aluminum tube dome roof connectionsdesign

huangyhg

aluminum tube dome roof connections/design
once again i pusing the limits of my knowledge and i need some direction.  i have modeled a 56' diameter dome 11' high in risa which is to be constructed of aluminum for corrosion resistance.  20 segmented arches connected with x bracing.  every member is angled and twisted.  the structure will be supporting an existing fiberglass dome which has not completely failed yet, the snow is coming though.  it is a clarifier cover so no field welding due to methane.  so far i have a 4kip frame dead load after a lot of work.  i have the 2005 aluminum design manual to check my model with but i am not sure about the tubular connection design.  should i approach these connections as if designed with steel (using aisc tube steel design methods) using the properties of aluminum or is there another reference which would be more appropriate.  i'm trying to avoid building a fem of each connection as my fee will certainly not cover that.
thanks
paul irwin pe
ripple design studio
check out our whitepaper library.
you can't design aluminum structures with steel codes.  try getting your hands on design information from the aluminum association.   one thing to watch out for--welding aluminum reduces the strength by a lot, maybe 50%.
thanks hokie66, i have the aluminum design manual 2005 but there is nothing in there specific to connections with tubular members similar to steel design.  currently my approach is as follows:
1. model the structure with aluminum material and standard shapes in risa (risa will include aluminum next year, but connections are a different animal)
i have designed lots of industrial dome roofs.  traditionally, they use i-beams with dished connection plates in a triangular space frame configuration.  if this might suit you, there are several companies that make these structures, or we might be able to collaborate on one.
vato, i cannot stress hokie66's assertion enough - you cannot use steel codes to design aluminum connections.  the codes are written with a specific material in mind.  aluminum, with 1/2 the strength and 1/3 the stiffness of steel, will have failure modes in a connection situation that weren't even considered when writing the steel code.
maybe try to standardize your connections and fem a few of them...
if you "heard" it on the internet, it's guilty until proven innocent. - dcs
it is sounding like there is no published connection design method for aluminum similar to steel hollow structural section methods per aisc.  i have not heard back from the aluminum association yet.  i am using tubes and i still think that the connection methods and design procedures, using the correct allowable fy for the aluminum per adm05, which is different for almost every condition and   
vato, how feasible will it be to do some full load tests on a connection or two?  i've done this a couple of times through or local university and every time, we've all gained some new understanding of how things fail  
it is reasonable to use the aisc procedures, modified for aluminum properties.  please note that structural grades of tempered aluminum are not 1/2 the strength of steel.  as an example, using 6063-t6 aluminum, the allowable stress in bending is slightly more that 15 ksi; while a36 steel is slightly less than 22 ksi....that's more than 70 percent, not less than 50 percent.  lrfd results are similar.
as someone noted, the allowable stress on welded aluminum is significantly reduced.  again as an example, the 6063-t6 material within 1 inch of the weld has an allowable bending stress of 6.67 ksi....big reduction!  while that does not actually happen, it has been used by the aa for many years due to the variation and lack of predictability of the post-weld strength in the heat affected zone of the weld.
first off, thanks to all for chiming in on this one.  i have standardized the connection and i am building a fem of it.  the model of the dome with appropriate loading has given me the member forces to check with the adm05.  the connection stresses, which will be checked against the reduced due to weld values, are within reach now.  i have discussed constructing the entire frame and cutting it up into 10 "leafs".  there are a lot of ways to slice it.  anyway, it may give us an opportunity to load test it.