Need a book? Engineering books recommendations...

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

Re: Story Drift: 1994 UBC vs. 1997 UBC

[Subject Prev][Subject Next][Thread Prev][Thread Next]

Here's the scoop on 3Rw/8 being changed.

If we go way back in the blue book  (before 1988), back in the days when V
was still equal to ZIKCSW, the commentary noted that the design forces,
given by that equation were significantly lower than the actual anticipated
resopnse of the structure, and that probably, the real inelastic response
was 3 or 4 times larger than that calculated under the design forces.
Remember this was in the days before common use of dynamic analysis, or
even widespread knoweldge by engineers of response spectra.  In the code
itself, not much use was made of this, other than that attachment of fascia
panels had to accomodate 3/K times the computed drift under the design
forces.  Since K varied from a low of .67 t o a high of 1.33, 3/K was
roughly "3 to 4" times.

In 1988, the code was "rationalized" using the theory developed in
ATC-3.06.  ZIKCSW became ZIC/Rw, and the old structural quality factor, K,
became 8/Rw, so that and K=1 building in the 1985 UBC was an Rw=8 building
in the 1988 UBC, etc.  At this same time, it was recognized that the design
forces were reduced from the actual ground motion by the factor R and that
this was possible due to a number of factors, including overstrength,
ductility, hysteretic damping, etc. ,   While all this as true with regard
to the forces induced in a structure, it is not true with regards to
displacements.  The best estimate of the real displacement induced in a
structure by ground shaking is that computed with an R factor of 1.0.  By
1988 this had been demonstrated many times by Newmark, Hall, Bertero,
Krawinkler, and many others.  However, SEAOC held on to the "3 or 4" times
larger myth from the past.  Now "3 or 4" times larger became 3Rw/8 - with 8
varying between 12 and 6 still resulting in approixmately 3 to 4.

Now where the 1988 UBC went wrong is that it used 3Rw/8 both to estimate
the "real" displacement and also the maximum force that could develop in
teh structure do to overstrength.  It would have been better if it had used
a different factor for overstrenght than deflection amplification.  3Rw/8
is a reasonably good factor for overstrength but not for real deflection.
8Rw/8 would have been better for deflection.  In teh 1997 UBC, SEAOC
finally recognized reality and stopped using 3Rw/8 to estimate real
deflection.  Instead in 1997 we went to  0.7R.  Still not really enough,
but a lot closer.  The Omega -sub-0 has replaced 3/8Rw for force and is
approximatley equal to 3Rw/8

Seaintonln(--nospam--at) on 08/11/99 01:59:05 PM

Please respond to seaint(--nospam--at)

 To:      seaint(--nospam--at)                                   
 cc:      (bcc: Ron O. Hamburger/EQE)                         
 Subject: Re: Story Drift: 1994 UBC vs. 1997 UBC              

If I understand what you're saying, the old method allowed the designer to
design the column based on deflection matching an adjacent shear element,
when in reality, the column was expected to deflect much more. Therefore,
results of prior codes produced columns that were much more flexible than
adjacent walls and could, conceivably produce greater damage from shear
produced by torsion in the diaprhagm.

I was always under the assumption that the calculated shear was magnfied by
3Rw/8 time Pe  (if you use the '94 UBC Rw for cantilevered columns of 3
Table 16-P) or 1.125 times. Next I would design the deflection in the
using a fixed column deflection formula. I now fail to see how the
can be so much greater in the new code if the actual load applied (assuming
working stress design) was simply increased by this factor and static
formulat's to determine deflection were used to calculate the bending or
deflection in the cantilevered beam.

Now I'm confused??? I thought the 3Rw/8 term was to add a level of safety
into the design of the column by purposely making it stiffer.


In a message dated 8/11/99 12:17:17 PM Pacific Daylight Time,
mtv(--nospam--at) writes:

<< Bill:

 Yes, the required elastic stiffness is different.  However, because
 it is now based on 0.7 times the elastic displacement for the
 unreduced forces (because R cancels out), the values can be more or
 less stringent than in previous editions of the UBC.  That is, the
 old drift limit was based on REDUCED "elastic" design forces that are
 not really related to the total inelastic displacement that will
 occur.  It has been observed that the total displacement is almost
 unrelated to the level of inelastic response (whether R is 1 or 8).
 This has been dubbed the "equal displacement rule", although "rule"
 is too strong a word.

 In the example you provided (cantilever columns), more of that fake
 "service" displacement is now allowed because the displacement that
 really matters (for seismic response) is the total displacement
 including inelastic action.  For special steel moment frames, less
 "elastic" displacement is allowed using the 1997 UBC.  The total
 inelastic displacement allowed for both systems is the same using the
 1997 UBC.

 Here's a (non-dimensional) way to calculate the required elastic
 stiffness (to meet the drift limits):

 Kmin = F / d

 Under the 1994 UBC, F was 1/Rw and d was 0.005.  Therefore, Kmin is
 1/(0.005Rw) which varies from 66.7 to 16.7 as Rw varies from 3 to 12

 Under the 1997 UBC, F is 0.7 (because the check involves 0.7R
 times the displacement with F = 1/R; R cancels out) and d is 0.025.
 Therefore, Kmin is 28 regardless of the value of R.

 The answers would be the same for Rw = 7.14.  However, the new way is
 more consistent with the displacements that cause seismic performance
 problems.  One implication is that displacements that are really due
 to service-level loads (like wind) may now control.

 -Mike >>