Clarification on 2D Modelling and Design

[Manabat]

Hi guys,

Q: Should I do 2d modelling of a global system or local system for my design?

Background: Im currently doing 2d analysis of a timber building on Microstran for a project in order to get my design actions and check them manually.

What's happening is, I'm separating the structural components, analyzing them before designing them, eg. wall sections are separated and analyzed separately from floor framing sections, etc.

However, when I model the system as a whole and analyze it, I'm getting different reactions - especially in terms of my deflection which I want to limit for SLS checks.

Your advice would be helpful on this.

Cheers!

 

[XR250]

Check it by hand

 

[DaveAtkins]

There is nothing wrong with using 2D modelling for design, or designing by hand, unless the Code requires a 3D analysis. This is only required for seismic design in certain situations.

DaveAtkins

 

[milkshakelake]

If things are separated, the only reason they would have different reactions is if they have different boundary conditions or load transfer, or there's a mistake. For example, in the full model, it might be partially fixed instead of pinned.

I agree with XR250 that checking by hand is important. It should always be done for a representative sample of members, whether it's a single member or a complex model with thousands of members. A model is only as good as the engineer making it. This means that good engineers will spend extra time making sure it works by hand. In the end, though this takes more time, it still takes far less time than doing everything by hand.

It's very common to model individual components instead of the whole structure. It's a useful tool for drilling down to specific members.

 

[JoshPlumSE]

This is a matter of engineering judgment.

I tend to side with the old adage, "keep it simple, stupid" (KISS). That's not saying anything about your intelligence level or anyone else's.

Rather, the idea of keeping any model simple helps me (or engineers in general) to focus on a simple model that will tell us 90% of what we need to know about a structure. That might be a hand calc. That might be a 2D model, or it might be a simple 3D model. The point is that adding extra complexity to how we model a structure isn't helpful if we're not 100% sure that the extra complexity is behaving the way it should. If you didn't model it right, then your results could be garbage and you might not even realize it.

 

[milkshakelake]

Yeah, there's also separating models into gravity and lateral. Because all the complexity from gravity doesn't always need to go into the model for lateral. Doesn't work for steel, but it works for concrete, wood, and cold-formed steel. In wood, there's generally no reason to model anything except main beams. Modeling the studs could add complexity. I just model them as concrete walls to keep things simple, and it affects literally nothing for the gravity design.

 

[rowingengineer]

Microstran is a beam type modelling software, how are you handling the floors, walls and ceiling diaphragms? If you are representing these are strap bracing then your distribution of loads will be different to the load distribution assumed thru the timber codes.

If your doing grillage type representing of any item, be careful of restraint setups and close members to not for a coupling moment.

Can we see a sketch of your building, as we are all guessing as to type and design. If a portal frame building the discussion will be different to a standard brace framed.