paring down sesmic loads

huangyhg

paring down sesmic loads
at my current place of employment when there is a problem with the seismic loads they like to reduce the dead loads to "what is actual there".  
example:  
we design a roof for a gravity dead load of 24psf.
deck - 3  psf
joist - 6 psf
roofing - 10 psf
mep - 5 psf
then we design the weight for seismic using a dead load of 20 psf
deck - 2  psf
joist - 5 psf
roofing - 8 psf
mep - 5 psf
we use the logic that the second set of loads are more acurate becasue they more closely reflect what is really there.  we then proceed to put the 24 psf dead load on our plans as what we design for, but have the seismic load really correspond to less.  
i don't feel this is correct.  is this type of thing standard practice?  is there a code reference (ibc) that tells me this is or is not allowed?  

i have done that.
the one thing to worry about is re-roofing, new ceiling, new hvac units etc.
our local code enforcement buys off on it.
is it right or more accurate - kind of an engineering judgement.
i beleive the code allows you to use actaul weights if you can define them.  most people pick conservative loads for gravity - i do.  but when it comes to seismic - i often use actaul based on the assumption the really big one will never hit.  maybe not the greatest decision.
we do the same thing.  for instance in warehouse design it is common to add 3-4 psf of fire sprinkler load into the gravity design for bar joists so that the sprinkler trunk lines (3" pipes) can be hung just about anywhere.  if you had a 100,000 sf warehouse this would mean a mass of 300,000 lbs.  more accurately a 3" pipe at 30 ft or so should give you a mass of around 100,000 lbs.  gives you flexibility in gravity design without penalizing your seismic design.
this what is called "sharpening your pencil".  using more accurate values for assumed loadings and is done only when the first assumption of load values gives unworkable stress.  if the first assumption works, then stop designing!  foolish to spend 150 dollars worth of engineering to save 125 dollars worth of material unless you plan to manufacture a few thousand of this design.