.

ebe79442114 · Post by **ebe79442114** » Fri Jul 10, 2009 10:59 pm

in the integrator I say:

integrator Newmark 0.55 0.275625

I want know, if my model didn't converge, how can i reduce time step for integration.

excuse me, perhaps my question is very simple, but really I don't know.

e.roknabadi · Post by **e.roknabadi** » Sat Jul 11, 2009 12:45 pm

I am ebe79442114, excuse me but I have been bonded from this site.

one another question:

should I filter Acceleration or displacement output?

---------------------------------------------------------
if I don't filter output, results is:
xi is: 0.0392565750178
Wn(Hz) is: 1.30922183127

pic: freeimagehosting.net/uploads/318b099f93.jpg
---------------------------------------------------------
and if I filter output(0.5 to 2 Hz bond pass filter), results is:
xi is: 0.0396612321389
Wn(Hz) is: 1.28040973111

pic: freeimagehosting.net/uploads/11c8ea8e94.jpg
---------------------------------------------------------

the frequency is different.

silvia · Post by **silvia** » Sat Jul 11, 2009 3:52 pm

pretty close, though.
it's your call. you can ask frank to un ban you