Failing in STAAD but standing since years in real, members 16 17 35 53 are failing , in real they shouldnt be. Plz help

Question

Keynsham.std

Tank · Accepted Answer

You've got the bottom chord designed as having an unbraced length about the local z-axis as its full length, even though it is predominately in tension. The end segments which are failing are in compression. Due to the very long unbraced length the axial strength is being computed as a very low number. I don't think this is realistic. I did a buckling analysis on load case 3 (had to change it to repeat load) and got a buckling factor of 11.7. There's a clue why the structure is still standing. You need to define more realistic unbraced lengths for the top and bottom chords. Better yet, do a nonlinear analysis. I'm not familiar with the British code so I don't know how it feels about that. Just remember when doing nonlinear you don't take the results and apply code checks to them because you're doing some things analytically that the code is accounting for so you would in effect be double dipping.

Tank · Answer

Mahesh, the stress in member 35 due to major axis bending alone is 319 MPa (using plastic modulus) but the yield strength is only 275 MPa so you've definitely got a problem. Any combination of the following could be happening. 
 
1) The detail is such that the connection where member 35 ties into the column is not really as rigid as STAAD thinks it is. This would reduce the moment here, but remember, it has to go somewhere. 
 
2) A plastic hinge has formed, yet the structure still remained stable. Just for grins, I reduced the Mz stiffness of the ends of members 35 and 53. I used MPZ 0.4 after some trial and error to get close to their moment capacity. All members in the structure now pass code check. Please understand, I'm not saying that makes everything OK. It just explains why it didn't fall down. 
 
3) Steel usually is stronger than its published minimum value. In the states, it's very common to see A36 steel with a minimum yield strength of 36 ksi to actually test above 50 ksi. 
 
Back to the unbraced lengths. I didn't mean to imply that the unbraced lengths of the compression members would equal their node to node lengths but somewhere between this and the full length. It's a fairly complicated problem. One that, unfortunately, I don't have the time to get into other than to suggest you do a more rigorous analysis using nonlinear. Somebody like Dr. Yura may have more classical solution but I don't know right off the top of my head. 
 
You haven't given us much back story, but it seems like it is definitely a cause for concern.