Model Tapered Members 1

[MKarr]

I'm helping another engineer in my office on a cantilevered canopy and he says he wants the model to be as accurate as possible. The canopy is to have sloped, tapered HSS members. I know RISA only allows you to model tapered wide flange members as tapered and other members have to be tapered in increments. For kicks and giggles I created a small model with the two sides of the canopy tapered differently. One side has all members modeled center line to center line (creating a non-level TOS) and the other side is modeled such that TOS for all members is continuous. For the second side I used rigid links to connect the member center lines. I applied the same loads to both arms and it looks like tip deflections are similar, with the continuous TOS values being greater in magnitude, but the intermediate deflection values get a little odd.

I currently have a total model of the structure with all members modeled by centerline that my supervisor wants me to change, so I'm wondering if it's worth changing the model to align everything TOS or are the small rigid links going to artificially increase my deflections?

The small model is attached.

Thanks,

 

[weab]

Not at the office so I can't review your model, but it sounds clear. Who cares if the the TOS is kept constant if RISA doesn't understand the reality. Is the bottom of steel constant? No, so why is that more realistic? What is realistic is that you have a tapered section in which the center line is constant.

If you have a slightly sloped tube to keep the top of steel constant, then your center line is also sloped. Keep the center line true and gradually change the cross section. This allows for the model calculations to be as close to correct as possible.

Isn't it most important that the analysis/calculations are correct? When you add eccentricities that don't really exist, that's when you get less accurate results. When you make approximations such as this for a tapered section, you do it because the software cannot process the real situation. Your job in this case is to safely make a close approximation that the software can properly use to provide the most realistic results.

If you want a nice picture for presentation purposes, then use your constant TOS model. Otherwise, go for accurate results.

 

[MKarr]

Thank you weab. That's what I was figuring would be the case and that was what I went ahead and modeled.

 

[JoshPlumSE]

I don't generally worry too much about maintaining the same top of steel elevation for tapered members like this. When I do worry about that, I will generally just change the slope of the member slightly so that modeling using pure centerline of the members will give me consistent TOS elevations.

There isn't anything wrong with using the rigid link concept like you show on the right side of the model. But, it does introduce extra complexity. Both in the modeling and the results interpretation.

Note:
I believe the AISC design guide on tapered members discusses this issue a bit. Specifically as section towards the end on "frame design" where they talk about a "curved centoidal axis" or a "centroidal axis offse" for a tapered member. But, my guess is that this won't apply to your model. Because your model is probably doubly symmetric at any point along its length.

 

[MKarr]

Yes, the total model is symmetric in both sides of the beam and in both directions.