how to predict 75 recovery

huangyhg

how to predict 75% recovery?
the "ansi/dasma 108-2002 standard method for testing sectional garage doors" does not have a numerical criteria for deflection of the door like l/120 (in the closed position for wind load).  the criteria is listed below as a deflection recovery of 75% of the total deflection under load.
11.0 pass/fail criteria
11.1 the door system shall sustain both the
design load and the test load for the predetermined
amount of time.
11.2 upon the conclusion of design load testing,
the door system shall recover at least 75% of its
maximum deflection at design load for positive loads
and the same for negative loads.
11.3 the door system shall remain in the opening
throughout the duration of the test.
11.4 the door system shall be operable, or shall
be deemed to be operable, at the conclusion of the test.
my question is how does one calculate the design load at which you can expect the 75% recovery for a given door size?  the vertical or horizontal member's deflection under a particular applied load can be calculated and for a particular deflection (l/120) the applied load can be calculated.  but, given all the material and section properties, how do you go about predicting (calculating) one or the other which will result in a 75% recovery of original shape?
ok... i'll put in 2 cents worth...
start at the bottom of the requirements and work up.
1. make an engineering judgment if a door that is deformed at all is operable?
2. make an engineering judgment if a door that is deformed (during the test) will stay in the opening?
3. since things could get unpredictable very quickly if the elastic properties of the materials are exceeded, compute the test load so that in the worse case (weakest direction) you reach, but do not exceed, the material yield strength. this means calculating for 100% recovery (but just barely), for now.
4. determine if this test load magnitude realistically represents conditions during use (i.e. probably no need to have a test load that represents, say a 300 mph wind load)
5. since you now should have a load that a door should survive, run tests with your selected load and make revisions to the load (up or down) as needed. imho it would be a waste of time to calculate an initial test load that could cause failure - a door destroyed during the test will tell you little, other than the initial test load was too big.
that's how i would have approached this problem in my r&d days.
slideruleeara:
thanks for the comments.  i guess a full scale test is certainly the way to verify the door deflection and recovery. i was hoping to find a way to estimate this via calculation.
for example, if we said the structural frame was a c3x4.1 rolled channel.  i know the section properties, i know the allowable stress of a36 steel. so, i can calculate the wind loading that results in fy = 24 ksi, the "normal" allowable steel stress.  then, the resulting theoretical deflection is easy to calculate.  
i think the 75% recovery point actually occures at some unknown steel stress 24 ksi < fy <36 ksi. once the stress is above the yeild point, there is no deflection recovery. so, would you calculate the deflection at 24 ksi and 36 ksi and select a deflection of 25% between the two?
of course, this says nothing about whether the door stays in the opening.  it could possibly rotate out of the door track. so, am i back to a guess if i don't actually do a test on the door?
this is my take on it. imagine if your door remained elastic throughout its test.  then you would recover 100% of your deflection.  in others words, the deflection with no load, after a load cycle, would be zero.  if you recover only 75% of the deflection, that implies that your door system has undergone some plastic (permanent) deformation.  plastic deformation occurs when you have exceeded yield stress, not between allowable and yield.  while you may be able to estimate a load by back-calculating based on your parameters, it will not be as straight forward as with elastic deformations only.   
jheidt2543 - the problem with any structure loaded to failure is that if it is well built to start with, it ends up being stronger that calculations predict - as significant deflection begins,   
thanks for the thoughtful comments.  i guess to really be able to approximate the behavior in calculations does require some test data.  
after thinking about this some more, i do agree with both of you guys' comments.  it helps to be able to bounce some of these thoughts off others. i'm a one man office and the dog just gives me quizical looks when i ask him!
this discussion does raise another question through. is it in the design engineer's perogative to "push the envelope" because "limited failure" is acceptable in one code (ansi/dasma) while he exceeds the "allowable fy" of another code (aisc)?  if you prefece the design with a note indicating which code you intend to have preference, i would think it would be acceptable.
jheidt2543 - i know what you mean about being a one man office - me too. i don't see any problem or code conflict with a planned, "limited failure" - it is done all the time.
a good example is the crushable concrete used in engineered material arresting systems at the ends of aircraft runways