design of built-up w sections for combined torsion and bendi

huangyhg

design of built-up w sections for combined torsion and bendi
a current project in our office has required us to size a very large w-section. a height restriction of 24" has been set by the architect to make the beam more "interesting" for us. the beam spans 45' and with its cont. and point. loads must resist a 154k-ft moment.  we have just recently chosen to design a built-up beam that will consist of (2) w21x147 beams w/ a cont. 1/2" plate above and below. one of the point loads is 77k and must be set into the webs of one of the beams, thus generating a torsional moment on the beam. we have treated the built-up beam as an open and closed section while generating its properties(i,s,j, etc). we have been looking for additional information on such built-up/composite sections as we have not been able to calculate what we feel is an acceptable f.o.s. there may be alternative methods to generate the torsional properties of the beam or perhaps our approach has been incorrect. so far lrfd methods have been used along with some theory from timoshenko's mechanics book.
does anyone have any experience with any beams of similar construction? does anyone have additional concerns for the performance of the beam (welds, web stiffeners etc.)?
although the w21x147 is not considered a heavy section will it be compatable with e70 eloctrodes?
thank you
imho,
   for the torsion strenght i would take in account only the section properties of an equivalent rhs formed by the two w's inner half flanges and their respective webs.. but this aproach depends on the welding arrangement utilized to join the w's and the cover plates.  when dealing with open
shapes, the torion can be "broken down" into st. venant torsion (pure shear stresses) and warping torsion (both shear stresses and axial/bending stresses as result you avoid to get in a very complex issue.
  you should check  aisc's design guide 9 and also salmon & johnson here you can find a very good theory discussion on
torsional stresses.
   see also sci's advisory 249 design of members subjected to torsion
   
imho-
thank you for the response.
a full penetration weld at the flanges (top and bottom) should allow us to consider the webs and connecting portions of flanges as a closed section.  the additional plates stiffen the flanges and we feel that the beam will perform well against localized stresses(i.e. warping shear) at these locations.
the issue we have is primarily in the webs of the (2) w sections. we have reviewed the aisc tutorial on the "torsional analysis of steel members" and have determined that the warping shear is the main problem.
the combination of bending and torsion stress approaches the limits of the beam. we feel that web stiffeners at the locations of point loads on the beam will help stiffen the beam (the ends of the beam can also be considered torsionally pinned as the webs are braced).  
the outer edges of the flanges have been added to the calculation of our j-value by the lrfd equation: sum of (bt^3)/3.  we feel that this is acceptable when checking the beam under the combined stresses because the flanges do help resist the bending stress.
is there an alternative method of calculating the torsional property of the beam(j) that would more accurately reflect its behavior?
again, thank you for your response. we were hoping that some experience out there had some input.