Return to index: [Subject] [Thread] [Date] [Author]

Re: Non Linear Dynamic Analysis and the number of records

[Subject Prev][Subject Next][Thread Prev][Thread Next]
>In non linear time-history analysis, we face, amongst others, the following
>problems:
>
>1.- The selection of the records
>2.- The number of records to use, or
>3.- The allowed  one-sigma confidence bandwidth in the mean response of the
I've done some non-linear time-history work, and I find it a serious pain 
in the ass. I think it's a last desperate resort, if only because of the 
arm waving necessary to define a transient. The only thing you know for 
sure about existing earthquake records is that they represent events that 
will never happen again. All the statistical munging is nice, but my own 
feeling is that it's a forlorn attempt to put numbers on the 
unquantifiable. Sometimes you're forced to run a problem as a transient, 
but given the choice, I prefer response spectrum analysis, simply because 
it's straightforward, easy to check and doesn't carry any of the 
numerical problems associated with non-linear transient work. 

I think the number of records specified relates to synthetic records 
developed from ground response spectra--I don't know of very many sites 
that have more than a very few design-basis earthquake ground 
acceleration records. There's no magic number of records to make your 
analysis more accurate. Last time I ran one we were given three synthetic 
records to use because the earthquake wizard thought 3 was better than 2 
or 4. (Turned out they were crap--the motions were correlated, had no 
zero base correction and they'd been munged to provide a given ZPA by 
simply truncating the acceleration history peaks at the desired level. We 
only found this out after running a number of transients and noting that 
something looked strange.) 

CW has it that time history analysis is more accurate, but I think that's 
absurd for seismic analysis, given the indeterminacy in the loading and 
all the numerical hoops to jump through.  Bill Cain's comments are right 
on: 'Over reliance on analysis with out exercising good engineering 
judgement will not make your structure safer.' 

>If we need a 95% confidence level, and a +o- 10% bandwidth,  means
>use a very large number of records, if my figures are accurate, between 100
>to 324, being a function of dispersion.
First thing to ask yourself is whether Lima has ever experienced 100 to 
324 design basis earthquakes, ever. 

Christopher Wright P.E.    |"They couldn't hit an elephant from
chrisw(--nospam--at)skypoint.com        | this distance"   (last words of Gen.
___________________________| John Sedgwick, Spotsylvania 1864)
http://www.skypoint.com/~chrisw


******* ****** ******* ******** ******* ******* ******* *** 
*   Read list FAQ at: http://www.seaint.org/list_FAQ.asp 
*   This email was sent to you via Structural Engineers 
*   Association of Southern California (SEAOSC) server. To 
*   subscribe (no fee) to the list, send email to 
*   admin(--nospam--at)seaint.org and in the body of the message type 
*   "join seaint" (no quotes). To Unsubscribe, send email 
*   to admin(--nospam--at)seaint.org and in the body of the message 
*   type "leave seaint" (no quotes). For questions, send 
*   email to seaint-ad(--nospam--at)seaint.org. Remember, any email you 
*   send to the list is public domain and may be re-posted 
*   without your permission. Make sure you visit our web 
*   site at: http://www.seaint.org 
******* ****** ****** ****** ******* ****** ****** ********