Hello. Please go through attached PowerPoint and then the below discussion will make more sense. I am using STAAD version 20.07.11.70.
The main problem is: when modelling decks or bulkheads which include offset beam elements and plates attached to the same nodes to which the beams are attached, the plates seem to 'attract' a lot more load than an FEA check would suggest. Furthermore, they seem to skew the reported beam forces, sometimes hugely.
About slide 5 - axial force results for vertical structure vertically loaded:
The only difference between model 1 and model 3 is the fact that I deleted the beam offsets from model 1. However, even model 3 seems to be sending far too much axial load down the plates instead of the girders, when you consider that the axial load is applied directly at the girders. The relative stiffness of the plates would not be enough to attract nearly 50% of the compressive load that is travelling through the model.
About slide 6 - code check results:
The main issue here is that code check results report lower URs because less compression is present in the beams. I fear that the reduction in compressive stress in the beams is exaggerated and we are in fact missing potential issues in designs.
About slide 9 – axial force results for horizontal structure vertically loaded
When things become horizontal, it is easier to see that the axial force in the beams is strange. To me, there is no net external axial tension acting on the beam.
Here is my theory:
The plate elements are participating in the beam, acting as an extra flange (obviously). If you would consider this deck plate as an extra flange, you could locate a new neutral axis for the hybrid beam + plate (so long as you understood the effective flange width applicable), which would be higher up than the H-beam NA alone. Now, if you were to look at internal axial stress at the NA of the beam alone, you would see that it was tensile in nature since you’re looking now below the true combined NA. STAAD simply takes this axial stress number, and multiplies it by the H-beam’s cross-sectional area to ascertain these axial force results.
I was able to back up this theory by creating a solid finite element analysis in another software and read off the axial stress at the point in question.
Could the community offer some advice about what is the best way to create a hybrid model such as this so that the eccentricities of the real situation are captured but the forces within the beams (and therefore code check URs) are not so severely skewed?Plate Gate.stdHybrid Models.pptx